tsl_profits.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# в этом ноутбуке я считаю профиты для лонг трейдов со скользящим стоплоссом.\n",
    "# собираюсь использовать это как раметку данных.\n",
    "# todo: если посмотреть начало каждого участка, где лонг профит становится положительным, то можно заметить, что в \n",
    "# начале есть участок снижения цены. цена снижается, но размер снижения не выбивает скользящий стоп лосс. иногда \n",
    "# этот участок выглядит как ошибочно помеченным зеленым. видно просто по динамике, что до минимума динамика одна, а \n",
    "# после минимума меняется. то, что видно на глаз бывает затруднительно переложить в алгоритм без использования\n",
    "# нейронок. однако, в данном случае видно, что во время снижения цены, даже не смотря на то, что присутствуют колебания \n",
    "# туда-сюда, не снижается красная линия. не снижается она ровно до тех пор, пока не наступит минимум. надо включить \n",
    "# блять мозги и понять почему так происходит. я тупой. не понимаю. понятно только, что это можно использовать для того, \n",
    "# чтобы улучшить качество разметки и выбросить эти начальные участки, которые ошибочно размечаются как зеленые.\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import pandas_ta as ta\n",
    "import forecast\n",
    "from quote_chart import create_chart_app\n",
    "from plotly.subplots import make_subplots\n",
    "import plotly.graph_objects as go\n",
    "from datetime import datetime\n",
    "from tsl import get_trailing_stop_loss_long_profits, get_trailing_stop_loss_short_profits\n",
    "\n",
    "fee_percent = 0.07"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = forecast.get_df('BTCUSD', '', '1T')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# xx0 = datetime.strptime('2024-05-20 23:00:00.000000', '%Y-%m-%d %H:%M:%S.%f')\n",
    "# xx1 = datetime.strptime('2024-05-20 23:30:00.000000', '%Y-%m-%d %H:%M:%S.%f')\n",
    "# data = data[xx0:xx1]\n",
    "(long_profits, long_stop_loss_prices, maximums) = get_trailing_stop_loss_long_profits(data, 0.5 * 0.01)\n",
    "(short_profits, short_stop_loss_prices, minimums) = get_trailing_stop_loss_short_profits(data, 0.5 * 0.01)\n",
    "# fill None values with last not-None value to avoid issues on the chart.\n",
    "long_profits_df = pd.DataFrame(long_profits)\n",
    "long_profits_df.ffill(inplace=True)\n",
    "#long_profits_df[long_profits_df < 0] = 0\n",
    "data['long_profit'] = long_profits_df.to_numpy()\n",
    "short_profits_df = pd.DataFrame(short_profits)\n",
    "short_profits_df.ffill(inplace=True)\n",
    "#short_profits_df[short_profits_df < 0] = 0\n",
    "data['short_profit'] = short_profits_df.to_numpy()\n",
    "# sl_df = pd.DataFrame(stop_loss_prices)\n",
    "# sl_df.ffill(inplace=True)\n",
    "# data['stop_loss_price'] = sl_df.to_numpy()\n",
    "# maximums_df = pd.DataFrame(maximums)\n",
    "# maximums_df.ffill(inplace=True)\n",
    "# data['maximum'] = maximums_df.to_numpy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "rsi = ta.rsi(data['close'], 94)\n",
    "data['rsi'] = rsi\n",
    "rsi10 = ta.ema(rsi, 40)\n",
    "data['rsi10'] = rsi10\n",
    "rsi90 = ta.ema(rsi, 90)\n",
    "data['rsi90'] = rsi90"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "resampled_df = data.copy()\n",
    "\n",
    "# called when a period button is pressed under the plot.\n",
    "def on_period_change(button_id):\n",
    "    global resampled_df, selected_period\n",
    "    if button_id == '':\n",
    "        return\n",
    "    selected_period = button_id\n",
    "    resampled_df = data.resample(selected_period).agg({\n",
    "        'open': 'first',\n",
    "        'high': 'max',\n",
    "        'low': 'min',\n",
    "        'close': 'last',\n",
    "        'volume': 'sum',\n",
    "        'long_profit': 'first',\n",
    "        'short_profit': 'first',\n",
    "        'rsi': 'first',\n",
    "        'rsi10': 'first',\n",
    "        'rsi90': 'first',\n",
    "    })\n",
    "    resampled_df = resampled_df.dropna()\n",
    "\n",
    "def create_figure(x0, x1):\n",
    "    if x0 is not None:\n",
    "        #print(x0)\n",
    "        df = resampled_df[x0:x1]\n",
    "    else:\n",
    "        # df = resampled_df[:100]#\n",
    "        xx0 = datetime.strptime('2024-05-27 00:00:00.000000', '%Y-%m-%d %H:%M:%S.%f')\n",
    "        xx1 = datetime.strptime('2024-05-28 00:00:00.000000', '%Y-%m-%d %H:%M:%S.%f')\n",
    "        df = resampled_df[xx0:xx1]\n",
    "        \n",
    "    # define multiple panes. The top pane will be for the main price chart with candles. The second pane is for volumes.\n",
