docs/dataflow/all.build_simple_risk_model_dag.tutorial.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "16a64c12",
   "metadata": {},
   "source": [
    "Implement a simple risk model estimation framework."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4646d956",
   "metadata": {},
   "source": [
    "# Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "52b6e9bf",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.034328Z",
     "start_time": "2023-06-04T18:41:10.922299Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/app/core/signal_processing/decorrelation.py:12: TqdmExperimentalWarning: Using `tqdm.autonotebook.tqdm` in notebook mode. Use `tqdm.tqdm` instead to force console mode (e.g. in jupyter console)\n",
      "  from tqdm.autonotebook import tqdm\n"
     ]
    }
   ],
   "source": [
    "import logging\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import sklearn.linear_model as slmode\n",
    "\n",
    "import core.artificial_signal_generators as carsigen\n",
    "import core.config as cconfig\n",
    "import core.features as cofeatur\n",
    "import core.finance as cofinanc\n",
    "import core.finance.market_data_example as cfmadaex\n",
    "import core.signal_processing as csigproc\n",
    "import dataflow.core as dtfcore\n",
    "import dataflow.system as dtfsys\n",
    "import helpers.hdbg as hdbg\n",
    "import helpers.hprint as hprint"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "eff83ad1",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.057826Z",
     "start_time": "2023-06-04T18:41:15.036898Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[0m\u001b[36mINFO\u001b[0m: > cmd='/venv/lib/python3.8/site-packages/ipykernel_launcher.py -f /home/.local/share/jupyter/runtime/kernel-352f8814-e59a-4edc-abe8-f56b7fe7240b.json'\n",
      "WARNING: Running in Jupyter\n"
     ]
    }
   ],
   "source": [
    "hdbg.init_logger(verbosity=logging.INFO)\n",
    "\n",
    "_LOG = logging.getLogger(__name__)\n",
    "\n",
    "hprint.config_notebook()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "51564267",
   "metadata": {},
   "source": [
    "# Generate input data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "0d7927e1",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.064841Z",
     "start_time": "2023-06-04T18:41:15.062100Z"
    }
   },
   "outputs": [],
   "source": [
    "# Assume there are `n` assets\n",
    "# - Generate returns for the `n` assets at each time `t` (normalized)\n",
    "# - Generate `k` features for each asset (centered, normalized)\n",
    "# - Wrap in a dataframe source node"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "e593a67d",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.071360Z",
     "start_time": "2023-06-04T18:41:15.066708Z"
    }
   },
   "outputs": [],
   "source": [
    "n_assets = 4\n",
    "n_features = 2\n",
    "n_periods = 10\n",
    "freq = \"B\"\n",
    "period_start = \"2023-05-01\"\n",
    "rng_seed = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "c3700637",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.083071Z",
     "start_time": "2023-06-04T18:41:15.074436Z"
    }
   },
   "outputs": [],
   "source": [
    "def get_random_data(\n",
    "    n_assets: int,\n",
    "    n_features: int,\n",
    "    n_periods: int,\n",
    "    period_start: str,\n",
    "    freq: str,\n",
    "    seed: int\n",
    ") -> pd.DataFrame():\n",
    "    \"\"\"\n",
    "    Generate a dataframe of random returns and random features.\n",
    "    \"\"\"\n",
    "    # Create datetime index for dataframe.\n",
    "    idx = pd.date_range(start=period_start, periods=n_periods, freq=freq)\n",
    "    # Create columns names for the X matrix.\n",
    "    cols = [f\"x_{k}\" for k in range(1, n_features + 1)]\n",
    "    cols = cols + [\"ret_0\"]\n",
    "    # Instantiate random number generator.\n",
    "    rng = np.random.default_rng(rng_seed)\n",
    "    dfs = {}\n",
    "    for n in range(n_assets):\n",
    "        asset_id = 100 + n\n",
    "        dfs[asset_id] = pd.DataFrame(rng.standard_normal(\n",
    "            (n_periods, n_features + 1)\n",
    "        ), idx, cols)\n",
    "    df = pd.concat(dfs, axis=1).swaplevel(axis=1).sort_index(axis=1)\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "027b7713",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.127564Z",
     "start_time": "2023-06-04T18:41:15.085835Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0</th>\n",
       "      <th colspan=\"4\" halign=\"left\">x_1</th>\n",
       "      <th colspan=\"4\" halign=\"left\">x_2</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2023-05-01</th>\n",
       "      <td>0.330437</td>\n",
       "      <td>-0.377605</td>\n",
       "      <td>-0.982188</td>\n",
       "      <td>-0.094828</td>\n",
       "      <td>0.345584</td>\n",
       "      <td>2.117839</td>\n",
       "      <td>0.054102</td>\n",
       "      <td>0.876242</td>\n",
       "      <td>0.821618</td>\n",
       "      <td>-1.112021</td>\n",
       "      <td>0.272791</td>\n",
       "      <td>0.256486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-02</th>\n",
       "      <td>0.446375</td>\n",
       "      <td>0.663063</td>\n",
       "      <td>-0.466750</td>\n",
       "      <td>-2.250854</td>\n",
       "      <td>-1.303157</td>\n",
       "      <td>2.042772</td>\n",
       "      <td>-1.107373</td>\n",
       "      <td>-0.258848</td>\n",
       "      <td>0.905356</td>\n",
       "      <td>0.646703</td>\n",
       "      <td>0.199585</td>\n",
       "      <td>1.055743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-03</th>\n",
       "      <td>0.364572</td>\n",
       "      <td>0.167465</td>\n",
       "      <td>-1.648787</td>\n",
       "      <td>-1.425349</td>\n",
       "      <td>-0.536953</td>\n",
       "      <td>-0.514006</td>\n",
       "      <td>0.235506</td>\n",
       "      <td>-0.138655</td>\n",
       "      <td>0.581118</td>\n",
       "      <td>-1.648075</td>\n",
       "      <td>0.759520</td>\n",
       "      <td>0.033000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-04</th>\n",
       "      <td>0.546713</td>\n",
       "      <td>-0.683227</td>\n",
       "      <td>-0.297527</td>\n",
       "      <td>0.862445</td>\n",
       "      <td>0.294132</td>\n",
       "      <td>0.109014</td>\n",
       "      <td>0.254388</td>\n",
       "      <td>0.332814</td>\n",
       "      <td>0.028422</td>\n",
       "      <td>-1.227352</td>\n",
       "      <td>1.224647</td>\n",
       "      <td>-0.651281</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-05</th>\n",
       "      <td>-0.482119</td>\n",
       "      <td>-0.098270</td>\n",
       "      <td>0.253447</td>\n",
       "      <td>1.218844</td>\n",
       "      <td>-0.736454</td>\n",
       "      <td>-0.072044</td>\n",
       "      <td>-0.810815</td>\n",
       "      <td>-0.125592</td>\n",
       "      <td>-0.162910</td>\n",
       "      <td>-0.944752</td>\n",
       "      <td>0.752244</td>\n",
       "      <td>0.669153</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               ret_0                                     x_1                                     x_2                              \n",
       "                 100       101       102       103       100       101       102       103       100       101       102       103\n",
       "2023-05-01  0.330437 -0.377605 -0.982188 -0.094828  0.345584  2.117839  0.054102  0.876242  0.821618 -1.112021  0.272791  0.256486\n",
       "2023-05-02  0.446375  0.663063 -0.466750 -2.250854 -1.303157  2.042772 -1.107373 -0.258848  0.905356  0.646703  0.199585  1.055743\n",
       "2023-05-03  0.364572  0.167465 -1.648787 -1.425349 -0.536953 -0.514006  0.235506 -0.138655  0.581118 -1.648075  0.759520  0.033000\n",
       "2023-05-04  0.546713 -0.683227 -0.297527  0.862445  0.294132  0.109014  0.254388  0.332814  0.028422 -1.227352  1.224647 -0.651281\n",
       "2023-05-05 -0.482119 -0.098270  0.253447  1.218844 -0.736454 -0.072044 -0.810815 -0.125592 -0.162910 -0.944752  0.752244  0.669153"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# The innermost column level consists of the asset ids.\n",
    "# The outermost column level is comprised of the column names (features,\n",
    "#  returns, price data, etc.).\n",
    "data = get_random_data(\n",
    "    n_assets,\n",
    "    n_features,\n",
    "    n_periods,\n",
    "    period_start,\n",
    "    freq,\n",
    "    rng_seed\n",
    ")\n",
    "display(data.head())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bd788e65",
   "metadata": {},
   "source": [
    "## Instantiate a node and DAG"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "566fb449",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.134622Z",
     "start_time": "2023-06-04T18:41:15.130450Z"
    }
   },
   "outputs": [],
   "source": [
    "# Here use instantiate a `DataSource` object, which is a subclass of a\n",
    "#. dataflow `Node`.\n",
    "# Dataflow supports various source nodes, such as\n",
    "# - a node that wraps a dataframe (as we use here)\n",
    "# - a node that wraps a function (where the output of the function is\n",
    "#   a dataframe)\n",
    "# - a node that loads data from disk\n",
    "# - a node that supplies data from our `MarketData` object\n",
