demo.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Forex direction predictor\n",
    "\n",
    "This project is a proof-of-concept of how historical forex data can be used to determine the direction of the price (UP/DOWN) using deep learning.\n",
    "We use historical EURUSD data from 2021-2023 to train, validate and test a model which predicts the direction of new data better than chance (63% accuracy in the test set).\n",
    "\n",
    "Note that the model shown here is quite simple and much more can be done to improve the results.For more information and more advanced models contact me at omcandido(at)gmail.com."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from utils.dataset import *\n",
    "from utils.model import *\n",
    "from utils.utils import *\n",
    "\n",
    "import pathlib\n",
    "import plotly.graph_objects as go\n",
    "from torchvision import transforms\n",
    "from torch.utils.data import DataLoader\n",
    "\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "DEVICE = 'cuda' if torch.cuda.is_available() else 'cpu'\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## The historical data\n",
    "The folder `eurusd` contains the 1-min OHLC (Open, High, Low, Close) ticks of the EUR/USD forex pair from 2015 to 2023. The historical data has been downloaded from [HistData.com](https://www.histdata.com/).\n",
    "\n",
    "Let's load the data corresponding to the 2020 year and plot the first week of data using 15min candlesticks."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1/1 [00:00<00:00,  2.01it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sorting data...\n",
      "Done!\n"
     ]
    },
    {
     "data": {
      "application/vnd.plotly.v1+json": {
       "config": {
        "plotlyServerURL": "https://plot.ly"
       },
       "data": [
        {
         "close": [
          1.1214300394058228,
          1.121880054473877,
          1.1218299865722656,
          1.1220899820327759,
          1.1222200393676758,
          1.1213200092315674,
          1.1201200485229492,
          1.1204299926757812,
          1.12076997756958,
          1.1209800243377686,
          1.1211999654769897,
          1.120300054550171,
          1.11981999874115,
          1.1184799671173096,
          1.1193000078201294,
          1.119629979133606,
          1.1174099445343018,
          1.1200499534606934,
          1.116610050201416,
          1.116979956626892,
          1.1170200109481812,
          1.116629958152771,
          1.1169300079345703,
          1.1172200441360474,
          1.1172599792480469,
          1.1170300245285034,
          1.117550015449524,
          1.117550015449524,
          1.1172200441360474,
          1.117400050163269,
          1.117050051689148,
          1.1164799928665161,
          1.1168400049209595,
          1.1158100366592407,
          1.115149974822998,
          1.114259958267212,
          1.113360047340393,
          1.11312997341156,
          1.1141799688339233,
          1.1137200593948364,
          1.1157699823379517,
          1.116320013999939,
          1.1173100471496582,
          1.1171499490737915,
          1.1167199611663818,
          1.1165200471878052,
          1.1161500215530396,
          1.1164900064468384,
          1.1163400411605835,
          1.1165399551391602,
          1.1164799928665161,
          1.1163300275802612,
          1.1167399883270264,
          1.116569995880127,
          1.1163899898529053,
          1.116129994392395,
          1.1162500381469727,
          1.1165000200271606,
          1.1176799535751343,
          1.1190199851989746,
          1.1202800273895264,
          1.119920015335083,
          1.1198300123214722,
          1.1196999549865723,
          1.1187800168991089,
          1.1184600591659546,
          1.1189600229263306,
          1.1194000244140625,
          1.119189977645874,
          1.119089961051941,
          1.1193000078201294,
          1.1196000576019287,
          1.1196099519729614,
          1.1194700002670288,
          1.1191799640655518,
          1.119230031967163,
          1.119189977645874,
          1.119189977645874,
          1.1191999912261963,
          1.1187900304794312,
          1.1183199882507324,
          1.1176999807357788,
          1.11844003200531,
          1.1182700395584106,
          1.11735999584198,
          1.1179399490356445,
          1.1173700094223022,
          1.1159600019454956,
          1.1147199869155884,
          1.1143300533294678,
          1.114050030708313,
          1.114359974861145,
          1.1142499446868896,
          1.1146899461746216,
          1.1144399642944336,
          1.1148500442504883,
          1.1149799823760986,
          1.1156500577926636,
          1.1161500215530396,
          1.1154999732971191,
          1.1157499551773071,
          1.1154899597167969,
          1.1148699522018433,
          1.1150699853897095,
          1.1152000427246094,
          1.1152399778366089,
          1.1133999824523926,
          1.1134400367736816,
          1.1128400564193726,
          1.1126199960708618,
          1.111780047416687,
          1.1120100021362305,
          1.113010048866272,
          1.1115299463272095,
          1.112030029296875,
          1.1114599704742432,
          1.1111899614334106,
          1.1110299825668335,
          1.1109999418258667,
          1.1106799840927124,
          1.1108299493789673,
          1.1112699508666992,
          1.11149001121521,
          1.1118099689483643,
          1.1114699840545654,
          1.1113500595092773,
          1.1115299463272095,
          1.1117199659347534,
          1.1114100217819214,
          1.1107200384140015,
          1.1117899417877197,
          1.109779953956604,
          1.111050009727478,
          1.1111500263214111,
          1.1108700037002563,
          1.1108700037002563,
          1.1107900142669678,
          1.1096099615097046,
          1.1104600429534912,
          1.1103999614715576,
          1.1107800006866455,
          1.1106300354003906,
          1.110610008239746,
          1.1104700565338135,
          1.1107100248336792,
          1.110759973526001,
          1.1106699705123901,
          1.1104300022125244,
          1.1107800006866455,
          1.1107399463653564,
          1.1107900142669678,
          1.1109700202941895,
          1.1109399795532227,
          1.110550045967102,
          1.1096800565719604,
          1.10971999168396,
          1.1087499856948853,
