tutorial/Tutorial2_DeepScalper.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Tutorial 2\n",
    "We show how to train DeepScalper on BTC trading."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 1: Import Packages\n",
    "Modify the system path and load the corresponding packages and functions "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-03-04 18:50:27,916\tINFO services.py:1476 -- View the Ray dashboard at \u001b[1m\u001b[32mhttp://127.0.0.1:8266\u001b[39m\u001b[22m\n",
      "2023-03-04 18:50:30,042\tINFO worker.py:973 -- Calling ray.init() again after it has already been called.\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import sys\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "\n",
    "ROOT = os.path.dirname(os.path.abspath(\".\"))\n",
    "sys.path.append(ROOT)\n",
    "\n",
    "import torch\n",
    "import argparse\n",
    "import os.path as osp\n",
    "from mmcv import Config\n",
    "from trademaster.utils import replace_cfg_vals\n",
    "from trademaster.nets.builder import build_net\n",
    "from trademaster.environments.builder import build_environment\n",
    "from trademaster.datasets.builder import build_dataset\n",
    "from trademaster.agents.builder import build_agent\n",
    "from trademaster.optimizers.builder import build_optimizer\n",
    "from trademaster.losses.builder import build_loss\n",
    "from trademaster.trainers.builder import build_trainer\n",
    "from trademaster.transition.builder import build_transition\n",
    "from trademaster.utils import plot\n",
    "from trademaster.utils import set_seed\n",
    "set_seed(2023)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 2: Load Configs\n",
    "Load default config from the file `configs/algorithmic_trading/algorithmic_trading_BTC_deepscalper_deepscalper_adam_mse.py`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "parser = argparse.ArgumentParser(description='Download Alpaca Datasets')\n",
    "parser.add_argument(\"--config\", default=osp.join(ROOT, \"configs\", \"algorithmic_trading\", \"algorithmic_trading_BTC_deepscalper_deepscalper_adam_mse.py\"),\n",
    "                    help=\"download datasets config file path\")\n",
    "parser.add_argument(\"--task_name\", type=str, default=\"train\")\n",
    "args, _= parser.parse_known_args()\n",
    "\n",
    "cfg = Config.fromfile(args.config)\n",
    "task_name = args.task_name\n",
    "cfg = replace_cfg_vals(cfg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Config (path: /data/sunshuo/qml/TradeMaster/configs/algorithmic_trading/algorithmic_trading_BTC_deepscalper_deepscalper_adam_mse.py): {'data': {'type': 'AlgorithmicTradingDataset', 'data_path': 'data/algorithmic_trading/BTC', 'train_path': 'data/algorithmic_trading/BTC/train.csv', 'valid_path': 'data/algorithmic_trading/BTC/valid.csv', 'test_path': 'data/algorithmic_trading/BTC/test.csv', 'test_dynamic_path': 'data/algorithmic_trading/BTC/test_labeled_3_24_-0.15_0.15.csv', 'tech_indicator_list': ['high', 'low', 'open', 'close', 'adjcp', 'zopen', 'zhigh', 'zlow', 'zadjcp', 'zclose', 'zd_5', 'zd_10', 'zd_15', 'zd_20', 'zd_25', 'zd_30'], 'backward_num_day': 5, 'forward_num_day': 5, 'test_dynamic': '-1'}, 'environment': {'type': 'AlgorithmicTradingEnvironment'}, 'agent': {'type': 'AlgorithmicTradingDQN', 'max_step': 12345, 'reward_scale': 1, 'repeat_times': 1, 'gamma': 0.9, 'batch_size': 64, 'clip_grad_norm': 3.0, 'soft_update_tau': 0, 'state_value_tau': 0.005}, 'trainer': {'type': 'AlgorithmicTradingTrainer', 'epochs': 2, 'work_dir': 'work_dir/algorithmic_trading_BTC_deepscalper_deepscalper_adam_mse', 'seeds_list': (12345,), 'batch_size': 64, 'horizon_len': 128, 'buffer_size': 1000000.0, 'num_threads': 8, 'if_remove': False, 'if_discrete': True, 'if_off_policy': True, 'if_keep_save': True, 'if_over_write': False, 'if_save_buffer': False}, 'loss': {'type': 'MSELoss'}, 'optimizer': {'type': 'Adam', 'lr': 0.001}, 