    "node = dtfcore.DfDataSource(\"data\", data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "7e566e00",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.142740Z",
     "start_time": "2023-06-04T18:41:15.138663Z"
    }
   },
   "outputs": [],
   "source": [
    "# A DAG consists of (fully instantiated) Nodes along with the directed\n",
    "#  edges between them.\n",
    "# In this example notebook, we build a DAG incrementally.\n",
    "# For DAGs we re-use, we specify in code a `DagBuilder` object, which\n",
    "#  specifies how to build a DAG and provides a means to configure it.\n",
    "dag = dtfcore.DAG(name=\"simple_risk_model\", mode=\"loose\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "9ede3ad6",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.170576Z",
     "start_time": "2023-06-04T18:41:15.147820Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Here we use a convenience method for linear DAGs.\n",
    "dag.append_to_tail(node)\n",
    "# At each stage in this notebook, we draw the DAG.\n",
    "dtfcore.draw(dag)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "fd40d820",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.219107Z",
     "start_time": "2023-06-04T18:41:15.173157Z"
    }
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c29bbddcd0e0472f8c4ca65588ac7ef3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "run_leq_node:   0%|          | 0/1 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# We specify which node to run \"up to\" and whether to run in \"fit\" or\n",
    "#  \"predict\" mode.\n",
    "# All nodes that are less than or equal to the specified node\n",
    "#  (in the topological sense) are executed.\n",
    "data_df = dag.run_leq_node(\"data\", \"fit\")[\"df_out\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "993fab3d",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.239059Z",
     "start_time": "2023-06-04T18:41:15.221552Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0</th>\n",
       "      <th colspan=\"4\" halign=\"left\">x_1</th>\n",
       "      <th colspan=\"4\" halign=\"left\">x_2</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2023-05-01</th>\n",
       "      <td>0.330437</td>\n",
       "      <td>-0.377605</td>\n",
       "      <td>-0.982188</td>\n",
       "      <td>-0.094828</td>\n",
       "      <td>0.345584</td>\n",
       "      <td>2.117839</td>\n",
       "      <td>0.054102</td>\n",
       "      <td>0.876242</td>\n",
       "      <td>0.821618</td>\n",
       "      <td>-1.112021</td>\n",
       "      <td>0.272791</td>\n",
       "      <td>0.256486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-02</th>\n",
       "      <td>0.446375</td>\n",
       "      <td>0.663063</td>\n",
       "      <td>-0.466750</td>\n",
       "      <td>-2.250854</td>\n",
       "      <td>-1.303157</td>\n",
       "      <td>2.042772</td>\n",
       "      <td>-1.107373</td>\n",
       "      <td>-0.258848</td>\n",
       "      <td>0.905356</td>\n",
       "      <td>0.646703</td>\n",
       "      <td>0.199585</td>\n",
       "      <td>1.055743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-03</th>\n",
       "      <td>0.364572</td>\n",
       "      <td>0.167465</td>\n",
       "      <td>-1.648787</td>\n",
       "      <td>-1.425349</td>\n",
       "      <td>-0.536953</td>\n",
       "      <td>-0.514006</td>\n",
       "      <td>0.235506</td>\n",
       "      <td>-0.138655</td>\n",
       "      <td>0.581118</td>\n",
       "      <td>-1.648075</td>\n",
       "      <td>0.759520</td>\n",
       "      <td>0.033000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-04</th>\n",
       "      <td>0.546713</td>\n",
       "      <td>-0.683227</td>\n",
       "      <td>-0.297527</td>\n",
       "      <td>0.862445</td>\n",
       "      <td>0.294132</td>\n",
       "      <td>0.109014</td>\n",
       "      <td>0.254388</td>\n",
       "      <td>0.332814</td>\n",
       "      <td>0.028422</td>\n",
       "      <td>-1.227352</td>\n",
       "      <td>1.224647</td>\n",
       "      <td>-0.651281</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-05</th>\n",
       "      <td>-0.482119</td>\n",
       "      <td>-0.098270</td>\n",
       "      <td>0.253447</td>\n",
       "      <td>1.218844</td>\n",
       "      <td>-0.736454</td>\n",
       "      <td>-0.072044</td>\n",
       "      <td>-0.810815</td>\n",
       "      <td>-0.125592</td>\n",
       "      <td>-0.162910</td>\n",
       "      <td>-0.944752</td>\n",
       "      <td>0.752244</td>\n",
       "      <td>0.669153</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               ret_0                                     x_1                                     x_2                              \n",
       "                 100       101       102       103       100       101       102       103       100       101       102       103\n",
       "2023-05-01  0.330437 -0.377605 -0.982188 -0.094828  0.345584  2.117839  0.054102  0.876242  0.821618 -1.112021  0.272791  0.256486\n",
       "2023-05-02  0.446375  0.663063 -0.466750 -2.250854 -1.303157  2.042772 -1.107373 -0.258848  0.905356  0.646703  0.199585  1.055743\n",
       "2023-05-03  0.364572  0.167465 -1.648787 -1.425349 -0.536953 -0.514006  0.235506 -0.138655  0.581118 -1.648075  0.759520  0.033000\n",
       "2023-05-04  0.546713 -0.683227 -0.297527  0.862445  0.294132  0.109014  0.254388  0.332814  0.028422 -1.227352  1.224647 -0.651281\n",
       "2023-05-05 -0.482119 -0.098270  0.253447  1.218844 -0.736454 -0.072044 -0.810815 -0.125592 -0.162910 -0.944752  0.752244  0.669153"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# The data matches the input data.\n",
    "# In this notebook, all data is processed at once. However, dataflow\n",
    "#  supports the notion of a clock and supports running in an\n",
    "#  incremental mode.\n",
    "data_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "93b6f4c6",
   "metadata": {},
   "source": [
    "# Estimate model components"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2f0ed0f2",
   "metadata": {},
   "source": [
    "## Estimate beta"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "cea512f6",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.247463Z",
     "start_time": "2023-06-04T18:41:15.241197Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['ret_0', 'x_1', 'x_2'], dtype='object')"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cols = data.columns.levels[0]\n",
    "display(cols)\n",
    "x_vars = list(cols.difference([\"ret_0\"]))\n",
    "y_vars = [\"ret_0\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "da3c8d1c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.255209Z",
     "start_time": "2023-06-04T18:41:15.250669Z"
    }
   },
   "outputs": [],
   "source": [
    "# Use one of the dataflow nodes that wraps sklearn.\n",
    "# This node\n",
    "# - wraps supervised models\n",
    "# - learns a model for each asset independently (we also have a wrapper\n",
    "#   that pools across all assets)\n",
    "# The \"model_func\" specifies which sklearn model to use.\n",
    "# The sklearn model is configured through \"model_kwargs\", which are\n",
    "#  forwarded to the model.\n",
    "node_config = {\n",
    "    \"in_col_groups\": [(x,) for x in cols],\n",
    "    \"out_col_group\": (),\n",
    "    \"x_vars\": x_vars,\n",
    "    \"y_vars\": y_vars,\n",
    "    \"steps_ahead\": 1,\n",
    "    \"model_func\": slmode.LinearRegression,\n",
    "    \"model_kwargs\": {\n",
    "        \"fit_intercept\": False,\n",
    "    },\n",
    "    \"nan_mode\": \"drop\",\n",
    "}\n",
    "node = dtfcore.MultiindexSkLearnModel(\n",
    "    \"sklearn\",\n",
    "    **node_config,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "65d67a11",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.280019Z",
     "start_time": "2023-06-04T18:41:15.257364Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dag.append_to_tail(node)\n",
    "dtfcore.draw(dag)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "c2a9d797",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.424928Z",
     "start_time": "2023-06-04T18:41:15.282107Z"
    }
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f12a1e1866c1424b859a3c76c5b8556f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "run_leq_node:   0%|          | 0/2 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sklearn_df = dag.run_leq_node(\"sklearn\", \"fit\")[\"df_out\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "4f74039e",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.453701Z",
     "start_time": "2023-06-04T18:41:15.427435Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0.shift_-1</th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0.shift_-1_hat</th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0</th>\n",
       "      <th colspan=\"4\" halign=\"left\">x_1</th>\n",
       "      <th colspan=\"4\" halign=\"left\">x_2</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2023-05-01</th>\n",
       "      <td>0.446375</td>\n",
       "      <td>0.663063</td>\n",
       "      <td>-0.466750</td>\n",
       "      <td>-2.250854</td>\n",
       "      <td>0.039724</td>\n",
       "      <td>-0.003303</td>\n",
       "      <td>0.082740</td>\n",
       "      <td>-0.529041</td>\n",
       "      <td>0.330437</td>\n",
       "      <td>-0.377605</td>\n",
       "      <td>-0.982188</td>\n",