          1.1092699766159058,
          1.1093599796295166,
          1.1096800565719604,
          1.1108399629592896,
          1.111240029335022,
          1.1123499870300293,
          1.112089991569519,
          1.1119600534439087,
          1.1120599508285522,
          1.1121599674224854,
          1.1119999885559082
         ],
         "high": [
          1.1216599941253662,
          1.1221799850463867,
          1.121899962425232,
          1.122439980506897,
          1.1224499940872192,
          1.1223100423812866,
          1.1214300394058228,
          1.1204400062561035,
          1.12076997756958,
          1.121019959449768,
          1.1213699579238892,
          1.1214100122451782,
          1.1203900575637817,
          1.12021005153656,
          1.1194299459457397,
          1.1196999549865723,
          1.1196500062942505,
          1.120229959487915,
          1.120300054550171,
          1.1171900033950806,
          1.1172300577163696,
          1.1170300245285034,
          1.1171499490737915,
          1.1173499822616577,
          1.117300033569336,
          1.1173399686813354,
          1.1175999641418457,
          1.117840051651001,
          1.1177400350570679,
          1.117419958114624,
          1.1174900531768799,
          1.117050051689148,
          1.1169500350952148,
          1.1171499490737915,
          1.1160399913787842,
          1.1152199506759644,
          1.1143200397491455,
          1.113360047340393,
          1.114359974861145,
          1.1148799657821655,
          1.1166900396347046,
          1.1167399883270264,
          1.1179800033569336,
          1.1173900365829468,
          1.117319941520691,
          1.1167700290679932,
          1.1165800094604492,
          1.11667001247406,
          1.1164900064468384,
          1.1167700290679932,
          1.1168500185012817,
          1.116529941558838,
          1.1167900562286377,
          1.1170899868011475,
          1.116610050201416,
          1.1164900064468384,
          1.1163599491119385,
          1.1167000532150269,
          1.1181399822235107,
          1.119439959526062,
          1.120460033416748,
          1.1205500364303589,
          1.1200200319290161,
          1.1203199625015259,
          1.1197500228881836,
          1.1190199851989746,
          1.1190099716186523,
          1.119629979133606,
          1.119789958000183,
          1.1192799806594849,
          1.1193900108337402,
          1.1198899745941162,
          1.1197500228881836,
          1.119629979133606,
          1.1195299625396729,
          1.1196999549865723,
          1.119339942932129,
          1.1194299459457397,
          1.119339942932129,
          1.119230031967163,
          1.1189700365066528,
          1.1185400485992432,
          1.1189600229263306,
          1.1187000274658203,
          1.1185599565505981,
          1.1184899806976318,
          1.1183099746704102,
          1.1175700426101685,
          1.1162199974060059,
          1.1151599884033203,
          1.114359974861145,
          1.1147300004959106,
          1.1149499416351318,
          1.1147700548171997,
          1.1147099733352661,
          1.1153600215911865,
          1.1151599884033203,
          1.115689992904663,
          1.116819977760315,
          1.116320013999939,
          1.116129994392395,
          1.1158599853515625,
          1.1155699491500854,
          1.1151299476623535,
          1.1153899431228638,
          1.1154899597167969,
          1.1152900457382202,
          1.1144700050354004,
          1.1134999990463257,
          1.1131099462509155,
          1.1130399703979492,
          1.1122699975967407,
          1.1130800247192383,
          1.113070011138916,
          1.1121400594711304,
          1.112589955329895,
          1.1115800142288208,
          1.1112899780273438,
          1.1114799976348877,
          1.1110199689865112,
          1.1114599704742432,
          1.1114399433135986,
          1.111590027809143,
          1.1119500398635864,
          1.111899971961975,
          1.1117899417877197,
          1.1115800142288208,
          1.111899971961975,
          1.1117299795150757,
          1.1120400428771973,
          1.111899971961975,
          1.1118299961090088,
          1.1111899614334106,
          1.1112600564956665,
          1.1116700172424316,
          1.1117299795150757,
          1.1112099885940552,
          1.1110700368881226,
          1.1107200384140015,
          1.1111299991607666,
          1.110859990119934,
          1.1109800338745117,
          1.1107300519943237,
          1.1106899976730347,
          1.1107300519943237,
          1.1109399795532227,
          1.1109100580215454,
          1.1107300519943237,
          1.1108299493789673,
          1.1108700037002563,
          1.111109972000122,
          1.1109999418258667,
          1.1110999584197998,
          1.1109700202941895,
          1.1108399629592896,
          1.1100000143051147,
          1.1098400354385376,
          1.1096199750900269,
          1.1099599599838257,
          1.11135995388031,
          1.1108499765396118,
          1.1117899417877197,
          1.1123499870300293,
          1.1129100322723389,
          1.112149953842163,
          1.1123199462890625,
          1.1122499704360962,
          1.1122299432754517
         ],
         "low": [
          1.1210600137710571,
          1.1214200258255005,
          1.1215699911117554,
          1.1217999458312988,
          1.121840000152588,
          1.1212400197982788,
          1.1200799942016602,
          1.1200900077819824,
          1.120360016822815,
          1.1204299926757812,
          1.1205099821090698,
          1.120110034942627,
          1.1192699670791626,
          1.1184200048446655,
          1.1182899475097656,
          1.1189600229263306,
          1.1169999837875366,
          1.1170799732208252,
          1.116510033607483,
          1.1163500547409058,
          1.1163599491119385,
          1.116420030593872,
          1.1165900230407715,
          1.1168899536132812,
          1.1168700456619263,
          1.116979956626892,
          1.116979956626892,
          1.1172200441360474,
          1.1171499490737915,
          1.1169500350952148,
          1.1168500185012817,
          1.116379976272583,
          1.1164300441741943,
          1.1152000427246094,
          1.114989995956421,
          1.1139899492263794,
          1.113070011138916,
          1.1124900579452515,
          1.1128100156784058,
