standard_betting_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "ead66db1",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.layers import Dense, Dropout, Input, Concatenate, Lambda, Conv1D, Activation\n",
    "from tensorflow.keras import Model\n",
    "from tensorflow.keras.models import load_model\n",
    "from tensorflow.keras.callbacks import ModelCheckpoint, LearningRateScheduler\n",
    "\n",
    "from model_util.losses import get_fractional_loss, get_unit_bet_loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "39d54ffc",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load validation data\n",
    "save_path = \"data/validation_data/\"\n",
    "X_train = np.load(save_path + \"X_train.npy\")\n",
    "Y_train = np.load(save_path + \"Y_train.npy\")\n",
    "X_test = np.load(save_path + \"X_test.npy\")\n",
    "Y_test = np.load(save_path + \"Y_test.npy\")\n",
    "X_val = np.load(save_path + \"X_val.npy\")\n",
    "Y_val = np.load(save_path + \"Y_val.npy\")\n",
    "\n",
    "# Define odds inputs\n",
    "O_train, O_test, O_val = Y_train[:, 1:], Y_test[:, 1:], Y_val[:, 1:]\n",
    "scaler = StandardScaler()\n",
    "O_train = scaler.fit_transform(O_train)\n",
    "O_test = scaler.transform(O_test)\n",
    "O_val = scaler.transform(O_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "c66b1e21",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load in pretrained win-loss models\n",
    "wl_models = [load_model(f\"saved_models/win_loss_model/model_{i}\") for i in range(9)]\n",
    "\n",
    "# Build an averaging model\n",
    "wl_input = Input(shape=(X_train.shape[1]))\n",
    "wl_outputs = [m(wl_input) for m in wl_models]\n",
    "avg_wl_output = tf.keras.layers.Average()(wl_outputs)\n",
    "\n",
    "avg_wl_model = Model(inputs=wl_input, outputs=avg_wl_output, name=\"wl_avg_model\")\n",
    "avg_wl_model.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "1c780cae",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tf.keras.utils.plot_model(avg_wl_model, to_file='model_images/avg_wl_model.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "27f344c1",
   "metadata": {},
   "outputs": [],
   "source": [
    "epochs = 400\n",
    "learning_rate = 1e-4\n",
    "learning_rate_decay = 0 # learning_rate / 100\n",
    "batch_size = 16\n",
    "n_models = 13\n",
    "regularizer = tf.keras.regularizers.L1(l1=0.005)\n",
    "\n",
    "loss = get_unit_bet_loss(batch_size=batch_size, day_bet_num=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "2c191113",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define the kelly function\n",
    "def tf_kelly(layer):\n",
    "    # Split the layer\n",
    "    home_win, home_o, away_o = tf.split(layer, [1,1,1], 1)\n",
    "    away_win = 1 - home_win\n",
    "    # Do the calculations\n",
    "    home_frac = ((home_o-1)*home_win - away_win)/(home_o-1)\n",
    "    away_frac = ((away_o-1)*away_win - home_win)/(away_o-1)\n",
    "    # Return the layer\n",
    "    return Concatenate()([home_frac, away_frac])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "6c12de7d",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "ename": "TypeError",
     "evalue": "list indices must be integers or slices, not str",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mTypeError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-7-c1a9315e3f08>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m     12\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     13\u001b[0m \u001b[0;31m# Do a kelly transformation\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 14\u001b[0;31m \u001b[0mkelly_layer\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mConcatenate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"wl_output_odds\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mwl_output\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0modds_input\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     15\u001b[0m \u001b[0mkelly_layer\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mLambda\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtf_kelly\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'kelly_fn'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mkelly_layer\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     16\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/.local/lib/python3.8/site-packages/keras/utils/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     65\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# pylint: disable=broad-except\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     66\u001b[0m       \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 67\u001b[0;31m       \u001b[0;32mraise\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfiltered_tb\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     68\u001b[0m     \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     69\u001b[0m       \u001b[0;32mdel\u001b[0m \u001b[0mfiltered_tb\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/.local/lib/python3.8/site-packages/keras/layers/merging/concatenate.py\u001b[0m in \u001b[0;36mbuild\u001b[0;34m(self, input_shape)\u001b[0m\n\u001b[1;32m    102\u001b[0m     \u001b[0mshape_set\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mset\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    103\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0mi\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mreduced_inputs_shapes\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 104\u001b[0;31m       \u001b[0;32mdel\u001b[0m \u001b[0mreduced_inputs_shapes\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0maxis\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    105\u001b[0m       \u001b[0mshape_set\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0madd\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtuple\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mreduced_inputs_shapes\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    106\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mTypeError\u001b[0m: list indices must be integers or slices, not str"
     ]
    }
   ],
   "source": [
    "# Make the model\n",
    "main_input = Input(shape=(X_train.shape[1]), name=\"main_input\")\n",
    "odds_input = Input(shape=(2), name=\"odds_input\")\n",
    "\n",
    "odds_concat = Concatenate(name=\"main_odds_input\")([main_input, odds_input])\n",
    "\n",
    "# Define the main branch\n",
    "hidden_layer = Dense(10, activation=\"relu\", kernel_regularizer=regularizer, name=\"main_odds_hl_1\")(odds_concat)\n",
    "\n",
    "# Define the wl branch\n",
    "wl_output = avg_wl_model(main_input)\n",
    "\n",
    "# Do a kelly transformation\n",
    "kelly_layer = Concatenate(name=\"wl_output_odds\")([wl_output, odds_input])\n",
    "kelly_layer = Lambda(tf_kelly, name='kelly_fn')(kelly_layer)\n",
    "\n",
    "# Append output to averaging model\n",
    "concat_layer = Concatenate(name=\"summary_layer\")([hidden_layer, wl_output, kelly_layer, odds_input])\n",
    "hidden_layer_2 = Dense(5, activation=\"relu\", kernel_regularizer=regularizer, name=\"summary_hl_1\")(concat_layer)\n",
    "output_layer = Dense(2, activation=\"sigmoid\", name=\"output\")(hidden_layer_2)\n",
    "\n",
    "# Define the model object\n",
    "model = Model(inputs=[main_input, odds_input], outputs=output_layer)\n",
    "\n",
    "# Compile the model\n",
    "optimizer = tf.keras.optimizers.Adam(learning_rate=learning_rate, decay=learning_rate_decay)\n",
    "model.compile(loss=loss, optimizer=optimizer)\n",
    "\n",
    "model.summary()\n",
    "tf.keras.utils.plot_model(model, to_file='model_images/betting_model.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1b24bfba",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_input, val_input = [X_train, O_train], [X_val, O_val]\n",
    "history = model.fit(x=train_input, y=Y_train, validation_data=(val_input, Y_val), \n",
    "                    batch_size=batch_size, epochs=epochs, verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cc33edc0",
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3667d769",
   "metadata": {},
   "outputs": [],
   "source": [
    "def betting_sim(preds, true):\n",
    "    won, home_odds, away_odds = true[:, 0], true[:, 1], true[:, 2]\n",
    "    \n",
    "    bankroll = [100]\n",
    "    unit_size = 10\n",
    "    for i in range(len(preds)):\n",
    "        \n",
    "        if preds[i,1] < 0.5:\n",
    "            bankroll.append(bankroll[-1])\n",
    "            continue\n",
    "        \n",
    "        if preds[i, 0] > 0.5:\n",
    "            # We want to bet home\n",
    "            bet_won, bet_odds = won[i] == True, home_odds[i]\n",
    "        else:\n",
    "            # We want to bet away\n",
    "            bet_won, bet_odds = won[i] == False, away_odds[i]\n",
    "            \n",
    "        if bet_won: \n",
    "            bankroll.append(bankroll[-1] + unit_size*(bet_odds - 1))\n",
    "        else:\n",
    "            bankroll.append(bankroll[-1] - unit_size)\n",
    "            \n",
    "    return bankroll\n",
    "\n",
    "train_preds, test_preds, val_preds = model.predict(X_train), model.predict(X_test), model.predict(X_val)\n",
    "\n",
    "bankroll = betting_sim(train_preds, Y_train)\n",
    "plt.plot(bankroll)\n",
    "plt.show()\n",
    "\n",
    "bankroll = betting_sim(test_preds, Y_test)\n",
    "plt.plot(bankroll)\n",
    "plt.show()\n",
    "\n",
    "bankroll = betting_sim(val_preds, Y_val)\n",
    "plt.plot(bankroll)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "34214913",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}