'act': {'type': 'QNet', 'state_dim': 82, 'action_dim': 3, 'dims': (64, 32), 'explore_rate': 0.25}, 'cri': None, 'transition': {'type': 'Transition'}, 'task_name': 'algorithmic_trading', 'dataset_name': 'BTC', 'optimizer_name': 'adam', 'loss_name': 'mse', 'net_name': 'deepscalper', 'agent_name': 'deepscalper', 'work_dir': 'work_dir/algorithmic_trading_BTC_deepscalper_deepscalper_adam_mse', 'batch_size': 64}"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cfg"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 3: Build Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = build_dataset(cfg)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 4: Build Reinforcement Learning Environments\n",
    "Build environments based on cfg and predefined dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_environment = build_environment(cfg, default_args=dict(dataset=dataset, task=\"train\"))\n",
    "valid_environment = build_environment(cfg, default_args=dict(dataset=dataset, task=\"valid\"))\n",
    "test_environment = build_environment(cfg, default_args=dict(dataset=dataset, task=\"test\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "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 th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>tic</th>\n",
       "      <th>date</th>\n",
       "      <th>high</th>\n",
       "      <th>low</th>\n",
       "      <th>open</th>\n",
       "      <th>close</th>\n",
       "      <th>adjcp</th>\n",
       "      <th>zopen</th>\n",
       "      <th>zhigh</th>\n",
       "      <th>zlow</th>\n",
       "      <th>zadjcp</th>\n",
       "      <th>zclose</th>\n",
       "      <th>zd_5</th>\n",
       "      <th>zd_10</th>\n",
       "      <th>zd_15</th>\n",
       "      <th>zd_20</th>\n",
       "      <th>zd_25</th>\n",
       "      <th>zd_30</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>BTC</td>\n",
       "      <td>2016-01-24 23:59:59</td>\n",
       "      <td>405.484985</td>\n",
       "      <td>387.510010</td>\n",
       "      <td>388.101990</td>\n",
       "      <td>402.971008</td>\n",
       "      <td>402.971008</td>\n",
       "      <td>-0.036898</td>\n",
       "      <td>0.006239</td>\n",
       "      <td>-0.038368</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.039949</td>\n",
       "      <td>-0.005662</td>\n",
       "      <td>-0.030965</td>\n",
       "      <td>0.009124</td>\n",
       "      <td>0.032291</td>\n",
       "      <td>0.040384</td>\n",
       "      <td>0.042514</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>BTC</td>\n",
       "      <td>2016-01-25 23:59:59</td>\n",
       "      <td>402.316986</td>\n",
       "      <td>388.553986</td>\n",
       "      <td>402.316986</td>\n",
       "      <td>391.726013</td>\n",
       "      <td>391.726013</td>\n",
       "      <td>0.027037</td>\n",
       "      <td>0.027037</td>\n",
       "      <td>-0.008098</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-0.027905</td>\n",
       "      <td>0.008328</td>\n",
       "      <td>0.003846</td>\n",
       "      <td>0.028517</td>\n",
       "      <td>0.056789</td>\n",
       "      <td>0.066283</td>\n",
       "      <td>0.070267</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>BTC</td>\n",
       "      <td>2016-01-26 23:59:59</td>\n",
       "      <td>397.765991</td>\n",
       "      <td>390.575012</td>\n",
       "      <td>392.002014</td>\n",
       "      <td>392.153015</td>\n",
       "      <td>392.153015</td>\n",
       "      <td>-0.000385</td>\n",
       "      <td>0.014313</td>\n",
       "      <td>-0.004024</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001090</td>\n",
       "      <td>-0.002005</td>\n",
       "      <td>0.003930</td>\n",
       "      <td>0.017830</td>\n",
       "      <td>0.050927</td>\n",
       "      <td>0.060819</td>\n",
       "      <td>0.066495</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>BTC</td>\n",