          1.1135300397872925,
          1.1136900186538696,
          1.1153600215911865,
          1.1162699460983276,
          1.1167900562286377,
          1.1164499521255493,
          1.1161199808120728,
          1.1160000562667847,
          1.1152299642562866,
          1.1158900260925293,
          1.1160000562667847,
          1.11627995967865,
          1.1161199808120728,
          1.1162400245666504,
          1.1164699792861938,
          1.1161600351333618,
          1.1160600185394287,
          1.1157100200653076,
          1.1159700155258179,
          1.116320013999939,
          1.1175099611282349,
          1.1189899444580078,
          1.1198999881744385,
          1.1195100545883179,
          1.1191099882125854,
          1.1182500123977661,
          1.118209958076477,
          1.118090033531189,
          1.1189700365066528,
          1.11913001537323,
          1.1187900304794312,
          1.1188900470733643,
          1.1192799806594849,
          1.1195100545883179,
          1.1193000078201294,
          1.1189099550247192,
          1.1191200017929077,
          1.119089961051941,
          1.1191099882125854,
          1.1190199851989746,
          1.118090033531189,
          1.1178200244903564,
          1.1172599792480469,
          1.1173700094223022,
          1.1180499792099,
          1.1164799928665161,
          1.1172000169754028,
          1.1170799732208252,
          1.1154799461364746,
          1.114150047302246,
          1.113420009613037,
          1.1133500337600708,
          1.1140300035476685,
          1.114169955253601,
          1.1141999959945679,
          1.1142300367355347,
          1.1144200563430786,
          1.1144700050354004,
          1.1147500276565552,
          1.1155099868774414,
          1.1153899431228638,
          1.1153099536895752,
          1.1153600215911865,
          1.1145000457763672,
          1.1147300004959106,
          1.114609956741333,
          1.1143200397491455,
          1.113010048866272,
          1.112529993057251,
          1.1125400066375732,
          1.1119600534439087,
          1.1113300323486328,
          1.111109972000122,
          1.1117600202560425,
          1.1115200519561768,
          1.1108700037002563,
          1.1113100051879883,
          1.111009955406189,
          1.1108100414276123,
          1.110949993133545,
          1.110159993171692,
          1.1106599569320679,
          1.11080002784729,
          1.111009955406189,
          1.1110999584197998,
          1.1114100217819214,
          1.111340045928955,
          1.1112699508666992,
          1.1115100383758545,
          1.1110999584197998,
          1.1105999946594238,
          1.110509991645813,
          1.1096700429916382,
          1.109660029411316,
          1.1103299856185913,
          1.1108200550079346,
          1.1105799674987793,
          1.1103800535202026,
          1.1092400550842285,
          1.1093499660491943,
          1.1103299856185913,
          1.1103500127792358,
          1.1104999780654907,
          1.1103800535202026,
          1.1102800369262695,
          1.1104400157928467,
          1.1106300354003906,
          1.1105899810791016,
          1.110200047492981,
          1.110360026359558,
          1.110640048980713,
          1.1106799840927124,
          1.1107499599456787,
          1.1106799840927124,
          1.1104199886322021,
          1.109320044517517,
          1.1091200113296509,
          1.1085000038146973,
          1.108680009841919,
          1.1092599630355835,
          1.1089799404144287,
          1.108620047569275,
          1.110450029373169,
          1.111109972000122,
          1.111930012702942,
          1.1117099523544312,
          1.1118500232696533,
          1.1119799613952637,
          1.1114100217819214
         ],
         "open": [
          1.1211999654769897,
          1.1214300394058228,
          1.121880054473877,
          1.1218199729919434,
          1.1220999956130981,
          1.1222399473190308,
          1.1213200092315674,
          1.120110034942627,
          1.1204400062561035,
          1.1207799911499023,
          1.1209900379180908,
          1.1211899518966675,
          1.120300054550171,
          1.1198400259017944,
          1.1184699535369873,
          1.1192899942398071,
          1.119629979133606,
          1.1173900365829468,
          1.1200599670410156,
          1.116610050201416,
          1.116979956626892,
          1.1170200109481812,
          1.1166199445724487,
          1.116919994354248,
          1.1172200441360474,
          1.1172599792480469,
          1.1170300245285034,
          1.117550015449524,
          1.117550015449524,
          1.1172200441360474,
          1.1174099445343018,
          1.1170400381088257,
          1.1164900064468384,
          1.1168400049209595,
          1.115820050239563,
          1.1151299476623535,
          1.1142799854278564,
          1.113360047340393,
          1.1131199598312378,
          1.1141799688339233,
          1.1137100458145142,
          1.1157599687576294,
          1.116320013999939,
          1.1173299551010132,
          1.1171400547027588,
          1.1167099475860596,
          1.116529941558838,
          1.116170048713684,
          1.1164900064468384,
          1.1163300275802612,
          1.116529941558838,
          1.1164799928665161,
          1.1163400411605835,
          1.1167399883270264,
          1.1165800094604492,
          1.1164100170135498,
          1.116129994392395,
          1.1162300109863281,
          1.1165000200271606,
          1.1176899671554565,
          1.1190099716186523,
          1.120270013809204,
          1.119920015335083,
          1.11981999874115,
          1.11968994140625,
          1.1187700033187866,
          1.1184699535369873,
          1.118980050086975,
          1.1194000244140625,
          1.1192100048065186,
          1.1190999746322632,
          1.1193100214004517,
          1.1196000576019287,
          1.1196099519729614,
          1.1194900274276733,
          1.1191699504852295,
          1.1192400455474854,
          1.119189977645874,
          1.1191999912261963,
          1.1191999912261963,
          1.1187800168991089,
          1.1183199882507324,
          1.1176999807357788,
          1.1184500455856323,
          1.118280053138733,
          1.1173700094223022,
          1.1179399490356445,
          1.11735999584198,
          1.1159600019454956,
