\uc790\ub3d9\ub9e4\ub9e4 \uc2dc\uc2a4\ud15c \uad6c\ucd95\uc740 \uae08\uc735 \uc2dc\uc7a5\uc5d0\uc11c \uc810\uc810 \ub354 \uc911\uc694\ud574\uc9c0\uace0 \uc788\uc2b5\ub2c8\ub2e4. \uadf8 \uc911\uc5d0\uc11c\ub3c4 \uba38\uc2e0\ub7ec\ub2dd \ubc0f \ub525\ub7ec\ub2dd \uc54c\uace0\ub9ac\uc998\uc740 \ub370\uc774\ud130\uc5d0\uc11c \ud328\ud134\uc744 \ubc1c\uacac\ud558\uace0, \uc774\ub97c \uae30\ubc18\uc73c\ub85c \uc758\uc0ac\uacb0\uc815\uc744 \uc9c0\uc6d0\ud558\ub294 \uac15\ub825\ud55c \ub3c4\uad6c\ub85c \uc790\ub9ac\uc7a1\uace0 \uc788\uc2b5\ub2c8\ub2e4. \uc774\ubc88 \uae00\uc5d0\uc11c\ub294 \ud30c\uc774\ud1a0\uce58 1.4\ub97c \ud65c\uc6a9\ud558\uc5ec \uc54c\uace0\ub9ac\uc998 \ud2b8\ub808\uc774\ub529\uc744 \uad6c\ud604\ud558\ub294 \ubc29\ubc95\uc5d0 \ub300\ud574 \uc0c1\uc138\ud788 \uc124\uba85\ud558\uaca0\uc2b5\ub2c8\ub2e4.<\/p>\n
\uba38\uc2e0\ub7ec\ub2dd\uc740 \ub370\uc774\ud130\ub85c\ubd80\ud130 \ud559\uc2b5\ud558\uc5ec \uc608\uce21 \ub610\ub294 \ubd84\ub958 \uc791\uc5c5\uc744 \uc218\ud589\ud558\ub294 \uae30\uc220\uc785\ub2c8\ub2e4. \ub525\ub7ec\ub2dd\uc740 \uba38\uc2e0\ub7ec\ub2dd\uc758 \ud558\uc704 \uc9d1\ud569\uc73c\ub85c, \uc778\uacf5\uc2e0\uacbd\ub9dd\uc744 \uc774\uc6a9\ud55c \ubcf5\uc7a1\ud55c \ubaa8\ub378\ub9c1\uc774 \uac00\ub2a5\ud569\ub2c8\ub2e4. \ucd5c\uadfc \uba87 \ub144 \ub3d9\uc548 \uae08\uc735 \uc2dc\uc7a5\uc5d0\uc11c \ub525\ub7ec\ub2dd\uc774 \uc8fc\ubaa9\ubc1b\ub294 \uc774\uc720\ub294 \ub300\ub7c9\uc758 \ub370\uc774\ud130\ub97c \ucc98\ub9ac\ud558\uace0, \ube44\uc120\ud615 \uad00\uacc4\ub97c \ud0d0\uc9c0\ud558\ub294 \ub370 \ud6a8\uacfc\uc801\uc774\uae30 \ub54c\ubb38\uc785\ub2c8\ub2e4.<\/p>\n
\uba38\uc2e0\ub7ec\ub2dd\uc740 \uad00\ucc30\ud55c \ub370\uc774\ud130\ub97c \ud1b5\ud574 \ud589\ub3d9\uc744 \ud559\uc2b5\ud558\ub294 \uc2dc\uc2a4\ud15c\uc785\ub2c8\ub2e4. \uc6b0\ub9ac\ub294 \uc8fc\uc5b4\uc9c4 \uc785\ub825(X)\uc640 \ucd9c\ub825(Y) \uac04\uc758 \uad00\uacc4\ub97c \ubaa8\ub378\ub9c1\ud558\uace0, \uc0c8\ub85c\uc6b4 \uc785\ub825\uc5d0 \ub300\ud574 \uc608\uce21\uc744 \uc0dd\uc131\ud569\ub2c8\ub2e4. \uba38\uc2e0\ub7ec\ub2dd \ubaa8\ub378\uc740 \ud06c\uac8c \uc138 \uac00\uc9c0\ub85c \ubd84\ub958\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4:<\/p>\n