       "      <td>2016-01-27 23:59:59</td>\n",
       "      <td>396.842987</td>\n",
       "      <td>391.782013</td>\n",
       "      <td>392.444000</td>\n",
       "      <td>394.971985</td>\n",
       "      <td>394.971985</td>\n",
       "      <td>-0.006400</td>\n",
       "      <td>0.004737</td>\n",
       "      <td>-0.008076</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.007188</td>\n",
       "      <td>-0.002809</td>\n",
       "      <td>-0.000027</td>\n",
       "      <td>0.003693</td>\n",
       "      <td>0.035441</td>\n",
       "      <td>0.049353</td>\n",
       "      <td>0.056578</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>BTC</td>\n",
       "      <td>2016-01-28 23:59:59</td>\n",
       "      <td>395.502014</td>\n",
       "      <td>379.734985</td>\n",
       "      <td>395.145996</td>\n",
       "      <td>380.289001</td>\n",
       "      <td>380.289001</td>\n",
       "      <td>0.039068</td>\n",
       "      <td>0.040004</td>\n",
       "      <td>-0.001457</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-0.037175</td>\n",
       "      <td>0.031905</td>\n",
       "      <td>0.036773</td>\n",
       "      <td>0.033315</td>\n",
       "      <td>0.065830</td>\n",
       "      <td>0.084638</td>\n",
       "      <td>0.092754</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   tic                 date        high         low        open       close  \\\n",
       "0  BTC  2016-01-24 23:59:59  405.484985  387.510010  388.101990  402.971008   \n",
       "1  BTC  2016-01-25 23:59:59  402.316986  388.553986  402.316986  391.726013   \n",
       "2  BTC  2016-01-26 23:59:59  397.765991  390.575012  392.002014  392.153015   \n",
       "3  BTC  2016-01-27 23:59:59  396.842987  391.782013  392.444000  394.971985   \n",
       "4  BTC  2016-01-28 23:59:59  395.502014  379.734985  395.145996  380.289001   \n",
       "\n",
       "        adjcp     zopen     zhigh      zlow  zadjcp    zclose      zd_5  \\\n",
       "0  402.971008 -0.036898  0.006239 -0.038368     0.0  0.039949 -0.005662   \n",
       "1  391.726013  0.027037  0.027037 -0.008098     0.0 -0.027905  0.008328   \n",
       "2  392.153015 -0.000385  0.014313 -0.004024     0.0  0.001090 -0.002005   \n",
       "3  394.971985 -0.006400  0.004737 -0.008076     0.0  0.007188 -0.002809   \n",
       "4  380.289001  0.039068  0.040004 -0.001457     0.0 -0.037175  0.031905   \n",
       "\n",
       "      zd_10     zd_15     zd_20     zd_25     zd_30  \n",
       "0 -0.030965  0.009124  0.032291  0.040384  0.042514  \n",
       "1  0.003846  0.028517  0.056789  0.066283  0.070267  \n",
       "2  0.003930  0.017830  0.050927  0.060819  0.066495  \n",
       "3 -0.000027  0.003693  0.035441  0.049353  0.056578  \n",
       "4  0.036773  0.033315  0.065830  0.084638  0.092754  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_environment.df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 5: Build Net \n",
    "Update information about the state and action dimension. Create networks and optimizer for DeepScalper\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "action_dim = train_environment.action_dim\n",
    "state_dim = train_environment.state_dim\n",
    "\n",
    "cfg.act.update(dict(action_dim=action_dim, state_dim=state_dim))\n",
    "act = build_net(cfg.act)\n",
    "act_optimizer = build_optimizer(cfg, default_args=dict(params=act.parameters()))\n",
    "if cfg.cri:\n",
    "    cfg.cri.update(dict(action_dim=action_dim, state_dim=state_dim))\n",
    "    cri = build_net(cfg.cri)\n",
    "    cri_optimizer = build_optimizer(cfg, default_args=dict(params=cri.parameters()))\n",
    "else:\n",
    "    cri = None\n",