       "      <td>-0.094828</td>\n",
       "      <td>0.345584</td>\n",
       "      <td>2.117839</td>\n",
       "      <td>0.054102</td>\n",
       "      <td>0.876242</td>\n",
       "      <td>0.821618</td>\n",
       "      <td>-1.112021</td>\n",
       "      <td>0.272791</td>\n",
       "      <td>0.256486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-02</th>\n",
       "      <td>0.364572</td>\n",
       "      <td>0.167465</td>\n",
       "      <td>-1.648787</td>\n",
       "      <td>-1.425349</td>\n",
       "      <td>0.408234</td>\n",
       "      <td>0.329971</td>\n",
       "      <td>-1.242679</td>\n",
       "      <td>-1.199168</td>\n",
       "      <td>0.446375</td>\n",
       "      <td>0.663063</td>\n",
       "      <td>-0.466750</td>\n",
       "      <td>-2.250854</td>\n",
       "      <td>-1.303157</td>\n",
       "      <td>2.042772</td>\n",
       "      <td>-1.107373</td>\n",
       "      <td>-0.258848</td>\n",
       "      <td>0.905356</td>\n",
       "      <td>0.646703</td>\n",
       "      <td>0.199585</td>\n",
       "      <td>1.055743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-03</th>\n",
       "      <td>0.546713</td>\n",
       "      <td>-0.683227</td>\n",
       "      <td>-0.297527</td>\n",
       "      <td>0.862445</td>\n",
       "      <td>0.197211</td>\n",
       "      <td>-0.370850</td>\n",
       "      <td>0.326803</td>\n",
       "      <td>-0.004435</td>\n",
       "      <td>0.364572</td>\n",
       "      <td>0.167465</td>\n",
       "      <td>-1.648787</td>\n",
       "      <td>-1.425349</td>\n",
       "      <td>-0.536953</td>\n",
       "      <td>-0.514006</td>\n",
       "      <td>0.235506</td>\n",
       "      <td>-0.138655</td>\n",
       "      <td>0.581118</td>\n",
       "      <td>-1.648075</td>\n",
       "      <td>0.759520</td>\n",
       "      <td>0.033000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-04</th>\n",
       "      <td>-0.482119</td>\n",
       "      <td>-0.098270</td>\n",
       "      <td>0.253447</td>\n",
       "      <td>1.218844</td>\n",
       "      <td>-0.059698</td>\n",
       "      <td>-0.226907</td>\n",
       "      <td>0.384520</td>\n",
       "      <td>0.695936</td>\n",
       "      <td>0.546713</td>\n",
       "      <td>-0.683227</td>\n",
       "      <td>-0.297527</td>\n",
       "      <td>0.862445</td>\n",
       "      <td>0.294132</td>\n",
       "      <td>0.109014</td>\n",
       "      <td>0.254388</td>\n",
       "      <td>0.332814</td>\n",
       "      <td>0.028422</td>\n",
       "      <td>-1.227352</td>\n",
       "      <td>1.224647</td>\n",
       "      <td>-0.651281</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-05</th>\n",
       "      <td>-0.292457</td>\n",
       "      <td>-0.506292</td>\n",
       "      <td>-1.481818</td>\n",
       "      <td>-1.514319</td>\n",
       "      <td>0.136685</td>\n",
       "      <td>-0.190286</td>\n",
       "      <td>-0.862632</td>\n",
       "      <td>-0.769797</td>\n",
       "      <td>-0.482119</td>\n",
       "      <td>-0.098270</td>\n",
       "      <td>0.253447</td>\n",
       "      <td>1.218844</td>\n",
       "      <td>-0.736454</td>\n",
       "      <td>-0.072044</td>\n",
       "      <td>-0.810815</td>\n",
       "      <td>-0.125592</td>\n",
       "      <td>-0.162910</td>\n",
       "      <td>-0.944752</td>\n",
       "      <td>0.752244</td>\n",
       "      <td>0.669153</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           ret_0.shift_-1                               ret_0.shift_-1_hat                                   ret_0                                     x_1                                     x_2                              \n",
       "                      100       101       102       103                100       101       102       103       100       101       102       103       100       101       102       103       100       101       102       103\n",
       "2023-05-01       0.446375  0.663063 -0.466750 -2.250854           0.039724 -0.003303  0.082740 -0.529041  0.330437 -0.377605 -0.982188 -0.094828  0.345584  2.117839  0.054102  0.876242  0.821618 -1.112021  0.272791  0.256486\n",
       "2023-05-02       0.364572  0.167465 -1.648787 -1.425349           0.408234  0.329971 -1.242679 -1.199168  0.446375  0.663063 -0.466750 -2.250854 -1.303157  2.042772 -1.107373 -0.258848  0.905356  0.646703  0.199585  1.055743\n",
       "2023-05-03       0.546713 -0.683227 -0.297527  0.862445           0.197211 -0.370850  0.326803 -0.004435  0.364572  0.167465 -1.648787 -1.425349 -0.536953 -0.514006  0.235506 -0.138655  0.581118 -1.648075  0.759520  0.033000\n",
       "2023-05-04      -0.482119 -0.098270  0.253447  1.218844          -0.059698 -0.226907  0.384520  0.695936  0.546713 -0.683227 -0.297527  0.862445  0.294132  0.109014  0.254388  0.332814  0.028422 -1.227352  1.224647 -0.651281\n",
       "2023-05-05      -0.292457 -0.506292 -1.481818 -1.514319           0.136685 -0.190286 -0.862632 -0.769797 -0.482119 -0.098270  0.253447  1.218844 -0.736454 -0.072044 -0.810815 -0.125592 -0.162910 -0.944752  0.752244  0.669153"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sklearn_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "91949bfa",
   "metadata": {},
   "source": [
    "## Estimate residual returns variances"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "23408b9e",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.485352Z",
     "start_time": "2023-06-04T18:41:15.456541Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0.shift_-1</th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0.shift_-1_hat</th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0</th>\n",
       "      <th colspan=\"4\" halign=\"left\">x_1</th>\n",
       "      <th colspan=\"4\" halign=\"left\">x_2</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2023-05-01</th>\n",
       "      <td>0.446375</td>\n",
       "      <td>0.663063</td>\n",
       "      <td>-0.466750</td>\n",
       "      <td>-2.250854</td>\n",
       "      <td>0.039724</td>\n",
       "      <td>-0.003303</td>\n",
       "      <td>0.082740</td>\n",
       "      <td>-0.529041</td>\n",
       "      <td>0.330437</td>\n",
       "      <td>-0.377605</td>\n",
       "      <td>-0.982188</td>\n",
       "      <td>-0.094828</td>\n",
       "      <td>0.345584</td>\n",
       "      <td>2.117839</td>\n",
       "      <td>0.054102</td>\n",
       "      <td>0.876242</td>\n",
       "      <td>0.821618</td>\n",
       "      <td>-1.112021</td>\n",
       "      <td>0.272791</td>\n",
       "      <td>0.256486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-02</th>\n",
       "      <td>0.364572</td>\n",
       "      <td>0.167465</td>\n",
       "      <td>-1.648787</td>\n",
       "      <td>-1.425349</td>\n",
       "      <td>0.408234</td>\n",
       "      <td>0.329971</td>\n",
       "      <td>-1.242679</td>\n",
       "      <td>-1.199168</td>\n",
       "      <td>0.446375</td>\n",
       "      <td>0.663063</td>\n",
       "      <td>-0.466750</td>\n",
       "      <td>-2.250854</td>\n",
       "      <td>-1.303157</td>\n",
       "      <td>2.042772</td>\n",
       "      <td>-1.107373</td>\n",
       "      <td>-0.258848</td>\n",
       "      <td>0.905356</td>\n",
       "      <td>0.646703</td>\n",
       "      <td>0.199585</td>\n",
       "      <td>1.055743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-03</th>\n",
       "      <td>0.546713</td>\n",
       "      <td>-0.683227</td>\n",
       "      <td>-0.297527</td>\n",
       "      <td>0.862445</td>\n",
       "      <td>0.197211</td>\n",
       "      <td>-0.370850</td>\n",
       "      <td>0.326803</td>\n",
       "      <td>-0.004435</td>\n",
       "      <td>0.364572</td>\n",
       "      <td>0.167465</td>\n",
       "      <td>-1.648787</td>\n",
       "      <td>-1.425349</td>\n",
       "      <td>-0.536953</td>\n",
       "      <td>-0.514006</td>\n",
       "      <td>0.235506</td>\n",
       "      <td>-0.138655</td>\n",
       "      <td>0.581118</td>\n",
       "      <td>-1.648075</td>\n",
       "      <td>0.759520</td>\n",
       "      <td>0.033000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-04</th>\n",
       "      <td>-0.482119</td>\n",
       "      <td>-0.098270</td>\n",
       "      <td>0.253447</td>\n",
       "      <td>1.218844</td>\n",
       "      <td>-0.059698</td>\n",
       "      <td>-0.226907</td>\n",
       "      <td>0.384520</td>\n",
       "      <td>0.695936</td>\n",
       "      <td>0.546713</td>\n",
       "      <td>-0.683227</td>\n",
       "      <td>-0.297527</td>\n",
       "      <td>0.862445</td>\n",
       "      <td>0.294132</td>\n",
       "      <td>0.109014</td>\n",
       "      <td>0.254388</td>\n",
       "      <td>0.332814</td>\n",
       "      <td>0.028422</td>\n",
       "      <td>-1.227352</td>\n",
       "      <td>1.224647</td>\n",
       "      <td>-0.651281</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-05</th>\n",
       "      <td>-0.292457</td>\n",
       "      <td>-0.506292</td>\n",
       "      <td>-1.481818</td>\n",
       "      <td>-1.514319</td>\n",
       "      <td>0.136685</td>\n",
       "      <td>-0.190286</td>\n",
       "      <td>-0.862632</td>\n",
       "      <td>-0.769797</td>\n",
       "      <td>-0.482119</td>\n",