          1.1147199869155884,
          1.1143399477005005,
          1.1140300035476685,
          1.114359974861145,
          1.1142499446868896,
          1.1146899461746216,
          1.1144299507141113,
          1.1148500442504883,
          1.1149799823760986,
          1.115630030632019,
          1.1161400079727173,
          1.1154999732971191,
          1.1157499551773071,
          1.1154899597167969,
          1.1148799657821655,
          1.1150799989700317,
          1.1152000427246094,
          1.1152399778366089,
          1.1133999824523926,
          1.1134400367736816,
          1.1128300428390503,
          1.1125999689102173,
          1.1117899417877197,
          1.1120100021362305,
          1.1130199432373047,
          1.1115399599075317,
          1.1120400428771973,
          1.1114699840545654,
          1.1111799478530884,
          1.1110299825668335,
          1.111009955406189,
          1.1106799840927124,
          1.1108299493789673,
          1.1112699508666992,
          1.11149001121521,
          1.1118199825286865,
          1.1114699840545654,
          1.111340045928955,
          1.1115200519561768,
          1.1117299795150757,
          1.1114200353622437,
          1.1107300519943237,
          1.111780047416687,
          1.1097899675369263,
          1.1110600233078003,
          1.1111600399017334,
          1.1108800172805786,
          1.1108700037002563,
          1.1107900142669678,
          1.1096099615097046,
          1.110450029373169,
          1.1103899478912354,
          1.11080002784729,
          1.110640048980713,
          1.110610008239746,
          1.1104799509048462,
          1.1106799840927124,
          1.1107699871063232,
          1.1106599569320679,
          1.1104300022125244,
          1.1107699871063232,
          1.1107499599456787,
          1.1107900142669678,
          1.1109700202941895,
          1.1109299659729004,
          1.1105400323867798,
          1.1096700429916382,
          1.1097300052642822,
          1.1087499856948853,
          1.1092699766159058,
          1.1093699932098389,
          1.1096899509429932,
          1.110859990119934,
          1.1112300157546997,
          1.1123499870300293,
          1.112089991569519,
          1.1119600534439087,
          1.1120699644088745,
          1.112149953842163
         ],
         "type": "candlestick",
         "x": [
          "2020-01-01T17:00:00",
          "2020-01-01T18:00:00",
          "2020-01-01T19:00:00",
          "2020-01-01T20:00:00",
          "2020-01-01T21:00:00",
          "2020-01-01T22:00:00",
          "2020-01-01T23:00:00",
          "2020-01-02T00:00:00",
          "2020-01-02T01:00:00",
          "2020-01-02T02:00:00",
          "2020-01-02T03:00:00",
          "2020-01-02T04:00:00",
          "2020-01-02T05:00:00",
          "2020-01-02T06:00:00",
          "2020-01-02T07:00:00",
          "2020-01-02T08:00:00",
          "2020-01-02T09:00:00",
          "2020-01-02T10:00:00",
          "2020-01-02T11:00:00",
          "2020-01-02T12:00:00",
          "2020-01-02T13:00:00",
          "2020-01-02T14:00:00",
          "2020-01-02T15:00:00",
          "2020-01-02T16:00:00",
          "2020-01-02T17:00:00",
          "2020-01-02T18:04:00",
          "2020-01-02T19:04:00",
          "2020-01-02T20:04:00",
          "2020-01-02T21:04:00",
          "2020-01-02T22:04:00",
          "2020-01-02T23:04:00",
          "2020-01-03T00:04:00",
          "2020-01-03T01:04:00",
          "2020-01-03T02:04:00",
          "2020-01-03T03:04:00",
          "2020-01-03T04:04:00",
          "2020-01-03T05:04:00",
          "2020-01-03T06:04:00",
          "2020-01-03T07:04:00",
          "2020-01-03T08:04:00",
          "2020-01-03T09:04:00",
          "2020-01-03T10:04:00",
          "2020-01-03T11:04:00",
          "2020-01-03T12:04:00",
          "2020-01-03T13:04:00",
          "2020-01-03T14:04:00",
          "2020-01-03T15:04:00",
          "2020-01-03T16:05:00",
          "2020-01-05T17:06:00",
          "2020-01-05T18:08:00",
          "2020-01-05T19:08:00",
          "2020-01-05T20:08:00",
          "2020-01-05T21:08:00",
          "2020-01-05T22:08:00",
          "2020-01-05T23:08:00",
          "2020-01-06T00:08:00",
          "2020-01-06T01:08:00",
          "2020-01-06T02:08:00",
          "2020-01-06T03:08:00",
          "2020-01-06T04:08:00",
          "2020-01-06T05:08:00",
          "2020-01-06T06:08:00",
          "2020-01-06T07:08:00",
          "2020-01-06T08:08:00",
          "2020-01-06T09:08:00",
          "2020-01-06T10:08:00",
          "2020-01-06T11:08:00",
          "2020-01-06T12:08:00",
          "2020-01-06T13:08:00",
          "2020-01-06T14:08:00",
          "2020-01-06T15:08:00",
          "2020-01-06T16:08:00",
          "2020-01-06T17:08:00",
          "2020-01-06T18:09:00",
          "2020-01-06T19:09:00",
          "2020-01-06T20:09:00",
          "2020-01-06T21:09:00",
          "2020-01-06T22:09:00",
          "2020-01-06T23:09:00",
          "2020-01-07T00:09:00",
          "2020-01-07T01:09:00",
          "2020-01-07T02:09:00",
          "2020-01-07T03:09:00",
          "2020-01-07T04:09:00",
          "2020-01-07T05:09:00",
          "2020-01-07T06:09:00",
          "2020-01-07T07:09:00",
          "2020-01-07T08:09:00",
          "2020-01-07T09:09:00",
          "2020-01-07T10:09:00",
          "2020-01-07T11:09:00",
          "2020-01-07T12:09:00",
          "2020-01-07T13:09:00",
          "2020-01-07T14:09:00",
          "2020-01-07T15:09:00",
          "2020-01-07T16:09:00",
          "2020-01-07T17:09:00",
          "2020-01-07T18:09:00",
          "2020-01-07T19:09:00",
          "2020-01-07T20:09:00",
          "2020-01-07T21:09:00",
          "2020-01-07T22:09:00",
          "2020-01-07T23:09:00",
          "2020-01-08T00:09:00",
          "2020-01-08T01:09:00",
          "2020-01-08T02:09:00",
          "2020-01-08T03:09:00",
          "2020-01-08T04:09:00",
          "2020-01-08T05:09:00",
          "2020-01-08T06:09:00",
          "2020-01-08T07:09:00",
          "2020-01-08T08:09:00",
          "2020-01-08T09:09:00",
          "2020-01-08T10:09:00",
          "2020-01-08T11:09:00",
          "2020-01-08T12:09:00",
          "2020-01-08T13:09:00",
          "2020-01-08T14:09:00",
          "2020-01-08T15:09:00",
          "2020-01-08T16:09:00",
          "2020-01-08T17:10:00",
          "2020-01-08T18:10:00",
          "2020-01-08T19:10:00",
          "2020-01-08T20:10:00",
          "2020-01-08T21:10:00",