\ub525\ub7ec\ub2dd \ubaa8\ub378\uc740 \uce35(Layer)\uc73c\ub85c \uad6c\uc131\ub418\uc5b4 \uc788\uc73c\uba70, \uac01 \uce35\uc740 \uc785\ub825 \ub370\uc774\ud130\ub97c \ubcc0\ud658\ud558\uc5ec \ucd5c\uc885 \ucd9c\ub825\uc744 \uc0dd\uc131\ud569\ub2c8\ub2e4. \uc2ec\uce35 \uc2e0\uacbd\ub9dd(Deep Neural Networks, DNN)\uc740 \uc774\ub7ec\ud55c \uaddc\uce59\uc744 \uae30\ubc18\uc73c\ub85c \ud558\uba70, CNN(Convolutional Neural Networks)\uacfc RNN(Recurrent Neural Networks)\uc640 \uac19\uc740 \ub2e4\uc591\ud55c \uc544\ud0a4\ud14d\ucc98\uac00 \uc874\uc7ac\ud569\ub2c8\ub2e4.<\/p>\n
\ud30c\uc774\ud1a0\uce58\ub294 \ud398\uc774\uc2a4\ubd81\uc774 \uac1c\ubc1c\ud55c \uc624\ud508\uc18c\uc2a4 \uba38\uc2e0\ub7ec\ub2dd \ub77c\uc774\ube0c\ub7ec\ub9ac\ub85c, \ud2b9\ud788 \ub525\ub7ec\ub2dd \ubaa8\ub378\uc744 \ucd5c\uc801\ud654\ud558\uace0 \uad6c\ucd95\ud558\uae30 \uc704\ud574 \uc124\uacc4\ub418\uc5c8\uc2b5\ub2c8\ub2e4. \uc9c1\uad00\uc801\uc778 \ubb38\ubc95\uacfc \ub3d9\uc801 \uacc4\uc0b0 \uadf8\ub798\ud504\ub85c \uc778\ud574 \ub370\uc774\ud130 \uacfc\ud559\uc790\uc640 \uc5f0\uad6c\uc790\ub4e4\uc5d0\uac8c \uc778\uae30\uac00 \ub192\uc2b5\ub2c8\ub2e4.<\/p>\n
\ud30c\uc774\ud1a0\uce58\ub97c \uc124\uce58\ud558\ub824\uba74, \ud30c\uc774\uc36c\uacfc pip\uac00 \ud544\uc694\ud569\ub2c8\ub2e4. \ub2e4\uc74c \uba85\ub839\uc5b4\ub85c \ud30c\uc774\ud1a0\uce58\ub97c \uc124\uce58\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4:<\/p>\n
pip install torch==1.4.0 torchvision==0.5.0<\/code><\/pre>\n\uc774\uc81c \ud30c\uc774\ud1a0\uce58\ub97c \ud65c\uc6a9\ud55c \uba38\uc2e0\ub7ec\ub2dd \ud504\ub85c\uc81d\ud2b8\ub97c \uc2dc\uc791\ud560 \uc900\ube44\uac00 \ub418\uc5c8\uc2b5\ub2c8\ub2e4.<\/p>\n
3. \ub370\uc774\ud130 \uc218\uc9d1 \ubc0f \uc804\ucc98\ub9ac<\/h2>\n
\uc790\ub3d9\ub9e4\ub9e4 \uc2dc\uc2a4\ud15c\uc5d0 \ud544\uc694\ud55c \ub370\uc774\ud130\ub294 \uc5ec\ub7ec \ucd9c\ucc98\uc5d0\uc11c \uc218\uc9d1\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4. Yahoo Finance API, Alpha Vantage, Quandl \ub4f1\uc758 API\ub97c \uc774\uc6a9\ud558\uba74 \uc8fc\uc2dd \ub370\uc774\ud130\ub97c \uc27d\uac8c \uac00\uc838\uc62c \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
3.1 \ub370\uc774\ud130 \uc218\uc9d1<\/h3>\n