       "      <td>-0.098270</td>\n",
       "      <td>0.253447</td>\n",
       "      <td>1.218844</td>\n",
       "      <td>-0.736454</td>\n",
       "      <td>-0.072044</td>\n",
       "      <td>-0.810815</td>\n",
       "      <td>-0.125592</td>\n",
       "      <td>-0.162910</td>\n",
       "      <td>-0.944752</td>\n",
       "      <td>0.752244</td>\n",
       "      <td>0.669153</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           ret_0.shift_-1                               ret_0.shift_-1_hat                                   ret_0                                     x_1                                     x_2                              \n",
       "                      100       101       102       103                100       101       102       103       100       101       102       103       100       101       102       103       100       101       102       103\n",
       "2023-05-01       0.446375  0.663063 -0.466750 -2.250854           0.039724 -0.003303  0.082740 -0.529041  0.330437 -0.377605 -0.982188 -0.094828  0.345584  2.117839  0.054102  0.876242  0.821618 -1.112021  0.272791  0.256486\n",
       "2023-05-02       0.364572  0.167465 -1.648787 -1.425349           0.408234  0.329971 -1.242679 -1.199168  0.446375  0.663063 -0.466750 -2.250854 -1.303157  2.042772 -1.107373 -0.258848  0.905356  0.646703  0.199585  1.055743\n",
       "2023-05-03       0.546713 -0.683227 -0.297527  0.862445           0.197211 -0.370850  0.326803 -0.004435  0.364572  0.167465 -1.648787 -1.425349 -0.536953 -0.514006  0.235506 -0.138655  0.581118 -1.648075  0.759520  0.033000\n",
       "2023-05-04      -0.482119 -0.098270  0.253447  1.218844          -0.059698 -0.226907  0.384520  0.695936  0.546713 -0.683227 -0.297527  0.862445  0.294132  0.109014  0.254388  0.332814  0.028422 -1.227352  1.224647 -0.651281\n",
       "2023-05-05      -0.292457 -0.506292 -1.481818 -1.514319           0.136685 -0.190286 -0.862632 -0.769797 -0.482119 -0.098270  0.253447  1.218844 -0.736454 -0.072044 -0.810815 -0.125592 -0.162910 -0.944752  0.752244  0.669153"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sklearn_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "43dc655d",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.495036Z",
     "start_time": "2023-06-04T18:41:15.488396Z"
    }
   },
   "outputs": [],
   "source": [
    "# The sklearn modeling node does not automatically perform the\n",
    "#  residualization. (We do have a different sklearn node that\n",
    "#  learns a factor model (e.g., PCA) and automatically\n",
    "#  residualizes.)\n",
    "# Here we create a node to do the column arithmetic.\n",
    "node_config = {\n",
    "    \"in_col_groups\": [\n",
    "        (\"ret_0.shift_-1\",), \n",
    "        (\"ret_0.shift_-1_hat\",),\n",
    "    ],\n",
    "    \"out_col_group\": (),\n",
    "    \"transformer_func\": cofeatur.combine_columns,\n",
    "    \"transformer_kwargs\": {\n",
    "        \"term1_col\": \"ret_0.shift_-1\",\n",
    "        \"term2_col\": \"ret_0.shift_-1_hat\",\n",
    "        \"out_col\": \"residual.shift_-1\",\n",
    "        \"operation\": \"sub\"\n",
    "    }\n",
    "}\n",
    "node = dtfcore.GroupedColDfToDfTransformer(\n",
    "    \"residualize\",\n",
    "    **node_config,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "9ee2ca6b",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.523294Z",
     "start_time": "2023-06-04T18:41:15.497636Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAK8AAAD7CAYAAAAcjcSUAAAABmJLR0QA/wD/AP+gvaeTAAAgAElEQVR4nO2de1xU1d7/P3NzLjAygDDcUUQoEw0IFMVAAdGCwFt5UrKLHfNYnlOPZadz8ulk2Sk7ZVbnZJ2np+sxM4W8Kyp5QTHjaioXReIOw20cLgPMzPf3hz/24zh4w2E2e9zv12u/GNZee63P7PnsNWvvWWt9BURE4OHhIEK2BfDwDBTevDychTcvD2cRsy3AHtDpdKirq4NOp0NraysAoKOjAz09PRAIBFCpVAAAR0dHKJVKuLm5wc3NDQKBgE3ZnIc3701y8eJFFBQUoKysjNnKy8uh0Wig1+tvuTyRSAQ3Nzd4eXkhKCgIY8aMQXBwMMaOHYuQkBCIxfxHcyME/NMGS3p7e3HixAlkZWXh1KlT+Pnnn6HRaCAQCODr64sxY8ZgzJgxGDVqFDw8PODm5gYPDw84OzvDyckJAKBQKCCVSmEymaDVagEAnZ2daGtrg0ajQV1dHTQaDaqqqlBaWorS0lKUl5ejp6cHcrkcoaGhiIyMRHR0NOLi4pjWm+f/4M37/9FoNNi6dSv27t2LQ4cOQafTwd/fH1FRUYiMjERkZCRCQ0OhUCgGTYPRaERxcTFzwZw8eRKFhYUAgMjISCQmJmLu3LkYN27coGngEne0eTs6OrB161Zs2rQJBw4cgEwmQ3x8PBISEjBjxgwEBgayLREtLS04ePAg9u/fj71796K6uhohISH43e9+h4ULF8LPz49tiexBdyAXLlygVatWkaurK4lEIoqPj6cvv/ySdDod29JuyC+//EIrVqwgtVpNQqGQ4uPjafv27WQymdiWZnPuKPOePXuW5s+fT0KhkPz8/Oitt94ijUbDtqwB0dPTQ5s3b6bo6GgCQBMmTKCMjAy2ZdmUO8K8v/32Gy1evJhEIhGFhITQli1bqLe3l21ZViM3N5fmzZtHAoGAJk6cSAcPHmRbkk2wa/P29vbSu+++Sw4ODhQYGEjffvstGY1GtmUNGrm5uTRr1iwCQAsXLqSGhga2JQ0qdmve06dP07333ktSqZT+9re/kV6vZ1uSzdi+fTv5+fmRi4sLffvtt2zLGTTs0rxfffUVOTg40OTJk6m4uJhtOayg0+noueeeI4FAQM8884xdXrx2ZV6j0ch8YC+88IJd9WsHytatW8nJyYkiIiLsrhthN+bt7e2lhQsXklQqpe+//55tOUOK0tJSCgwMpODgYKqsrGRbjtWwix8pTCYTHn74YezduxcZGRmIj49nW9KQo76+HjNmzMClS5eQnZ0Nb29vtiXdNnZh3pdeegkffvghMjMzER0dzbacIUtLSwumTp0KmUyGo0ePDupP3TaB3Yb/9vnqq69IIBDQ119/zbYUTnDhwgVyc3Oj+fPnsy3ltuG0eWtqakilUtGf/vSnQatj06ZNBIAAkFQqHbR6bMmhQ4dIIBDQd999x7aU24LT3YY5c+agqKgIRUVFg/4VGB8fj2PHjg1o7O5QZOnSpfjxxx9x9uxZuLi4sC1nQHB2GlBubi7S09Px0Ucfcabv5ujoOGT65G+//TaMRiM2bNjAtpQBw1nzvvfeexg/fjwSExPZlsJJVCoVli1bhn/+85/o6upiW86A4KR529rasGXLFrzwwgv8PLDbYPny5dBqtUhPT2dbyoDgpHmzsrJgMpmQnJxs9bKLi4uRmpoKJycnODg4YOrUqTh27Fi/eQ0GAzZv3oyEhAR4eHhALpcjJCQEH3zwAUwmE5Pv3XffhUAgQEdHB7KzsyEQCCAQCMzmqd1sWdZErVYjKioKBw4cGJTyBx227xgHwrPPPksRERFWL7esrIxUKhV5e3vT/v37SafTUVFREc2YMYNGjhxp8bRhx44dBIDWrl1LLS0tpNFoaMOGDSQUCmnlypUW5Ts4ONCUKVP6rftWy7IWa9asIV9f30ErfzDhpHlnzpxJTz75pNXLnT9/PgGgH374wSy9pqaGpFJpv+aNjY21KGfRokUkkUhIq9Wapd/IvLdSlrXIyMggAJwcuMPJbkNjYyPc3d2tXu7evXsBwOImsG96+tUkJSUhKyvLIn3ChAno7e3FmTNnbrpua5Z1K/Sdx8bGxkEpfzDh5OIAHR0dVn881t3dDZ1OB5lMBkdHR4v97u7uKC0tNUvTarX4xz/+gfT0dFRXV6Otrc1sf2dn503Xb82ybgWlUgkAaG9vH5TyBxNOtrwjRoxAc3OzVcuUSqVQKpXQ6/X9fpAtLS0WacnJyVizZg2efvpplJaWwmQygYjw/vvvAwDoqt9/rvdk5FbLshYajQYA4ObmNijlDyacNK+7uzvq6+utXu6sWbMA/F/3oY+mpiaUlJSYpRmNRmRnZ8PDwwMrVqwwW77pWs9NFQoFenp6mP+Dg4Px6aefDqgsa1FXVwexWMzJX9k4ad7Q0FBkZ2dbvdy1a9fCxcUFf/rTn5CZmYn29nacPXsWixYtsuhKiEQixMbGor6+HuvWrUNTUxO6urqQlZWFTz75pN/yw8LCUFpaiqqqKpw4cQLl5eWYOnXqgMqyFsePH8e9994LoZCDVmD3fnFgHD9+nAAMyhSfkpISSk1NpeHDh5NcLqeIiAjauXMnxcXFMQN0nnrqKSIi0mg0tHTpUvL19SWJREJqtZoef/xxevnll5m84eHhTNnFxcU0depUcnBwIF9fX/r444+ZfbdalrW46667aNWqVVYv1xZwcmCOwWCAj48PlixZgjfeeINtOZzl5MmTmDRpEo4dO4YpU6awLeeW4aR5AWDNmjVYv349Kisr4eDgwLYcTjJ//nxUVFTg1KlTbEsZEBzs6Fxm2bJl6O7uxvr169mWwkny8vKQnp6OF198kW0pA4azLS8ArFu3Dq+++iry8/Nx9913sy2HMxiNRkycOBEODg746aefODu4idPmNRgMmDRpEiQSCbKysiCTydiWxAn++7//G++88w4KCgoQHBzMtpyBw969onUoLi4mZ2dnWrhw4R25UuKtsmnTJhIIBLRx40a2pdw2nDcvEVFmZiaJxWJ66aWX2JYypNm/fz/JZDJ6/vnn2ZZiFezCvEREX3/9NYnFYlq+fLldL6Y3UNLT00kqlVJaWhoZDAa25VgFuzEvEdG2bdtIKpXS/Pnz6dKlS2zLGTJ88MEHJBaL6Q9/+INdXdh2ZV6iy9O63d3dKTg4mE6fPs22HFa5dOkSPfzwwyQSiWjt2rVsy7E6dmdeIqLq6mqKjo4muVxOf//736mnp4dtSTZn7969FBAQQGq12m4Xm7ZL8xJdXvb+zTffJLlcTuPHj6fs7Gy2JdmE2tpaWrRoEQGgefPmUV1dHduSBg27NW8fZWVlFB8fTwAoOTmZCgsL2ZY0KDQ1NdFLL71ECoWC/Pz8aPv27WxLGnTs3rx97Nmzh8LDw0koFNLcuXPp8OHDbEuyChcvXqQXX3yRhg8fTm5ubvTee+9RV1cX27Jswh1jXiIik8lEW7dupaioKCaCzsaNG6mlpYVtabdEb28v7dmzh2bPnk0ikYh8fHzorbfe4kQoLmtyR5n3Sk6dOkVpaWkkk8lIKpXSQw89RJs2bRq0Wbq3S29vL/3000/0hz/8gdzc3EggENDUqVNp8+bNd+wK8Jwe22ANLl26hIyMDCYKJgBERUUhMTERcXFxCAsLw7Bhw1jRVlZWhqysLOzbtw8HDx6EVqtFSEgIFixYgN/97ncYNWoUK7qGCne8ea+kubkZmZmZ2L9/P/bt24fa2lpIpVJMmDCBiT0cHByMoKAgq05Y7OjoYCLJFxUVMbGHW1tb4eDggNjYWCQmJiIxMbHfKfh3Krx5r0NfEOs+MxUVFTETIlUqFUaNGgVPT0+4ubnB3d0drq6uUCqVEIvFkEqlUCgU6O3tZWYjt7W1oaOjAw0NDaivr4dGo0FlZSWqq6sBXJ4XFxgYiIiICERGRiIiIgKhoaGQSqWsnYOhDG/eW4CIUFVVhdLSUpSVleHixYtobGyERqNBQ0MDWltb0d7ejt7eXuj1enR1dUEsFjNrIzg7O0OhUMDd3R0eHh5wc3ODt7c3xowZg+DgYIwePZq1LgoX4c07SJhMJohEImzZsgXz5s1jW45dwtlpQDw8vHl5OAtvXh7OwpuXh7Pw5uXhLLx5eTgLb14ezsKbl4ez8Obl4Sy8eXk4C29eHs7Cm5eHs/Dm5eEsvHl5OAtvXh7OwpuXh7Pw5uXhLLx5eTgLb14ezsKbl4ez8Obl4Sy8eXk4C29eHs7Cm5eHs/Dm5eEsvHl5OAtvXh7OwpuXh7Pw5uXhLLx5eTgLb14ezsKbl4ez8ItLW4m0tDQUFBSYpZWWlsLLywuOjo5MmkQiwe7du+Hh4WFriXaHmG0B9kJQUBC++eYbi/SKigqz/0NCQnjjWgm+22AlHn30UQgEguvmkUgkWLx4sY0U2T98t8GKhIWFobCwECaTqd/9AoEAlZWV8PHxsbEy+4Rvea3IY489BqGw/1MqFAoRFRXFG9eK8Oa1IgsWLLhuq/vYY4/ZWJF9w5vXinh4eGDq1KkQiUT97udDWlkX3rxWJi0tzSJNJBIhISEBrq6uLCiyX3jzWpm5c+da9HuJqF9T89wevHmtjEqlwsyZMyEW/98jdIlEgoceeohFVfYJb95BYNGiRTAajQAAsViM1NRUs1/ZeKwDb95B4KGHHoJcLgcAGI1GLFy4kGVF9glv3kFAJpNh9uzZAABHR0ckJiayrMg+4cc2DBC9Xo+2tjZotVq0t7ejvb0dvb29AACDwQA/Pz8AQGRkJPbt28e0xADg7OyMYcOGQaVSQaVSQalUsvIeuA7/8/BVaLValJWVobKyEtXV1aitrUVtbS1qampQV1eHlpYWaLVa6PV6q9UpFAqhUqng4uICtVoNLy8veHl5wdvbG56envDz80NgYCC8vLysVqc9cMeat7GxEfn5+SgsLERpaSmzNTQ0ALj8i5harYanpye8vb0ZQ7m4uMDJyQkqlYr5q1QqIZfLIZPJmGNVKhXWrFmDV155Bd3d3eju7gZwuQ986dIl6PV6aLVapvVua2tDS0sLGhoamAuluroa9fX16OnpAXC5CxIUFIQxY8ZgzJgxCAkJQWhoKAIDA284KMgeuSPMq9PpcOzYMZw4cQL5+fnIz89HTU0NAMDHxwd33XUXgoKCzDY/Pz9IJJLbqtdgMJg9MhsodXV1KCsrQ2lpKfO3pKQEZWVlMBgMGD58OCZMmIDQ0FBEREQgJiYGvr6+t13vUMcuzdve3o6srCwcPnwYR44cQV5eHoxGI4KCghAaGoqwsDCEhoYiNDQUI0aMYFvugNHr9SgqKkJ+fj7y8vKYb5Kenh6MGjUKMTExiImJQVxcnF2a2W7MW1lZib1792LHjh3IzMxEd3c3AgICEB8fjylTpmDatGl2+QFeTWdnJ/Ly8pCdnY0DBw4gOzsbXV1dGDt2LJKTk5GUlITJkydfc/Qbl+C0eevr6/Gf//wH33zzDfLz86FUKjFjxgwkJSXhgQcegLu7O9sSWUev1+Pw4cPYvn07du7cicrKSnh6euKRRx7BY489htDQULYlDhziGN3d3fTdd9/RAw88QGKxmJycnGjJkiW0b98+0uv1bMsb8hQUFNBrr71GgYGBBIDGjRtH69ato4aGBral3TKcMW9bWxutX7+efH19SSQSUXx8PH355ZfU3t7OtjTO8ssvv9CKFSvIzc2Nhg0bRmlpaVRUVMS2rJtmyJu3urqali1bRgqFglQqFb300ktUWVnJtiy7orOzkzZu3Eh33303CQQCSkhIoGPHjrEt64YMWfO2tbXRn//8Z1IoFOTn50cffvgh6XQ6tmXZNSaTifbs2UOxsbEEgFJSUujMmTNsy7omQ868JpOJPv30U3J1dSVXV1d69913qauri