          "2020-01-08T22:10:00",
          "2020-01-08T23:10:00",
          "2020-01-09T00:10:00",
          "2020-01-09T01:10:00",
          "2020-01-09T02:10:00",
          "2020-01-09T03:10:00",
          "2020-01-09T04:10:00",
          "2020-01-09T05:10:00",
          "2020-01-09T06:10:00",
          "2020-01-09T07:10:00",
          "2020-01-09T08:10:00",
          "2020-01-09T09:10:00",
          "2020-01-09T10:10:00",
          "2020-01-09T11:10:00",
          "2020-01-09T12:10:00",
          "2020-01-09T13:10:00",
          "2020-01-09T14:10:00",
          "2020-01-09T15:10:00",
          "2020-01-09T16:10:00",
          "2020-01-09T17:11:00",
          "2020-01-09T18:11:00",
          "2020-01-09T19:11:00",
          "2020-01-09T20:11:00",
          "2020-01-09T21:11:00",
          "2020-01-09T22:11:00",
          "2020-01-09T23:11:00",
          "2020-01-10T00:11:00",
          "2020-01-10T01:11:00",
          "2020-01-10T02:11:00",
          "2020-01-10T03:11:00",
          "2020-01-10T04:11:00",
          "2020-01-10T05:11:00",
          "2020-01-10T06:11:00",
          "2020-01-10T07:11:00",
          "2020-01-10T08:11:00",
          "2020-01-10T09:11:00",
          "2020-01-10T10:11:00",
          "2020-01-10T11:11:00",
          "2020-01-10T12:11:00",
          "2020-01-10T13:11:00",
          "2020-01-10T14:11:00",
          "2020-01-10T15:11:00",
          "2020-01-10T16:11:00"
         ]
        }
       ],
       "layout": {
        "template": {
         "data": {
          "bar": [
           {
            "error_x": {
             "color": "#2a3f5f"
            },
            "error_y": {
             "color": "#2a3f5f"
            },
            "marker": {
             "line": {
              "color": "#E5ECF6",
              "width": 0.5
             },
             "pattern": {
              "fillmode": "overlay",
              "size": 10,
              "solidity": 0.2
             }
            },
            "type": "bar"
           }
          ],
          "barpolar": [
           {
            "marker": {
             "line": {
              "color": "#E5ECF6",
              "width": 0.5
             },
             "pattern": {
              "fillmode": "overlay",
              "size": 10,
              "solidity": 0.2
             }
            },
            "type": "barpolar"
           }
          ],
          "carpet": [
           {
            "aaxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "white",
             "linecolor": "white",
             "minorgridcolor": "white",
             "startlinecolor": "#2a3f5f"
            },
            "baxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "white",
             "linecolor": "white",
             "minorgridcolor": "white",
             "startlinecolor": "#2a3f5f"
            },
            "type": "carpet"
           }
          ],
          "choropleth": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "choropleth"
           }
          ],
          "contour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "contour"
           }
          ],
          "contourcarpet": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "contourcarpet"
           }
          ],
          "heatmap": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmap"
           }
          ],
          "heatmapgl": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmapgl"
           }
          ],
          "histogram": [
           {
            "marker": {
             "pattern": {
              "fillmode": "overlay",
              "size": 10,
              "solidity": 0.2
             }
            },
            "type": "histogram"
           }
          ],
          "histogram2d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2d"
           }
          ],
          "histogram2dcontour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2dcontour"
           }
          ],
          "mesh3d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "mesh3d"
           }
          ],
          "parcoords": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "parcoords"
           }
          ],
          "pie": [
           {
            "automargin": true,
            "type": "pie"
           }
          ],
          "scatter": [
           {
            "fillpattern": {
             "fillmode": "overlay",
             "size": 10,
             "solidity": 0.2
            },
            "type": "scatter"
           }
          ],
          "scatter3d": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatter3d"
           }
          ],
          "scattercarpet": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattercarpet"
           }
          ],
          "scattergeo": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergeo"
           }
          ],
          "scattergl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergl"
           }
          ],
          "scattermapbox": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattermapbox"
           }
          ],
          "scatterpolar": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolar"
           }
          ],
          "scatterpolargl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolargl"
           }
          ],
          "scatterternary": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterternary"
           }
          ],
          "surface": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "surface"
           }
          ],
          "table": [
           {
            "cells": {
             "fill": {
              "color": "#EBF0F8"
             },
             "line": {
              "color": "white"
             }
            },
            "header": {
             "fill": {
              "color": "#C8D4E3"
             },
             "line": {
              "color": "white"
             }
            },
            "type": "table"
           }
          ]
         },
         "layout": {
          "annotationdefaults": {
           "arrowcolor": "#2a3f5f",
           "arrowhead": 0,
           "arrowwidth": 1
          },
          "autotypenumbers": "strict",
          "coloraxis": {
           "colorbar": {
            "outlinewidth": 0,
            "ticks": ""
           }
          },
          "colorscale": {
           "diverging": [
            [
             0,
             "#8e0152"
            ],
            [
             0.1,
             "#c51b7d"
            ],
            [
             0.2,