\uc608\ub97c \ub4e4\uc5b4, yfinance \ub77c\uc774\ube0c\ub7ec\ub9ac\ub97c \uc0ac\uc6a9\ud558\uc5ec \ud2b9\uc815 \uc8fc\uc2dd\uc758 \ub370\uc774\ud130\ub97c \uc218\uc9d1\ud558\ub294 \ubc29\ubc95\uc740 \ub2e4\uc74c\uacfc \uac19\uc2b5\ub2c8\ub2e4:<\/p>\n
import yfinance as yf\n\nstock_data = yf.download(\"AAPL\", start=\"2020-01-01\", end=\"2023-01-01\")<\/code><\/pre>\n3.2 \ub370\uc774\ud130 \uc804\ucc98\ub9ac<\/h3>\n
\uc218\uc9d1\ud55c \ub370\uc774\ud130\ub294 \uc885\uc885 \uacb0\uce21\uce58\uac00 \ud3ec\ud568\ub418\uc5b4 \uc788\uc73c\ubbc0\ub85c, \uc774\ub97c \ucc98\ub9ac\ud574\uc57c \ud569\ub2c8\ub2e4. Pandas \ub77c\uc774\ube0c\ub7ec\ub9ac\ub97c \ud65c\uc6a9\ud558\uc5ec \ub370\uc774\ud130\ud504\ub808\uc784\uc744 \uc870\uc791\ud558\uace0 \uacb0\uce21\uce58\ub97c \ucc98\ub9ac\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
import pandas as pd\n\nstock_data = stock_data.fillna(method='ffill')<\/code><\/pre>\n4. \ud2b9\uc9d5 \uc120\ud0dd \ubc0f \ubaa8\ub378 \uad6c\ucd95<\/h2>\n
\ud2b9\uc9d5 \uc120\ud0dd\uc740 \uba38\uc2e0\ub7ec\ub2dd \ubaa8\ub378\uc758 \uc131\ub2a5\uc5d0 \ud070 \uc601\ud5a5\uc744 \ubbf8\uce69\ub2c8\ub2e4. \uc6b0\ub9ac\ub294 \uc8fc\uac00 \uc608\uce21\uc744 \uc704\ud574 \uc5ec\ub7ec \uae08\uc735 \uc9c0\ud45c\uc640 \uae30\uc220\uc801 \uc9c0\ud45c\ub97c \ud2b9\uc9d5\uc73c\ub85c \uc0ac\uc6a9\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
4.1 \ud2b9\uc9d5 \uc0dd\uc131<\/h3>\n
\uac04\ub2e8\ud55c \uc774\ub3d9 \ud3c9\uade0(Simple Moving Average, SMA)\uacfc \uc0c1\ub300 \uac15\ub3c4 \uc9c0\uc218(Relative Strength Index, RSI)\uc640 \uac19\uc740 \uae30\uc220\uc801 \uc9c0\ud45c\ub97c \uc0dd\uc131\ud558\uc5ec \ubaa8\ub378\uc774 \ud559\uc2b5\ud560 \uc218 \uc788\ub294 \uc720\uc6a9\ud55c \uc815\ubcf4\ub97c \uc81c\uacf5\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
stock_data['SMA'] = stock_data['Close'].rolling(window=20).mean()\nstock_data['RSI'] = compute_rsi(stock_data['Close'], window=14)<\/code><\/pre>\n4.2 \ub370\uc774\ud130\uc14b \ubd84\ud560<\/h3>\n
\ub370\uc774\ud130\ub97c \ud6c8\ub828 \uc138\ud2b8\uc640 \ud14c\uc2a4\ud2b8 \uc138\ud2b8\ub85c \ub098\ub214\uc73c\ub85c\uc368 \ubaa8\ub378\uc758 \uc77c\ubc18\ud654 \uc131\ub2a5\uc744 \ud3c9\uac00\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \uc77c\ubc18\uc801\uc73c\ub85c 80-20 \ube44\uc728\uc744 \uc0ac\uc6a9\ud569\ub2c8\ub2e4.<\/p>\n