21Zdxx79uyhCRMmkEgkomeeeYa0Wi3bkiwYUuatqKighIQEEolE9F//9V/U2trKtqQ7GqPRSF9++SW5u7uTr68v7d27l21JZgwZ8/7nP/8hpVJJY8eOpZMnT7Ith+cKGhsb6ZFHHiGBQEDLli2jnp4etiUR0RAwr8lkoldffZUEAgH98Y9/5B93DWG2bNlCSqWSYmNjqampiW057Jq3t7eXHnnkEZJIJPTZZ5+xKYXnJiksLCR/f38KDAykiooKVrWwZl6TyURPPvkkKRQKOnTokM3rX7duHQEgAOTt7W21vHcC9fX1dO+991JwcDBpNBrWdLBm3j//+c8kkUho165dbEkgIqIJEybctCFvJa+9U1dXRwEBARQZGUkdHR2saGBldMbhw4fx97//HRs3bsQDDzzAhgSe28TDwwP79u1DWVkZVq9ezYoGm5u3u7sby5Ytw6xZs/DEE0/YunoeKxIYGIh169Zh/fr1+OWXX2xev83N++mnn6KqqgobN260ddU8g8CTTz6JqVOnYtWqVTav2+bm/fzzz/Hoo48O6mqJ3d3dWL16Ne666y4oFAq4uLggOTkZ27dvZ9ZTuBbffPMNBAKB2VZfX3/dYzQaDVasWIGRI0di2LBhcHNzw5w5cyxWSjcYDNi8eTMSEhLg4eEBuVyOkJAQfPDBB2YL9GVkZJjVX1JSgocffhiurq5M2r///W+zPBUVFXjkkUegUqng6uqKpKQkXLhwYeAn8SYRCARYuXIlsrKybFKfGbbsYBcUFBAAOn78+KDWs2TJEnJycqL9+/dTZ2cn1dfX08qVKwkAZWVlmeW9+ibMYDDQCy+8QAkJCdTS0nLdvEREtbW15O/vT2q1mnbt2kU6nY5+/fVXiomJIZlMZvZed+zYQQBo7dq11NLSQhqNhjZs2EBCoZBWrlxp8T5SUlIIAMXExFBWVhZ1dHRQTk4OiUQi5i6/L09KSgodP36c2tvbKTMzk+RyOUVERNzuqbwpDAYDeXl50d/+9jeb1NeHTc370UcfkYuLy6DXM2rUKJo8ebJFelBQ0HXN29raSomJifTHP/6RDAaDxfH9mXfx4sUEgL799luz9Lq6OpJKpRQeHs6k7dixg2JjYy3KXbRoEUkkEovxA33G3L179zXfa1+eHTt2mKXPmzePANjsUdYTTzxB8fHxNqmrD5t2GyoqKhAYGDjo9cycORPHjx/H73//e+Tk5DBdhZKSEsTGxvZ7TElJCSZOnAihUIj169dfc5nSq8nIyIBQKERSUpJZuoeHB+655x7k5uaiuroaAJCUlISsrCyLMiZMmIDe3l6cOXOm3zoiIyNvqCMiIsLs/74pT7W1tTf1Pm6XMWPG4OLFizapqw+bmrerq8ts8Y3B4uOPP8ZXX32F8vJyxMXFYfjw4Zg5cybS09P7zd/a2orU1FT4+Phgz549/QZG6Y/u7m5otVqYTCY4OTlZ9JXz8vIAAGVlZQAurwmxevVqhISEwNnZmcn34osvArg8/6w/HBwcbqjFycnJ7P9hw4YBwDUXu7Y2CoUCXV1dNqmrD5uaV6VSoaWlZdDrEQgESEtLw4EDB9DW1oaMjAwQEebMmYP33nvPIr9YLMaBAwfw448/IiQkBE8//TROnTp1w3qkUilUKhXEYjF6e3tBl7thFtu0adMAAMnJyVizZg2efvpplJaWwmQygYjw/vvvA7i8FCpXaW5uhrOzs03rtKl5x44di5KSkkG/QlUqFYqLiwFcXl40ISGBuYPftWuXRX6lUglvb284Ojpi+/btcHR0RGpqKurq6m5Y15w5c2AwGJCdnW2x7+2334afnx8MBgOMRiOys7Ph4eGBFStWwM3NjVkoxNYt1mCQn5+PsWPH2rROm5q3b8G5a319W5NnnnkGRUVF6O7uRmNjI9555x0QEaZPn37d40aOHIkffvgBGo0Gc+bMYVa6uRZvvfUWRo8ejSeffBJ79uyBVqtFS0sLNm7ciNdffx3vvvsuxGIxRCIRYmNjUV9fj3Xr1qGpqQldXV3IysrCJ598Ys23bnNaWlqQmZmJBx980LYV2/T2kIhmz55NcXFxg1pHQUEBLV26lO6++25SKBTk4uJCkyZNos8++4xMJhMREW3atIkZbNO3vf/++3TixAmL9P62v/zlL0x9zc3N9MILL1BAQABJJBJyc3OjGTNmUGZmppkujUZDS5cuJV9fX5JIJKRWq+nxxx+nl19+mSk3PDz8mhqupL88fZquTn/wwQcH9Xxv2LCBHB0dbT5Ny+bmzczMJAC0Z88eW1fNMwi0tLSQp6cnLV++3OZ1s7LoyIIFC5CTk4Nff/2VXzGc4yxZsgS7du3C2bNnbX7DxsqQyPr6enJ1daX58+eT0WhkQwKPFfjiiy9IIBDQ1q1bWamftfG8J06cIIVCQc888wxbEnhug507d5JYLKa//vWvrGlgdRrQtm3bSCQS0bPPPku9vb1sSuG5BbZt20YKhYKeeuop5gaYDVifgPn999+TQqGgxMREamtrY1sOz3UwmUy0du1aEgqFtGzZMtYbHNbNS0R06tQp8vLyoqCgIDpx4gTbcnj6oaGhgebOnUtisZg+/PBDtuUQ0RAxLxFRTU0NJSYmkkgkopUrV1JnZyfbknj+P9999x25ubmRv78/K5Nlr8WQMS/R5a+lf//73+Tk5ESBgYH0/fffs9qnutMpKiqiBx54gAQCAS1dupQuXbrEtiQzhpR5+6iqqqJFixaRUCikyMjIIXW13wn89ttvtHjxYhIKhRQWFmYxBnqoMCTN28eZM2do/vz5zM+mX375Jes3CfZMQUEB/f73vyeZTEb+/v60cePGIf0cfkibt48jR45QSkoKCYVCCggIoPfff5+am5vZlmUX9PT0UEZGBk2fPp0AUEhICP3P//zPkFmP7Hpwwrx9lJWV0bPPPkuOjo40bNgwSk1NpW3btlF3dzfb0jjHzz//zKyKLhQKadasWbR//35O3WNwMqBKe3s7tm3bhq+//hqHDh2CSqVCSkoKkpKSMGPGDH68RD8YjUacPHkSO3bsQEZGBoqLixEcHIy0tDQsWrQI/v7+bEu8ZThp3iuprq7Gpk2bkJGRgZMnT0IsFiM2NhZJSUmIjY3FPffcc0dGhwQuBx88fPgw9u3bh127dkGj0WD06NFITk7GggULMHHiRLYl3hacN++VaDQa7Nq1Czt37kRmZiYuXbqEESNGIDo6GjExMYiOjsb48eOZ+V32BBHhwoULOH78OI4cOYIjR46grKwMYrEYEydORHJyMpKTk20+22EwsSvzXonRaER+fj6OHj2Kn376CceOHUNLSwskEgnuuecehIWFISwsDPfeey/uuusuuLq6si35ptHr9SgtLUVRURET+TI/Px9arRZSqRQRERG4//77cf/992PKlCl2242yW/NejclkQmlpKfLy8pgPPC8vD21tbQAAFxcXs9jD/v7+8PHxgYeHB3x9fW9qBq+16OnpQX19Paqrq1FXV4fKykqUlZUxW2VlJYgIw4YNY4Jn94WknTBhgk1maA8F7hjzXouKigrGFH2R38vKylBVVWU2f02pVMLHxwfOzs5MxPe+qO8qlYqJ9N7H8OHDmbUfLl26xKwd0dnZie7ubnR0dFhEfW9tbUVDQwMTeR4AhEIh1Gq1WbT3MWPGMBfZ7Qb35jJ3vHmvR2NjI+rr61FVVYW6ujrU1NSgtbWVMVuf8bRaLQwGA3Q6HXNsa2sr89rR0ZExmUwmg1wuh1wuZ4zfdxE4OztDrVabtfhqtfqONuj14M07SJhMJohEImzZsgXz5s1jW45dwv3Q3zx3LLx5eTgLb14ezsKbl4ez8Obl4Sy8eXk4C29eHs7Cm5eHs/Dm5eEsvHl5OAtvXh7OwpuXh7Pw5uXhLLx5eTgLb14ezsKbl4ez8Obl4Sy8eXk4C29eHs7Cm5eHs/Dm5eEsvHl5OAtvXh7OwpuXh7Pw5uXhLLx5eTgLb14ezsKbl4ez8Obl4Sy8eXk4C29eHs7Cm5eHs/CLS1uJtLQ0FBQUmKWVlpbCy8vLLKCJRCLB7t274eHhYWuJdoeYbQH2QlBQEL755huL9IqKCrP/Q0JCeONaCb7bYCUeffTRGwYrlEgkWLx4sY0U2T98t8GKhIWFobCwECaTqd/9AoEAlZWV8PHxsbEy+4Rvea3IY489BqGw/1MqFAoRFRXFG9eK8Oa1IgsWLLhuq/vYY4/ZWJF9w5vXinh4eGDq1KlM8MCr4UNaWRfevFYmLS3NIk0kEiEhIYFT8Y25AG9eKzN37lyLfi8R9WtqntuDN6+VUalUmDlzJsTi/3uELpFI8NBDD7Goyj7hzTsILFq0iAmULRaLkZqaavYrG4914M07CDz00EOQy+UAAKPRiIULF7KsyD7hzTsIyGQyzJ49G8DliO+JiYksK7JP+LENt0BbWxu6urrQ1dWF1tZWAIBOp4PBYLDI5+fnBwCIjIzEjz/+CKVSadYPBi73jwUCARwdHSGXy6FUKuHo6AiJRGKbN8Rx7rifh00mExoaGlBdXY2GhgY0NzejpaUFzc3NaG5uRlNTE/O6vb3dzLC2QiwWM0Z2dHSEq6srXFxc4OrqymwjRoyAq6sr3Nzc4OXlBW9vb0ilUptpHArYnXk7Ojpw/vx5XLhwARcvXkRVVRVqampQW1uLyspKNDQ0oLe3l8kvl8sZQ7i4uGDEiBGMMRwdHeHs7AyZTAa5XA6VSgWZTAaFQgEnJycIhULI5XLIZDIzDX2t7Jo1a/DKK69AJBKhra0NV55qo9GIS5cuAbjcend1daG9vR06nQ56vR46nQ7t7e1ob29nLqa+i6ypqQlNTU3M8X24u7vDy8sLPj4+8Pb2hpeXFwICAjB69GiMHj0a7u7ug3jmbQ8nzWsymXDhwgUUFhaiuLiYMev58+dRX18P4PJYAi8vL/j5+TEtk6+vLzw9PeHr6wtvb294eHhAoVAMmk6DwWDRVbAmvb290Gg0qK6uRl1dHSorK1FXV4fq6mpUV1ejtrYWFRUV6O7uBnD5ogoMDGTMHBwcjPHjx2Ps2LHMDSaXGPLm1Wq1yPQMfcIAAA+VSURBVM3NRVFREX799VcUFhbi7Nmz6OzshFAoxKhRoxAYGMh8KH2vAwIC7riv0f4wmUyoqqpiLu4r/5aUlECv10MkEiEwMBAhISEYP348QkJCEB4eDl9fX7blX5chZV6j0Yji4mLk5uYiNzcX2dnZyM/Ph8lkgrOzM8aOHYt77rkHY8eORXh4OEJDQ+Hg4MC2bE5TW1uL3NxcnD17FmfOnEFubi6Ki4thMpng6emJ8PBwZouOjoazszPbkhlYNa/JZEJ+fj4OHDiAAwcO4MSJE+jo6IBSqcR9992HSZMmYeLEiYiMjISnpydbMu842tvbkZeXh5MnTyInJwcnT55ETU0NRCIRxo8fj+nTpyM+Ph5Tp05ltfGwuXkrKyuxe/duHDx4EIcOHUJLSwvUajXi4uIQGxuLSZMmYezYsdccmcXDDtXV1cjJycHRo0dx8OBBnDlzBsOGDUNUVBTi4uKQmJiIiIiIG84msSY2MW95eTl27NiBLVu24Pjx45DL5Zg8eTLi4+MRHx+PsLAwm75pntunsbERhw8fxoEDB7Bv3z789ttvcHNzw8yZMzF//nzMnDlz0J9XD5p5L168iC+++AJbtmzBuXPn4O7ujpSUFMyePRvTp0/nb6bsjIKCAqSnpyM9PR2nT5+Gq6srUlJS8PjjjyM6OnpwGieyInq9njZt2kTx8fEkFArJ09OTnn/+eTpy5AgZDAZrVsUzhCkrK6O3336bwsLCCAAFBwfT22+/TfX19Vatxyrmra+vp1WrVpGrqyuJRCJKTk6mH3/8kXp7e61RPA+HycvLo+XLl5OzszNJJBKaN28enTp1yipl35Z5KyoqaPny5SSXy0mtVtMbb7xBNTU1VhHGY190dXXRN998Q/fddx8BoBkzZtBPP/10W2UOyLxNTU20dOlSkkgk5O/vTx9++CF1dnbelhCeO4d9+/ZRTEwMAaCpU6dSbm7ugMq5JfMajUb69NNPydXVlby8vOjzzz+nnp6eAVXMw3Ps2DGKjo4mkUhEy5cvp9bW1ls6/qbNe/78eZo4cSKJxWJ6/vnnSavV3rJYHp6rMZlM9MUXX5BarSZ3d3dKT0+/6WNvyry7d+8mZ2dnCgsLo8LCwgELtRc2bdpEAAgASaVSm9S5bt06pk5vb2+b1GlLWltbacmSJSQQCOiVV165qadTNzTvm2++SUKhkBYvXmzX/VqdTkeBgYH04IMP3vQxcXFxNjNvHxMmTLAw70C0D1U+//xzkslklJiYSG1tbdfNe91pQH/961+xevVqbNiwAV988QUnh83dLEQEk8l0zRVvhjJc1n41TzzxBI4ePYrTp09j1qxZ0Ol01858vStAIBDQF198YeVry34YKi2vPVJcXExqtZpSU1PJZDL1m6fflvf8+fNYvnw5Xn75ZX5JTh5WCA4Oxg8//IBdu3bhX//6V/+Z+nN0amoqhYSE2PwxWHp6OnNTAoCKi4tp/vz55OLiwqRpNBoiImpsbKTnnnuO/P39SSKR0IgRI2j27NmUn59vVqZer6dXX32VgoODSS6Xk7OzMyUlJdGPP/7I3BRcXW9XV5dZGefOnaOUlBQaPnw4KRQKio6OpqNHj1q0vGvWrGHKmDJlCpO+Z88eJt3V1dWs7N7eXvruu+8oPj6e1Go1yWQyGjduHK1fv56MRqPFObq65b2edicnJ7N9V24CgYCqqqqYvDd7Pm3NqlWrSKVS9ft0y8K85eXlJBQKadu2bTYR1x8pKSkEgGJiYigrK4s6OjooJyeHRCIRaTQaqq2tJX9/f1Kr1bRr1y7S6XT066+/UkxMDMlkMjp+/DhT1pIlS8jJyYn2799PnZ2dVF9fTytXriQAlJWV1W+9VxqgrKyMVCoVeXt70/79+0mn01FRURHNmDGDRo4c2W+3wcHBwcy8fYSHh1uYd8eOHQSA1q5dSy0tLaTRaGjDhg0kFApp5cqVFmVcq9vQn3YnJyfS6XRm+V5//XWmvj5u5Xzamra2NlIqlfTBBx9Y7LMw70cffUROTk6sjkvo+yB2797d7/7FixcTAPr222/N0uvq6kgqlVJ4eDiTNmrUKJo8ebJFGUFBQTdl3vnz5xMA+uGHH8zy1tTUkFQqtYp5Y2NjLfIuWrSIJBKJRYtzO+bdvHkzCQQCevzxx82OvZXzyQYLFy6k+Ph4i3QL8y5btqzfk2lL+j6Ipqamfvc7OTmRUCjs96ukbyRT31fismXLCAA9/fTTdOLEies+P+zPAEqlkgBYtGBERCEhIbdt3mvR91z36lbvVsx7JTk5OSSTySgmJoa6u7vN9t3K+WSD9evXk6enp0W6xQ1be3s7lErlALvZ1qW/KSbd3d3QarUwmUxwcnKCQCAw2/Ly8gAAZWVlAICPP/4YX331FcrLyxEXF4fhw4dj5syZSE9Pv2H93d3d0Ol0kMlk/a41Zo2p5FqtFqtXr0ZISAicnZ2Z9/Hiiy8CADo7O2+7jsrKSqSkpMDX1xfbtm3DsGHDmH23ej7ZYPjw4f0+MrMwr7u7O2pra20iaiBIpVKoVCqIxWL09vaCLn97WGzTpk0DcHlF8rS0NBw4cABtbW3IyMgAEWHOnDl47733bliXUqmEXq9He3u7xf6WlpZ+jxMKhejp6bFIb2trs0hLTk7GmjVr8PTTT6O0tBQmkwlEhPfffx8AzNZ6GAg6nQ5JSUno7e3Fzp074eLiYrb/Vs8nG9TU1ECtVlukW5g3KioKhYWFaG5utomwgTBnzhwYDAZkZ2db7Hv77bfh5+fHLMGkUqlQXFwM4PJSowkJCcjIyIBAIMCuXbtuWNesWbMAAHv37jVLb2pqQklJSb/HeHp6oqamxiytvr4elZWVZmlGoxHZ2dnw8PDAihUr4Obmxsw4sMYKPUajEQsWLEBxcTG2bt2KoKAgZt+8efOQkZEB4NbOJxscPHgQUVFRljuu7ke0t7eTs7Mzvfnmm4PTgbkJbtR/a2hooNGjR1NAQADt3r2b2traqLm5mT755BNSKBS0efNmJq+TkxPFxMRQYWEh6fV6amhooNdee40A0BtvvHHDes+fP08uLi5mTxvOnDlDiYmJ5O7u3m+f99lnnyUA9OGHH5JOp6Pz58/Tww8/TN7e3hZ93unTpxMAeuedd0ij0VBnZycdOnSI/Pz8CABlZmaa5b+VPu9zzz1HAOh///d/LfLPnTuXGQRzK+fT1hQVFZFAIOj35r3f57yvvfYaKZVKm3fST5w40e8zyf5obm6mF154gQICAkgikZCbmxvNmDHD4sMuKCigpUuX0t13300KhYJcXFxo0qRJ9NlnnzG/3Fz9rBQALVy4kCmjpKSEUlNTafjw4SSXyykiIoJ27txJcXFxTP6nnnqKyd/W1kZLliwhT09PksvlFB0dTadOnaLw8HAm/6pVq4iISKPR0NKlS8nX15ckEgmp1Wp6/PHH6eWXX2byhoeHmw3M6dv+8pe/XFP7L7/8cs1nvH3blSO4bvZ82hKTyUTTpk2j8PDwfn9l63cCpl6vR1hYGFQqFbKysvjJkjys8Prrr+ONN97A8ePHcd9991lmuJbrz507R05OTpSSksIPOOexOf/6179IIBDQP//5z2vmue6QyOPHj5NSqaSYmBhqaGiwukAenqsxGo30yiuvkEAgoDVr1lw37w3H8547d47uuusu8vb2phMnTlhNJA/P1Wi1WkpJSSGpVEqfffbZDfPf1EyKlpYWmjlzJkmlUlq9erVdD0rnYYdt27aRn58f+fr60s8//3xTx9z0HDaDwUD/+Mc/SKlUUkBAAG3fvn3AQnl4+igtLaVZs2aRQCCgtLQ0amxsvOljb3nqe01NDS1YsIAAUEJCAh0+fPhWi+Dhod9++42effZZkkqlFBISQkeOHLnlMga86EhWVhbdf//9zNjVXbt2XXPEOw9PH8XFxfTEE0+QRCIhPz8/+vjjjwc8gvG2l3s6evQo0+z3DaJubm6+3WJ57Iienh5KT0+npKQkEgqFFBwcbJU1P6y20F5eXh499dRTpFQqSSqV0oIFCygzM7Pf2QA8dwbFxcX00ksvkYeHBwmFQkpISKAffvjBap6w6iqRRJfXpPr+++8pPj6eBAIBjRgxgtLS0mj79u0W40h57I8LFy7Q+vXracqUKSQQCMjLy4tWrVpFFy5csHpdg7q49Llz57Blyxakp6ejoKAALi4uSE5ORkpKCqZNmwaVSjVYVfPYCIPBgJMnT2Lnzp1IT09HSUkJPDw8kJqairlz52L69OkQCgcn0KrNlvUvLy/Htm3bkJ6ejpycHAgEAtx3332Ii4tDXFwcJk+ebBHPjGfoQUT49ddfcfDgQRw8eBCHDx+GTqfDqFGjMHv2bMyZMwdRUVGDZtgrYSWgSnNzM7KysnDw4EEcOHAA58+fh1wux6RJkxAVFYWJEydi4sSJ/Q5A5rEtnZ2dyM3NRU5ODnJycpCdnY2Ghga4uLhg2rRpTONz5VhhWzEkQln99ttvOHjwII4ePYqTJ0+ipKQEJpMJI0eOZCIC9cUHc3NzY1uu3dLZ2YmzZ8+iqKgIv/zyC3JycnD69GkYDAZ4eHhg4sSJmDx5MqZPn46wsDCbtK7XY0iY92q0Wi1OnjzJbKdOnUJjYyMAQK1WIyQkBCEhIRg3bhzGjRuHwMBAi+ktPNdGr9fjwoULOHfuHE6fPo0zZ86gsLAQ5eXlMJlMUCgUuPfee5lvwEmTJsHf359t2RYMSfP2R0NDA06fPs1EwiwqKsLZs2eZ6TIuLi5M9MsrI2L6+flBrVbfcZHUGxsbUVtbi/Lycpw/f95sq66uBhFBJBIhICCA+VYbN24cxo8fj9GjR7Peqt4MnDFvfxiNRlRUVDAfSl9o0vPnz6O8vJyJuSsUCqFWq5kYxD4+PvD09ISXl5dZJPW+bajGgNNqtdBoNEwg7b4g2n0xh2tqapg4xFe+dx8fH4vwtoGBgRgzZgynF0/ktHmvh8lkYiK99wWT7ov+XlVVhdraWtTX1/c7K1ilUmHEiBFwcnKCUqmETCaDUqmEo6MjZDIZhg8fDgcHB2YKuUqlMgvVJJVKzQJy9/b2WtSj0+mYSY2tra3o6uqCXq9Ha2sr9Ho9urq60NbWhs7OTsaoV0+ClEqlGDFiBBMQ/Mpo731pI0eOtNuZMHZr3pulu7sbLS0tZq1Zc3MzNBoNLl26hPb2dnR1dUGn00Gn00Gv1zOvDQYDTCYTtFqtWZmdnZ1Mywdcnn5/9TNthULBmEqlUkEul0Mul1u8VigUZt8KI0aMYF73t5bEncQdb14e7jL0e+U8PNeANy8PZ+HNy8NZxAC2sC2Ch2cg/D9HlYIlBqFupgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dag.append_to_tail(node)\n",