             "#de77ae"
            ],
            [
             0.3,
             "#f1b6da"
            ],
            [
             0.4,
             "#fde0ef"
            ],
            [
             0.5,
             "#f7f7f7"
            ],
            [
             0.6,
             "#e6f5d0"
            ],
            [
             0.7,
             "#b8e186"
            ],
            [
             0.8,
             "#7fbc41"
            ],
            [
             0.9,
             "#4d9221"
            ],
            [
             1,
             "#276419"
            ]
           ],
           "sequential": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ],
           "sequentialminus": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ]
          },
          "colorway": [
           "#636efa",
           "#EF553B",
           "#00cc96",
           "#ab63fa",
           "#FFA15A",
           "#19d3f3",
           "#FF6692",
           "#B6E880",
           "#FF97FF",
           "#FECB52"
          ],
          "font": {
           "color": "#2a3f5f"
          },
          "geo": {
           "bgcolor": "white",
           "lakecolor": "white",
           "landcolor": "#E5ECF6",
           "showlakes": true,
           "showland": true,
           "subunitcolor": "white"
          },
          "hoverlabel": {
           "align": "left"
          },
          "hovermode": "closest",
          "mapbox": {
           "style": "light"
          },
          "paper_bgcolor": "white",
          "plot_bgcolor": "#E5ECF6",
          "polar": {
           "angularaxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           },
           "bgcolor": "#E5ECF6",
           "radialaxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           }
          },
          "scene": {
           "xaxis": {
            "backgroundcolor": "#E5ECF6",
            "gridcolor": "white",
            "gridwidth": 2,
            "linecolor": "white",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "white"
           },
           "yaxis": {
            "backgroundcolor": "#E5ECF6",
            "gridcolor": "white",
            "gridwidth": 2,
            "linecolor": "white",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "white"
           },
           "zaxis": {
            "backgroundcolor": "#E5ECF6",
            "gridcolor": "white",
            "gridwidth": 2,
            "linecolor": "white",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "white"
           }
          },
          "shapedefaults": {
           "line": {
            "color": "#2a3f5f"
           }
          },
          "ternary": {
           "aaxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           },
           "baxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           },
           "bgcolor": "#E5ECF6",
           "caxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           }
          },
          "title": {
           "x": 0.05
          },
          "xaxis": {
           "automargin": true,
           "gridcolor": "white",
           "linecolor": "white",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "white",
           "zerolinewidth": 2
          },
          "yaxis": {
           "automargin": true,
           "gridcolor": "white",
           "linecolor": "white",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "white",
           "zerolinewidth": 2
          }
         }
        }
       }
      }
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "PAIRS_PATH = pathlib.Path.cwd().resolve() / \"eurusd\"\n",
    "MODELS_PATH = pathlib.Path.cwd().resolve() / \"saved_models\"\n",
    "\n",
    "ticks, epochs = csv_to_ticks([f'{PAIRS_PATH}/DAT_ASCII_EURUSD_M1_2020.zip'])\n",
    "window = 60*24*7\n",
    "candle_width = 60\n",
    "candlesticks = ticks_to_candlesticks(ticks[:window], candle_width)\n",
    "\n",
    "datetime = pd.to_datetime(epochs, unit='s')\n",
    "datetime = datetime[range(0, window, candle_width)]\n",
    "\n",
    "\n",
    "fig = go.Figure(data=[go.Candlestick(x=datetime,\n",
    "                open=candlesticks[:, OHLC.OPEN],\n",
    "                high=candlesticks[:, OHLC.HIGH],\n",
    "                low=candlesticks[:, OHLC.LOW],\n",
    "                close=candlesticks[:, OHLC.CLOSE])])\n",
    "\n",
    "fig.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The figure above shows the EURUSD rate between Jan 2 and Jan 10 of 2020. We can observe a gap between Jan 4 to Jan 5, due to FOREX being closed on weekends."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preprocessing\n",
    "In order to work with these data we need to create examples our model can learn from. That can be achieved by iteratively going through a window of past ticks and noting how the price fluctuated in the past. We do not want our samples to contain very exhaustive data (i.e., long noisy sequences) but we do not want to do too much preprocessing of the data either because we want our model to figure out the best features on its own.\n",
    "As a proof of concept, we will be looking at samples with the following parameters:\n",
    "- `N_PAST`: number of past ticks the model uses to make a prediction. In our case we will be looking at sliding windows containing 1440 ticks (i.e.: the last 24h).\n",
    "- `TOLERANCE`: becase the dataset contains missing ticks (not only due to the market being closed), we want to be flexible regarding how much data can be missing. We use a value of 0.9, meaning that we require that 90% of the data be present in the `N_PAST` interval.\n",
    "- `CANDLE_WIDTH`: to understample our data and filter out noise, we look at series of candles of width `CANDLE_WIDTH`. In our case we use candles of width 15 mins.\n",
    "- `HORIZON`: how long in the future do we want to predict the price movement? We will look at the target price after 720 ticks (i.e.: 12h).\n",
    "- `THRESHOLD`: how much does the price need to go up/down with respect to the current price in order for it to be considered a positive example? In our case, we consider it a positive example when the price has moved up (down) by a magnitude of 0.003 (i.e., 30 pips), with respect to the highest (lowest) value of the last tick of the ticks window.\n",
    "\n",