    "    cri_optimizer = None"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 6: Build Loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "criterion = build_loss(cfg)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 7: Build Transition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "transition = build_transition(cfg)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 8: Build Agent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "agent = build_agent(cfg, default_args=dict(action_dim = action_dim,state_dim = state_dim,act = act,cri = cri,act_optimizer = act_optimizer,cri_optimizer = cri_optimizer, criterion = criterion,transition = transition,device=device))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 9: Build Trainer\n",
    "Build trainer from config and create work directionary to save the result, model and config"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| Arguments Keep work_dir: /data/sunshuo/qml/TradeMaster/work_dir/algorithmic_trading_BTC_deepscalper_deepscalper_adam_mse\n"
     ]
    }
   ],
   "source": [
    "trainer = build_trainer(cfg, default_args=dict(train_environment=train_environment,valid_environment=valid_environment,test_environment=test_environment,agent=agent,device=device))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 10: Train the RL Agent\n",
    "Train the trainer based on the config and save results in workdir"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Episode: [1/2]\n",
      "+--------------+-------------+------------+--------------+\n",
      "| Total Return | Sharp Ratio | Volatility | Max Drawdown |\n",
      "+--------------+-------------+------------+--------------+\n",
      "| 663.053338%  |   1.465309  | 3.726925%  |  63.966805%  |\n",
      "+--------------+-------------+------------+--------------+\n",
      "Valid Episode: [1/2]\n",
      "+--------------+-------------+------------+--------------+\n",
      "| Total Return | Sharp Ratio | Volatility | Max Drawdown |\n",
      "+--------------+-------------+------------+--------------+\n",
      "| -35.444878%  |   0.430537  | 42.560375% | 118.689588%  |\n",
      "+--------------+-------------+------------+--------------+\n",
      "Train Episode: [2/2]\n",
      "+--------------+-------------+------------+--------------+\n",
      "| Total Return | Sharp Ratio | Volatility | Max Drawdown |\n",
      "+--------------+-------------+------------+--------------+\n",
      "| 967.324329%  |   1.621369  | 3.868479%  |  64.052556%  |\n",
      "+--------------+-------------+------------+--------------+\n",
      "Valid Episode: [2/2]\n",
      "+--------------+-------------+------------+--------------+\n",
      "| Total Return | Sharp Ratio | Volatility | Max Drawdown |\n",
      "+--------------+-------------+------------+--------------+\n",
      "| -199.576611% |   0.292613  | 15.006834% | 289.010033%  |\n",
      "+--------------+-------------+------------+--------------+\n"
     ]
    }
   ],
   "source": [
    "trainer.train_and_valid()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 11: RL Agent Testing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Resume checkpoint /data/sunshuo/qml/TradeMaster/work_dir/algorithmic_trading_BTC_deepscalper_deepscalper_adam_mse/checkpoints/best.pth\n",
      "Test Best Episode\n",
      "+--------------+-------------+------------+--------------+\n",
      "| Total Return | Sharp Ratio | Volatility | Max Drawdown |\n",
      "+--------------+-------------+------------+--------------+\n",
      "| 218.177594%  |   1.966325  | 2.131242%  |  29.503340%  |\n",
      "+--------------+-------------+------------+--------------+\n"
     ]
    }
   ],
   "source": [
    "trainer.test();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot(trainer.test_environment.save_asset_memory(),alg=\"DeepScalper\")"
   ]
  }
 ],
 "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": "3.7.13"
  },
  "vscode": {
   "interpreter": {
    "hash": "a1f97403911abd3f02553c8f2b0c54537fddc7efadd9f5d3e31784db6e40c347"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}