    "dtfcore.draw(dag)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "0ac40a70",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.719647Z",
     "start_time": "2023-06-04T18:41:15.527024Z"
    }
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "54bed6de9d954f97b575f6cc9ce2e1e4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "run_leq_node:   0%|          | 0/3 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/app/helpers/hdbg.py:311: PerformanceWarning: indexing past lexsort depth may impact performance.\n",
      "  cond = value in valid_values\n"
     ]
    }
   ],
   "source": [
    "residualize_df = dag.run_leq_node(\"residualize\", \"fit\")[\"df_out\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "1d8b3812",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.755146Z",
     "start_time": "2023-06-04T18:41:15.722963Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"4\" halign=\"left\">residual.shift_-1</th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0.shift_-1</th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0.shift_-1_hat</th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0</th>\n",
       "      <th colspan=\"4\" halign=\"left\">x_1</th>\n",
       "      <th colspan=\"4\" halign=\"left\">x_2</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2023-05-01</th>\n",
       "      <td>0.406651</td>\n",
       "      <td>0.666367</td>\n",
       "      <td>-0.549490</td>\n",
       "      <td>-1.721813</td>\n",
       "      <td>0.446375</td>\n",
       "      <td>0.663063</td>\n",
       "      <td>-0.466750</td>\n",
       "      <td>-2.250854</td>\n",
       "      <td>0.039724</td>\n",
       "      <td>-0.003303</td>\n",
       "      <td>0.082740</td>\n",
       "      <td>-0.529041</td>\n",
       "      <td>0.330437</td>\n",
       "      <td>-0.377605</td>\n",
       "      <td>-0.982188</td>\n",
       "      <td>-0.094828</td>\n",
       "      <td>0.345584</td>\n",
       "      <td>2.117839</td>\n",
       "      <td>0.054102</td>\n",
       "      <td>0.876242</td>\n",
       "      <td>0.821618</td>\n",
       "      <td>-1.112021</td>\n",
       "      <td>0.272791</td>\n",
       "      <td>0.256486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-02</th>\n",
       "      <td>-0.043662</td>\n",
       "      <td>-0.162506</td>\n",
       "      <td>-0.406108</td>\n",
       "      <td>-0.226181</td>\n",
       "      <td>0.364572</td>\n",
       "      <td>0.167465</td>\n",
       "      <td>-1.648787</td>\n",
       "      <td>-1.425349</td>\n",
       "      <td>0.408234</td>\n",
       "      <td>0.329971</td>\n",
       "      <td>-1.242679</td>\n",
       "      <td>-1.199168</td>\n",
       "      <td>0.446375</td>\n",
       "      <td>0.663063</td>\n",
       "      <td>-0.466750</td>\n",
       "      <td>-2.250854</td>\n",
       "      <td>-1.303157</td>\n",
       "      <td>2.042772</td>\n",
       "      <td>-1.107373</td>\n",
       "      <td>-0.258848</td>\n",
       "      <td>0.905356</td>\n",
       "      <td>0.646703</td>\n",
       "      <td>0.199585</td>\n",
       "      <td>1.055743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-03</th>\n",
       "      <td>0.349502</td>\n",
       "      <td>-0.312376</td>\n",
       "      <td>-0.624330</td>\n",
       "      <td>0.866880</td>\n",
       "      <td>0.546713</td>\n",
       "      <td>-0.683227</td>\n",
       "      <td>-0.297527</td>\n",
       "      <td>0.862445</td>\n",
       "      <td>0.197211</td>\n",
       "      <td>-0.370850</td>\n",
       "      <td>0.326803</td>\n",
       "      <td>-0.004435</td>\n",
       "      <td>0.364572</td>\n",
       "      <td>0.167465</td>\n",
       "      <td>-1.648787</td>\n",
       "      <td>-1.425349</td>\n",
       "      <td>-0.536953</td>\n",
       "      <td>-0.514006</td>\n",
       "      <td>0.235506</td>\n",
       "      <td>-0.138655</td>\n",
       "      <td>0.581118</td>\n",
       "      <td>-1.648075</td>\n",
       "      <td>0.759520</td>\n",
       "      <td>0.033000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-04</th>\n",
       "      <td>-0.422422</td>\n",
       "      <td>0.128637</td>\n",
       "      <td>-0.131074</td>\n",
       "      <td>0.522907</td>\n",
       "      <td>-0.482119</td>\n",
       "      <td>-0.098270</td>\n",
       "      <td>0.253447</td>\n",
       "      <td>1.218844</td>\n",
       "      <td>-0.059698</td>\n",
       "      <td>-0.226907</td>\n",
       "      <td>0.384520</td>\n",
       "      <td>0.695936</td>\n",
       "      <td>0.546713</td>\n",
       "      <td>-0.683227</td>\n",
       "      <td>-0.297527</td>\n",
       "      <td>0.862445</td>\n",
       "      <td>0.294132</td>\n",
       "      <td>0.109014</td>\n",
       "      <td>0.254388</td>\n",
       "      <td>0.332814</td>\n",
       "      <td>0.028422</td>\n",
       "      <td>-1.227352</td>\n",
       "      <td>1.224647</td>\n",
       "      <td>-0.651281</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-05</th>\n",
       "      <td>-0.429141</td>\n",
       "      <td>-0.316006</td>\n",
       "      <td>-0.619186</td>\n",
       "      <td>-0.744521</td>\n",
       "      <td>-0.292457</td>\n",
       "      <td>-0.506292</td>\n",
       "      <td>-1.481818</td>\n",
       "      <td>-1.514319</td>\n",
       "      <td>0.136685</td>\n",
       "      <td>-0.190286</td>\n",
       "      <td>-0.862632</td>\n",
       "      <td>-0.769797</td>\n",
       "      <td>-0.482119</td>\n",
       "      <td>-0.098270</td>\n",
       "      <td>0.253447</td>\n",
       "      <td>1.218844</td>\n",
       "      <td>-0.736454</td>\n",
       "      <td>-0.072044</td>\n",
       "      <td>-0.810815</td>\n",
       "      <td>-0.125592</td>\n",
       "      <td>-0.162910</td>\n",
       "      <td>-0.944752</td>\n",
       "      <td>0.752244</td>\n",
       "      <td>0.669153</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           residual.shift_-1                               ret_0.shift_-1                               ret_0.shift_-1_hat                                   ret_0                                     x_1                                     x_2                              \n",
       "                         100       101       102       103            100       101       102       103                100       101       102       103       100       101       102       103       100       101       102       103       100       101       102       103\n",
       "2023-05-01          0.406651  0.666367 -0.549490 -1.721813       0.446375  0.663063 -0.466750 -2.250854           0.039724 -0.003303  0.082740 -0.529041  0.330437 -0.377605 -0.982188 -0.094828  0.345584  2.117839  0.054102  0.876242  0.821618 -1.112021  0.272791  0.256486\n",
       "2023-05-02         -0.043662 -0.162506 -0.406108 -0.226181       0.364572  0.167465 -1.648787 -1.425349           0.408234  0.329971 -1.242679 -1.199168  0.446375  0.663063 -0.466750 -2.250854 -1.303157  2.042772 -1.107373 -0.258848  0.905356  0.646703  0.199585  1.055743\n",
       "2023-05-03          0.349502 -0.312376 -0.624330  0.866880       0.546713 -0.683227 -0.297527  0.862445           0.197211 -0.370850  0.326803 -0.004435  0.364572  0.167465 -1.648787 -1.425349 -0.536953 -0.514006  0.235506 -0.138655  0.581118 -1.648075  0.759520  0.033000\n",
       "2023-05-04         -0.422422  0.128637 -0.131074  0.522907      -0.482119 -0.098270  0.253447  1.218844          -0.059698 -0.226907  0.384520  0.695936  0.546713 -0.683227 -0.297527  0.862445  0.294132  0.109014  0.254388  0.332814  0.028422 -1.227352  1.224647 -0.651281\n",
       "2023-05-05         -0.429141 -0.316006 -0.619186 -0.744521      -0.292457 -0.506292 -1.481818 -1.514319           0.136685 -0.190286 -0.862632 -0.769797 -0.482119 -0.098270  0.253447  1.218844 -0.736454 -0.072044 -0.810815 -0.125592 -0.162910 -0.944752  0.752244  0.669153"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "residualize_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "9e8a7e2a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.762829Z",
     "start_time": "2023-06-04T18:41:15.758505Z"
    }
   },
   "outputs": [],
   "source": [
    "# Next we use a simple EWMA to explicitly estimate the variances of the\n",