    "In order to make the samples stationary, instead of using the raw OHLC values we will look at the difference of the current tick values with respect to the OPEN price of the next tick. This will reduce the effects of low frequency signals and help keep the inputs within the same range (i.e., using a candle with of 15 mins, most of the times the price will have moved less than 50 pips)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "N_PAST = 60*24     # 1440 ticks\n",
    "HORIZON = 720      # 12 hours\n",
    "TOLERANCE = 0.8    # 80% of of the data must be present\n",
    "CANDLE_WIDTH = 15  # 15 minutes\n",
    "THRESHOLD = 30e-4  # 30 pips\n",
    "\n",
    "train_files = [f'{PAIRS_PATH}/DAT_ASCII_EURUSD_M1_2021.zip']\n",
    "valid_files = [f'{PAIRS_PATH}/DAT_ASCII_EURUSD_M1_2022.zip']\n",
    "test_files  = [f'{PAIRS_PATH}/DAT_ASCII_EURUSD_M1_2023.zip']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1/1 [00:00<00:00,  1.98it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sorting data...\n",
      "Done!\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 367282/367282 [00:03<00:00, 116503.10it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Samples identified: 254592 (68.9126%).\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1/1 [00:00<00:00,  2.15it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sorting data...\n",
      "Done!\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 370645/370645 [00:03<00:00, 119759.91it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Samples identified: 259642 (69.6455%).\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1/1 [00:00<00:00,  2.13it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sorting data...\n",
      "Done!\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 320418/320418 [00:01<00:00, 164082.18it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Samples identified: 143127 (44.3697%).\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "train_dataset = TicksDataset(train_files,\n",
    "                            N_PAST,\n",
    "                            HORIZON,\n",
    "                            TOLERANCE,\n",
    "                            transform=transforms.Compose([\n",
    "                                ToCandlesticks(CANDLE_WIDTH),\n",
    "                                FlipAxes(),\n",
    "                                ToPipDifference()\n",
    "                            ]),\n",
    "                            target_transform=transforms.Compose([\n",
    "                                Threshold(THRESHOLD),\n",
    "                                ToMulticlass()\n",
    "                            ]))\n",
    "valid_dataset = TicksDataset(valid_files,\n",
    "                            N_PAST,\n",
    "                            HORIZON,\n",
    "                            TOLERANCE,\n",
    "                            transform=transforms.Compose([\n",
    "                                ToCandlesticks(CANDLE_WIDTH),\n",
    "                                FlipAxes(),\n",
    "                                ToPipDifference(),\n",
    "                            ]),\n",
    "                            target_transform=transforms.Compose([\n",
    "                                # Threshold(THRESHOLD), # Only interested in direction, so no threshold\n",
    "                                ToMulticlass(),\n",
    "                            ]))\n",
    "test_dataset = TicksDataset(test_files,\n",
    "                            N_PAST,\n",
    "                            HORIZON,\n",
    "                            TOLERANCE,\n",
    "                            transform=transforms.Compose([\n",
    "                                ToCandlesticks(CANDLE_WIDTH),\n",
    "                                FlipAxes(),\n",
    "                                ToPipDifference(),\n",
    "                            ]),\n",
    "                            target_transform=transforms.Compose([\n",
    "                                # Threshold(THRESHOLD), # Only interested in direction, so no threshold\n",
    "                                ToMulticlass()\n",
    "                            ]))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataloader = DataLoader(train_dataset, batch_size=64, shuffle=True, num_workers=3)\n",
    "valid_dataloader = DataLoader(valid_dataset, batch_size=128, shuffle=True, num_workers=5)\n",
    "test_dataloader = DataLoader(test_dataset, batch_size=512, shuffle=False, num_workers=6)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visual inspection\n",
    "\n",
    "Let's take one batch from the training data and see what the distribution of the input values and labels look like:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "x, y = next(iter(train_dataloader))\n",
    "\n",
    "plt.hist(x.numpy().flatten(), bins=200)\n",
    "plt.title('Candlesticks')\n",
    "plt.ylabel('Pip difference')\n",
    "plt.show()\n",
    "\n",
    "plt.hist(y.numpy().flatten(), bins=np.arange(-0.5, 2.5+1, 1),rwidth=0.8)\n",
    "plt.title('Labels')\n",
    "plt.ylabel('Pip difference')\n",
    "plt.xticks([0, 1, 2])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The two first figures represent the distribution of pip differences in the window of past ticks. We can see that the data is concentrated around the 0 (this makes sense, as we market movements are expected to be a (quasi-)random walk). We confirm the observation made earlier that most tick value diffrences fall within the (-50, +50) pips range.\n",
    "\n",
    "The last histogram tells us that the dataset is unbalanced, with positive observations (either UP or DOWN) being about 40% of the total for the chosen threshold of 30 pips."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We train a 1D-CNN using early stopping and reducing the LR after reaching a plateau according to the parameters below. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "STOP_PATIENCE = 200\n",
    "PLATEAU_PATIENCE = STOP_PATIENCE // 4\n",
    "\n",
    "model = ConvNetMultiClass(N_PAST//CANDLE_WIDTH).to(DEVICE)\n",
    "\n",
    "\n",
    "criterion = nn.CrossEntropyLoss(weight=torch.tensor([1, 50, 50], dtype=torch.float32).to(DEVICE))\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=2e-4)\n",
    "scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(optimizer, mode='min', factor=0.25, patience=PLATEAU_PATIENCE, threshold=1e-3, threshold_mode='rel', cooldown=10, min_lr=1e-6, eps=1e-8)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [1/100], Step [25/3978], Train loss: 0.8872, Valid loss: 0.9339\n",
      "Epoch [1/100], Step [50/3978], Train loss: 0.9118, Valid loss: 0.8954\n",