    "    fig = make_subplots(rows=2, cols=1, shared_xaxes=True, \n",
    "                vertical_spacing=0.01,\n",
    "                row_heights=[0.8, 0.2],\n",
    "                specs=[[{\"secondary_y\": True}], [{\"secondary_y\": True}]])\n",
    "    fig.add_trace(go.Scatter(x=df.index, y=df['open'] - df['short_profit'], mode='lines', line=dict(color='red'),\n",
    "                             name='short_profit'), row=1, col=1)\n",
    "    fig.add_trace(go.Scatter(x=df.index, y=df['open'] + df['long_profit'], mode='lines', line=dict(color='green'),\n",
    "                             name='long_profit'), row=1, col=1)\n",
    "    fig.add_trace(go.Candlestick(x=df.index,\n",
    "                                 open=df['open'],\n",
    "                                 high=df['high'],\n",
    "                                 low=df['low'],\n",
    "                                 close=df['close'],\n",
    "                                 name='ohlc'), row=1, col=1)\n",
    "    # plot the main chart with price candles.\n",
    "    # fig.add_trace(go.Scatter(x=df.index, y=df['open'] + df['profit'] - (df['maximum'] - df['close']), \n",
    "    #                          mode='lines', line=dict(color='orange'), name='discounted profit'), row=1, col=1)\n",
    "    # fig.add_trace(go.Scatter(x=df.index, y=df['maximum'], mode='lines', line=dict(color='green'), name='maximum'),\n",
    "    #               row=1, col=1)\n",
    "\n",
    "    # fig.add_trace(go.Scatter(x=df.index, y=df['open'] * (1 + 0.01*fee_percent*2), mode='lines', line=dict(color='blue'),\n",
    "    # name='fee'), row=1, col=1)\n",
    "    # fig.add_trace(go.Scatter(x=df.index, y=df['stop_loss_price'], mode='lines', line=dict(color='red'),\n",
    "    #                          name='stop_loss_pirce'), row=1, col=1)\n",
    "    #fig.add_trace(go.Bar(x=df.index, y=df['volume'], name='Volume'), row=2, col=1)\n",
    "    fig.add_trace(go.Scatter(x=df.index, y=df['rsi'], name='rsi', line=dict(color='red')), row=2, col=1)\n",
    "    fig.add_trace(go.Scatter(x=df.index, y=df['rsi10'], name='rsi10', line=dict(color='blue')), row=2, col=1)\n",
    "    fig.add_trace(go.Scatter(x=df.index, y=df['rsi90'], name='rsi90', line=dict(color='black')), row=2, col=1)\n",
    "    # fig.add_trace(go.Scatter(x=df.index, \n",
    "    #                          y=df['open'] - df['profit'] + df['open'] * 0.01*fee_percent*2,\n",
    "    #                          mode='lines', line=dict(color='blue'), name='prof wo fee'), row=1, col=1)\n",
    "    # set the default dragmode to pan, remove the range slider because i use zoom/pan instead of it.\n",
    "    fig.update_layout(\n",
    "        dragmode='pan',\n",
    "        xaxis_rangeslider_visible=False,\n",
    "        width=1200, # px\n",
    "        height=600,\n",
    "        margin=dict(l=50, r=0, t=10, b=0),\n",
    "        yaxis=dict(side='right'),\n",
    "        yaxis2=dict(side='left'),\n",
    "        yaxis3=dict(side='right'),\n",
    "        yaxis4=dict(side='left'),\n",
    "        yaxis5=dict(side='right'),\n",
    "        yaxis6=dict(side='left'),\n",
    "        )\n",
    "    fig.update_xaxes(\n",
    "        ticklabelposition=\"outside right\",  # keep labels on the right so that they don't affect margin-left.\n",
    "    )\n",
    "    return fig\n",
    "\n",
    "app = create_chart_app(create_figure, on_period_change, period_buttons=None, debug=False)\n",
    "app.run_server(debug=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "n=int(data.shape[0]/1100)\n",
    "n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "shift = 1000\n",
    "n=10000"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "fig = make_subplots(rows=1, cols=1, shared_xaxes=True, \n",
    "            vertical_spacing=0.01,\n",
    "            row_heights=[1],\n",
    "            specs=[[{\"secondary_y\": True}]])\n",
    "rsi_diffs = (data['rsi90'][shift:shift+n] - data['rsi10'][shift:shift+n])\n",
    "fig.add_trace(go.Scatter(x=rsi_diffs,\n",
    "                         y=data['short_profit'][shift:shift+n],\n",
    "                         mode='markers',\n",
    "                         marker=dict(color='red', size=1)), row=1, col=1)\n",
    "fig.add_trace(go.Scatter(x=rsi_diffs,\n",
    "                         y=data['long_profit'][shift:shift+n],\n",
    "                         mode='markers',\n",
    "                         marker=dict(color='blue', size=1)), row=1, col=1)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "venv",
   "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}