    "#  residuals (though the sklearn model is internally computing this in\n",
    "#  the regression).\n",
    "node_config = {\n",
    "    \"in_col_group\": (\"residual.shift_-1\",), \n",
    "    \"out_col_group\": (\"smoothed_squared_residual.shift_-1\",),\n",
    "    \"transformer_func\": csigproc.compute_rolling_norm,\n",
    "    \"transformer_kwargs\": {\n",
    "        \"tau\": 10,\n",
    "    }\n",
    "}\n",
    "node = dtfcore.SeriesToSeriesTransformer(\n",
    "    \"compute_rolling_norm\",\n",
    "    **node_config,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "ba0cc13e",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:15.796822Z",
     "start_time": "2023-06-04T18:41:15.765539Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dag.append_to_tail(node)\n",
    "dtfcore.draw(dag)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "fd53bd40",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:16.025010Z",
     "start_time": "2023-06-04T18:41:15.798821Z"
    }
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "495a9ee3c8d74aeab1a2bd06cbc5f057",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "run_leq_node:   0%|          | 0/4 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/app/helpers/hdbg.py:311: PerformanceWarning: indexing past lexsort depth may impact performance.\n",
      "  cond = value in valid_values\n"
     ]
    }
   ],
   "source": [
    "compute_rolling_norm_df = dag.run_leq_node(\n",
    "    \"compute_rolling_norm\", \"fit\"\n",
    ")[\"df_out\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "095bf768",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:16.065071Z",
     "start_time": "2023-06-04T18:41:16.027440Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"4\" halign=\"left\">smoothed_squared_residual.shift_-1</th>\n",
       "      <th colspan=\"4\" halign=\"left\">residual.shift_-1</th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0.shift_-1</th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0.shift_-1_hat</th>\n",
       "      <th colspan=\"4\" halign=\"left\">ret_0</th>\n",
       "      <th colspan=\"4\" halign=\"left\">x_1</th>\n",
       "      <th colspan=\"4\" halign=\"left\">x_2</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "      <th>100</th>\n",
       "      <th>101</th>\n",
       "      <th>102</th>\n",
       "      <th>103</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2023-05-01</th>\n",
       "      <td>0.406651</td>\n",
       "      <td>0.666367</td>\n",
       "      <td>0.549490</td>\n",
       "      <td>1.721813</td>\n",
       "      <td>0.406651</td>\n",
       "      <td>0.666367</td>\n",
       "      <td>-0.549490</td>\n",
       "      <td>-1.721813</td>\n",
       "      <td>0.446375</td>\n",
       "      <td>0.663063</td>\n",
       "      <td>-0.466750</td>\n",
       "      <td>-2.250854</td>\n",
       "      <td>0.039724</td>\n",
       "      <td>-0.003303</td>\n",
       "      <td>0.082740</td>\n",
       "      <td>-0.529041</td>\n",
       "      <td>0.330437</td>\n",
       "      <td>-0.377605</td>\n",
       "      <td>-0.982188</td>\n",
       "      <td>-0.094828</td>\n",
       "      <td>0.345584</td>\n",
       "      <td>2.117839</td>\n",
       "      <td>0.054102</td>\n",
       "      <td>0.876242</td>\n",
       "      <td>0.821618</td>\n",
       "      <td>-1.112021</td>\n",
       "      <td>0.272791</td>\n",
       "      <td>0.256486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-02</th>\n",
       "      <td>0.282050</td>\n",
       "      <td>0.474125</td>\n",
       "      <td>0.479592</td>\n",
       "      <td>1.197966</td>\n",
       "      <td>-0.043662</td>\n",
       "      <td>-0.162506</td>\n",
       "      <td>-0.406108</td>\n",
       "      <td>-0.226181</td>\n",
       "      <td>0.364572</td>\n",
       "      <td>0.167465</td>\n",
       "      <td>-1.648787</td>\n",
       "      <td>-1.425349</td>\n",
       "      <td>0.408234</td>\n",
       "      <td>0.329971</td>\n",
       "      <td>-1.242679</td>\n",
       "      <td>-1.199168</td>\n",
       "      <td>0.446375</td>\n",
       "      <td>0.663063</td>\n",
       "      <td>-0.466750</td>\n",
       "      <td>-2.250854</td>\n",
       "      <td>-1.303157</td>\n",
       "      <td>2.042772</td>\n",
       "      <td>-1.107373</td>\n",
       "      <td>-0.258848</td>\n",
       "      <td>0.905356</td>\n",
       "      <td>0.646703</td>\n",
       "      <td>0.199585</td>\n",
       "      <td>1.055743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-03</th>\n",
       "      <td>0.308534</td>\n",
       "      <td>0.422001</td>\n",
       "      <td>0.537284</td>\n",
       "      <td>1.088170</td>\n",
       "      <td>0.349502</td>\n",
       "      <td>-0.312376</td>\n",
       "      <td>-0.624330</td>\n",
       "      <td>0.866880</td>\n",
       "      <td>0.546713</td>\n",
       "      <td>-0.683227</td>\n",
       "      <td>-0.297527</td>\n",
       "      <td>0.862445</td>\n",
       "      <td>0.197211</td>\n",
       "      <td>-0.370850</td>\n",
       "      <td>0.326803</td>\n",
       "      <td>-0.004435</td>\n",
       "      <td>0.364572</td>\n",
       "      <td>0.167465</td>\n",
       "      <td>-1.648787</td>\n",
       "      <td>-1.425349</td>\n",
       "      <td>-0.536953</td>\n",
       "      <td>-0.514006</td>\n",
       "      <td>0.235506</td>\n",
       "      <td>-0.138655</td>\n",
       "      <td>0.581118</td>\n",
       "      <td>-1.648075</td>\n",
       "      <td>0.759520</td>\n",
       "      <td>0.033000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-04</th>\n",
       "      <td>0.345286</td>\n",
       "      <td>0.362570</td>\n",
       "      <td>0.458593</td>\n",
       "      <td>0.959814</td>\n",
       "      <td>-0.422422</td>\n",
       "      <td>0.128637</td>\n",
       "      <td>-0.131074</td>\n",
       "      <td>0.522907</td>\n",
       "      <td>-0.482119</td>\n",
       "      <td>-0.098270</td>\n",
       "      <td>0.253447</td>\n",
       "      <td>1.218844</td>\n",
       "      <td>-0.059698</td>\n",
       "      <td>-0.226907</td>\n",
       "      <td>0.384520</td>\n",
       "      <td>0.695936</td>\n",
       "      <td>0.546713</td>\n",
       "      <td>-0.683227</td>\n",
       "      <td>-0.297527</td>\n",
       "      <td>0.862445</td>\n",
       "      <td>0.294132</td>\n",
       "      <td>0.109014</td>\n",
       "      <td>0.254388</td>\n",
       "      <td>0.332814</td>\n",
       "      <td>0.028422</td>\n",
       "      <td>-1.227352</td>\n",
       "      <td>1.224647</td>\n",
       "      <td>-0.651281</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023-05-05</th>\n",
       "      <td>0.367326</td>\n",
       "      <td>0.351874</td>\n",
       "      <td>0.502164</td>\n",
       "      <td>0.912414</td>\n",
       "      <td>-0.429141</td>\n",
       "      <td>-0.316006</td>\n",
       "      <td>-0.619186</td>\n",
       "      <td>-0.744521</td>\n",
       "      <td>-0.292457</td>\n",
       "      <td>-0.506292</td>\n",
       "      <td>-1.481818</td>\n",
       "      <td>-1.514319</td>\n",
       "      <td>0.136685</td>\n",
       "      <td>-0.190286</td>\n",
       "      <td>-0.862632</td>\n",
       "      <td>-0.769797</td>\n",
       "      <td>-0.482119</td>\n",
       "      <td>-0.098270</td>\n",
       "      <td>0.253447</td>\n",
       "      <td>1.218844</td>\n",
       "      <td>-0.736454</td>\n",
       "      <td>-0.072044</td>\n",
       "      <td>-0.810815</td>\n",
       "      <td>-0.125592</td>\n",
       "      <td>-0.162910</td>\n",
       "      <td>-0.944752</td>\n",
       "      <td>0.752244</td>\n",
       "      <td>0.669153</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           smoothed_squared_residual.shift_-1                               residual.shift_-1                               ret_0.shift_-1                               ret_0.shift_-1_hat                                   ret_0                                     x_1                                     x_2                              \n",
       "                                          100       101       102       103               100       101       102       103            100       101       102       103                100       101       102       103       100       101       102       103       100       101       102       103       100       101       102       103\n",
       "2023-05-01                           0.406651  0.666367  0.549490  1.721813          0.406651  0.666367 -0.549490 -1.721813       0.446375  0.663063 -0.466750 -2.250854           0.039724 -0.003303  0.082740 -0.529041  0.330437 -0.377605 -0.982188 -0.094828  0.345584  2.117839  0.054102  0.876242  0.821618 -1.112021  0.272791  0.256486\n",
       "2023-05-02                           0.282050  0.474125  0.479592  1.197966         -0.043662 -0.162506 -0.406108 -0.226181       0.364572  0.167465 -1.648787 -1.425349           0.408234  0.329971 -1.242679 -1.199168  0.446375  0.663063 -0.466750 -2.250854 -1.303157  2.042772 -1.107373 -0.258848  0.905356  0.646703  0.199585  1.055743\n",
       "2023-05-03                           0.308534  0.422001  0.537284  1.088170          0.349502 -0.312376 -0.624330  0.866880       0.546713 -0.683227 -0.297527  0.862445           0.197211 -0.370850  0.326803 -0.004435  0.364572  0.167465 -1.648787 -1.425349 -0.536953 -0.514006  0.235506 -0.138655  0.581118 -1.648075  0.759520  0.033000\n",