      "Epoch [1/100], Step [75/3978], Train loss: 0.8480, Valid loss: 0.8688\n",
      "Epoch [1/100], Step [100/3978], Train loss: 0.8714, Valid loss: 0.8538\n",
      "Epoch [1/100], Step [125/3978], Train loss: 0.8926, Valid loss: 0.8252\n",
      "Epoch [1/100], Step [150/3978], Train loss: 1.0427, Valid loss: 0.8155\n",
      "Epoch [1/100], Step [175/3978], Train loss: 0.8000, Valid loss: 0.8165\n",
      "Epoch [1/100], Step [200/3978], Train loss: 0.8472, Valid loss: 0.8215\n",
      "Epoch [1/100], Step [225/3978], Train loss: 0.7658, Valid loss: 0.8224\n",
      "Epoch [1/100], Step [250/3978], Train loss: 0.7769, Valid loss: 0.8253\n",
      "Epoch [1/100], Step [275/3978], Train loss: 0.8225, Valid loss: 0.8273\n"
     ]
    }
   ],
   "source": [
    "train_losses = []\n",
    "valid_losses = []\n",
    "\n",
    "# Train the model\n",
    "total_steps = len(train_dataloader)\n",
    "bad_epochs = 0\n",
    "best_loss = np.inf\n",
    "iter_valid_dataloader = iter(valid_dataloader)\n",
    "for epoch in range(100):\n",
    "    for i, (x_batch, y_batch) in enumerate(train_dataloader):\n",
    "        x_batch = x_batch.to(torch.float32).to(DEVICE)\n",
    "        x_batch = x_batch.to(torch.float32).to(DEVICE)\n",
    "        y_batch = y_batch.to(DEVICE).type(torch.int64)\n",
    "        \n",
    "        # Forward pass\n",
    "        model.train()\n",
    "        preds = model(x_batch)\n",
    "        train_loss = criterion(preds, y_batch)\n",
    "        \n",
    "        # Backward and optimize\n",
    "        optimizer.zero_grad()\n",
    "        train_loss.backward()\n",
    "        optimizer.step()\n",
    "        \n",
    "        train_losses.append(train_loss.item())\n",
    "        avg_train_loss = sum(train_losses[-100:])/len(train_losses[-100:])\n",
    "        \n",
    "        model.eval()\n",
    "        with torch.no_grad():\n",
    "            try:\n",
    "                x_batch, y_batch =  next(iter_valid_dataloader)\n",
    "            except StopIteration:\n",
    "                iter_valid_dataloader = iter(valid_dataloader)\n",
    "                x_batch, y_batch =  next(iter_valid_dataloader)\n",
    "            \n",
    "            x_batch = x_batch.to(torch.float32).to(DEVICE)\n",
    "            x_batch = x_batch.to(torch.float32).to(DEVICE)\n",
    "            y_batch = y_batch.to(DEVICE).type(torch.int64)\n",
    "                            \n",
    "            preds = model(x_batch)\n",
    "            \n",
    "            valid_loss = criterion(preds, y_batch)\n",
    "            valid_losses.append(valid_loss.item())\n",
    "            \n",
    "            avg_valid_loss = sum(valid_losses[-100:])/len(valid_losses[-100:])\n",
    "            \n",
    "            scheduler.step(valid_loss)\n",
    "            \n",
    "            if (i+1) % 25 == 0:\n",
    "                print ('Epoch [{}/{}], Step [{}/{}], Train loss: {:.4f}, Valid loss: {:.4f}'\n",
    "                                .format(epoch+1, 100, i+1, total_steps, train_loss.item(), avg_valid_loss))\n",
    "            \n",
    "            \n",
    "            torch.save(model.state_dict(), f\"{MODELS_PATH}/model.pt\")\n",
    "            \n",
    "            if valid_loss < best_loss:\n",
    "                best_loss = valid_loss\n",
    "                bad_epochs = 0\n",
    "                torch.save(model.state_dict(), f\"{MODELS_PATH}/model.pt\")\n",
    "            else:\n",
    "                bad_epochs += 1\n",
    "                # print(f\"Bad epochs: {bad_epochs} · Best loss: {scheduler.best:.4f} · LR: {scheduler.optimizer.param_groups[0]['lr']}\")\n",
    "                if bad_epochs > STOP_PATIENCE:\n",
    "                    break\n",
    "                \n",
    "    else:\n",
    "        continue\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# This is what the training and validation losses look like:\n",
    "plt.plot(train_losses, label='train')\n",
    "plt.plot(valid_losses, label='valid')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<All keys matched successfully>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load the best trained model\n",
    "model = ConvNetMultiClass(N_PAST//CANDLE_WIDTH).to(DEVICE)\n",
    "model.load_state_dict(torch.load(f\"{MODELS_PATH}/model.pt\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Of out 6205 trades, 63.38% were successful (net: 1661 successful trades)\n"
     ]
    }
   ],
   "source": [
    "counts_ = []\n",
    "for x_batch, y_batch in test_dataloader:\n",
    "    model.eval()\n",
    "    with torch.no_grad():\n",
    "        x_batch = x_batch.to(DEVICE)\n",
    "        x_batch = x_batch.to(torch.float32).to(DEVICE)\n",
    "        y_batch = y_batch.to(DEVICE).type(torch.int64)\n",
    "               \n",
    "        preds = model(x_batch)\n",
    "\n",
    "        metrics, counts = metrics_batch(preds, y_batch, 0.6)\n",
    "        \n",
    "        counts_.append(counts[0][1][1]) # TP class 0\n",
    "        counts_.append(-counts[0][0][1]) # FP class 0\n",
    "        counts_.append(counts[1][1][1]) # TP class 1\n",
    "        counts_.append(-counts[1][0][1]) # FP class 1\n",
    "\n",
    "total = sum([abs(c) for c in counts_])\n",
    "pos = sum([c for c in counts_ if c > 0])\n",
    "net = sum(counts_)\n",
    "\n",
    "print('Of out {:d} trades, {:.2%} were successful (net: {:d} successful trades)'.format(int(total), pos/total, int(net)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Conclusion\n",
    "In this notebook we train a machine learning model to predict the direction of the market after a given time horizon based exclusively on a window of past ticks.\n",
    "\n",
    "We show that we can predict new labels with an accuracy above chance (63% accuracy in the test set), a remarkable performance considering that the model takes about just one minute to be trained from scratch on my old GTX 1050 mobile. This suggests that pure technical analysis of historic FOREX data can indeed predict future short-term performance.\n",
    "\n",
    "Now, will you get rich using this model? **Absolutely not**. One of the reasons for that is that this naive model only predicts the market direction and does not take the reward-to-risk ratio into consideration."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "py11",
   "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": "3.11.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}