       "2023-05-04                           0.345286  0.362570  0.458593  0.959814         -0.422422  0.128637 -0.131074  0.522907      -0.482119 -0.098270  0.253447  1.218844          -0.059698 -0.226907  0.384520  0.695936  0.546713 -0.683227 -0.297527  0.862445  0.294132  0.109014  0.254388  0.332814  0.028422 -1.227352  1.224647 -0.651281\n",
       "2023-05-05                           0.367326  0.351874  0.502164  0.912414         -0.429141 -0.316006 -0.619186 -0.744521      -0.292457 -0.506292 -1.481818 -1.514319           0.136685 -0.190286 -0.862632 -0.769797 -0.482119 -0.098270  0.253447  1.218844 -0.736454 -0.072044 -0.810815 -0.125592 -0.162910 -0.944752  0.752244  0.669153"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "compute_rolling_norm_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "68433eb0",
   "metadata": {},
   "source": [
    "## Estimate factor returns variance-covariance matrices"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "6b6ffbe7",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:16.070139Z",
     "start_time": "2023-06-04T18:41:16.067708Z"
    }
   },
   "outputs": [],
   "source": [
    "# TODO: We can compute beta beta^transpose, linearizing the matrix as\n",
    "#  feature columns, and then apply EWMA smoothing as above."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b862f74e",
   "metadata": {},
   "source": [
    "# Extract estimates and compute statistics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "3407ca77",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:16.080456Z",
     "start_time": "2023-06-04T18:41:16.075639Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['_key_fit_state', \"_info['fit']\"])"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Information supplementary to the dataframes can be accessed\n",
    "# (such as any sklearn model info).\n",
    "sklearn_fit_state = dag.get_node(\"sklearn\").get_fit_state()\n",
    "display(sklearn_fit_state.keys())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "7fbdfaee",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:16.090695Z",
     "start_time": "2023-06-04T18:41:16.082288Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'_model': LinearRegression(fit_intercept=False),\n",
       " \"_info['fit']\": OrderedDict([('model_x_vars', ['x_1', 'x_2']),\n",
       "              ('model_params',\n",
       "               {'copy_X': True,\n",
       "                'fit_intercept': False,\n",
       "                'n_jobs': None,\n",
       "                'normalize': 'deprecated',\n",
       "                'positive': False}),\n",
       "              ('model_attributes',\n",
       "               OrderedDict([('fit_intercept', False),\n",
       "                            ('normalize', 'deprecated'),\n",
       "                            ('copy_X', True),\n",
       "                            ('n_jobs', None),\n",
       "                            ('positive', False),\n",
       "                            ('n_features_in_', 2),\n",
       "                            ('coef_', array([[-0.21643078,  0.13938234]])),\n",
       "                            ('rank_', 2),\n",
       "                            ('singular_', array([3.67883457, 2.18353991])),\n",
       "                            ('intercept_', 0.0)])),\n",
       "              ('model_score', 0.07223187389865093),\n",
       "              ('df_out_info',\n",
       "               \"<class 'pandas.core.frame.DataFrame'>\\nDatetimeIndex: 9 entries, 2023-05-01 to 2023-05-11\\nFreq: B\\nData columns (total 2 columns):\\n #   Column              Non-Null Count  Dtype  \\n---  ------              --------------  -----  \\n 0   ret_0.shift_-1      9 non-null      float64\\n 1   ret_0.shift_-1_hat  9 non-null      float64\\ndtypes: float64(2)\")])}"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Note that we may recover from sklearn information such as\n",
    "# - model coefficients\n",
    "# - model score\n",
    "# - full model specification\n",
    "sklearn_fit_state[\"_key_fit_state\"][100]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "88af80a3",
   "metadata": {},
   "source": [
    "# Alternative DAG construction method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "fcf0ea02",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-06-04T18:41:16.108727Z",
     "start_time": "2023-06-04T18:41:16.093447Z"
    }
   },
   "outputs": [],
   "source": [
    "# In transferring this DAG from a notebook to code, we would do something\n",
    "# similar to the following:\n",
    "\n",
    "class SimpleRiskModel_DagBuilder(dtfcore.DagBuilder):\n",
    "    \"\"\"\n",
    "    A pipeline similar to real feature processing.\n",
    "    \"\"\"\n",
    "\n",
    "    @staticmethod\n",
    "    def get_column_name(tag: str) -> str:\n",
    "        \"\"\"\n",
    "        See description in the parent class.\n",
    "        \"\"\"\n",
    "        raise NotImplementedError\n",
    "\n",
    "    def get_trading_period(\n",
    "        self, config: cconfig.Config, mark_key_as_used: bool\n",
    "    ) -> str:\n",
    "        \"\"\"\n",
    "        See description in the parent class.\n",
    "        \"\"\"\n",
    "        _ = self\n",
    "        raise NotImplementedError\n",
    "\n",
    "    def get_required_lookback_in_effective_days(\n",
    "        self, config: cconfig.Config, mark_key_as_used: bool\n",
    "    ) -> str:\n",
    "        \"\"\"\n",
    "        See description in the parent class.\n",
    "        \"\"\"\n",
    "        raise NotImplementedError\n",
    "\n",
    "    def set_weights(\n",
    "        self, config: cconfig.Config, weights: pd.Series\n",
    "    ) -> cconfig.Config:\n",
    "        \"\"\"\n",
    "        See description in the parent class.\n",
    "        \"\"\"\n",
    "        raise NotImplementedError\n",
    "\n",
    "    def convert_to_fast_prod_setup(\n",
    "        self, config: cconfig.Config\n",
    "    ) -> cconfig.Config:\n",
    "        \"\"\"\n",
    "        See description in the parent class.\n",
    "        \"\"\"\n",
    "        raise NotImplementedError\n",
    "\n",
    "    def get_config_template(self) -> cconfig.Config:\n",
    "        dict_ = {\n",
    "            self._get_nid(\"sklearn\"): {\n",
    "                \"in_col_groups\": [(x,) for x in cols],\n",
    "                \"out_col_group\": (),\n",
    "                \"x_vars\": x_vars,\n",
    "                \"y_vars\": y_vars,\n",
    "                \"steps_ahead\": 1,\n",
    "                \"model_kwargs\": {\n",
    "                    \"fit_intercept\": False,\n",
    "                },\n",
    "                \"nan_mode\": \"drop\",\n",
    "            },\n",
    "            self._get_nid(\"residualize\"): {\n",
    "                \"in_col_groups\": [\n",
    "                    (\"ret_0.shift_-1\",), \n",
    "                    (\"ret_0.shift_-1_hat\",),\n",
    "                ],\n",
    "                \"out_col_group\": (),\n",
    "                \"transformer_kwargs\": {\n",
    "                    \"term1_col\": \"ret_0.shift_-1\",\n",
    "                    \"term2_col\": \"ret_0.shift_-1_hat\",\n",
    "                    \"out_col\": \"residual.shift_-1\",\n",
    "                    \"operation\": \"sub\"\n",
    "                }\n",
    "            },\n",
    "            self._get_nid(\"compute_rolling_node\"): {\n",
    "                \"in_col_group\": (\"residual.shift_-1\",), \n",
    "                \"out_col_group\": (\"smoothed_squared_residual.shift_-1\",),\n",
    "                \"transformer_kwargs\": {\n",
    "                    \"tau\": 10,\n",
    "                }\n",
    "            },\n",
    "        }\n",
    "        config = cconfig.Config.from_dict(dict_)\n",
    "        return config\n",
    "\n",
    "    def _get_dag(\n",
    "        self,\n",
    "        config: cconfig.Config,\n",
    "        mode: str = \"strict\",\n",
    "    ) -> dtfcore.DAG:\n",
    "        dag = dtfcore.DAG(mode=mode)\n",
    "        _LOG.debug(\"%s\", config)\n",
    "        #\n",
    "        stage = \"sklearn\"\n",
    "        nid = self._get_nid(stage)\n",
    "        node = dtfcore.MultiindexSkLearnModel(\n",
    "            nid,\n",
    "            model_func=slmode.LinearRegression,\n",
    "            **config[nid].to_dict(),\n",
    "        )\n",
    "        dag.append_to_tail(node)\n",
    "        #\n",
    "        node = dtfcore.GroupedColDfToDfTransformer(\n",
    "            \"residualize\",\n",
    "            transformer_func=cofeatur.combine_columns,\n",
    "            **config[nid].to_dict(),\n",
    "        )\n",
    "        dag.append_to_tail(node)\n",
    "        #\n",
    "        node = dtfcore.SeriesToSeriesTransformer(\n",
    "            nid,\n",
    "            transformer_func=csigproc.compute_rolling_norm,\n",
    "            **config[nid].to_dict(),\n",
    "        )\n",
    "        dag.append_to_tail(node)\n",
    "        #\n",
    "        return dag"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c1593155",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.5"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}