from sklearn.model_selection import train_test_split\n\ntrain_data, test_data = train_test_split(stock_data, test_size=0.2, random_state=42)<\/code><\/pre>\n5. \ubaa8\ub378 \ud559\uc2b5<\/h2>\n
\uc774\uc81c \ubaa8\ub378\uc744 \uad6c\ucd95\ud558\uace0 \ud559\uc2b5\ud560 \ucc28\ub840\uc785\ub2c8\ub2e4. \ud30c\uc774\ud1a0\uce58\ub97c \uc0ac\uc6a9\ud558\uc5ec \uc778\uacf5\uc2e0\uacbd\ub9dd\uc744 \uad6c\ucd95\ud558\ub294 \ubc29\ubc95\uc5d0 \ub300\ud574 \uc124\uba85\ud558\uaca0\uc2b5\ub2c8\ub2e4.<\/p>\n
5.1 \ubaa8\ub378 \uad6c\uc131<\/h3>\n
\ud30c\uc774\ud1a0\uce58\uc5d0\uc11c\ub294 nn.Module\uc744 \uc0c1\uc18d\ud558\uc5ec \uc2e0\uacbd\ub9dd \ubaa8\ub378\uc744 \uc815\uc758\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \uc544\ub798\ub294 \uac04\ub2e8\ud55c \ud53c\ub4dc\ud3ec\uc6cc\ub4dc \uc2e0\uacbd\ub9dd\uc758 \uc608\uc785\ub2c8\ub2e4:<\/p>\n
import torch\nimport torch.nn as nn\n\nclass SimpleNN(nn.Module):\n def __init__(self, input_size, hidden_size, output_size):\n super(SimpleNN, self).__init__()\n self.fc1 = nn.Linear(input_size, hidden_size)\n self.fc2 = nn.Linear(hidden_size, output_size)\n self.relu = nn.ReLU()\n\n def forward(self, x):\n x = self.fc1(x)\n x = self.relu(x)\n x = self.fc2(x)\n return x\n\nmodel = SimpleNN(input_size=3, hidden_size=10, output_size=1)<\/code><\/pre>\n5.2 \ubaa8\ub378 \ud6c8\ub828<\/h3>\n
\ubaa8\ub378\uc744 \ud6c8\ub828\ud558\uae30 \uc704\ud574 \uc190\uc2e4 \ud568\uc218\uc640 \ucd5c\uc801\ud654 \uc54c\uace0\ub9ac\uc998\uc744 \uc815\uc758\ud574\uc57c \ud569\ub2c8\ub2e4. \uc544\ub798\ub294 \uc190\uc2e4 \ud568\uc218\ub85c MSE(Mean Squared Error)\uc640 Adam \ucd5c\uc801\ud654 \uc54c\uace0\ub9ac\uc998\uc744 \uc0ac\uc6a9\ud558\ub294 \uc608\uc785\ub2c8\ub2e4:<\/p>\n
criterion = nn.MSELoss()\noptimizer = torch.optim.Adam(model.parameters(), lr=0.001)\n\nfor epoch in range(100):\n model.train()\n inputs = torch.from_numpy(train_data[['SMA', 'RSI']].values).float()\n labels = torch.from_numpy(train_data['Close'].values).float().view(-1, 1)\n\n optimizer.zero_grad()\n outputs = model(inputs)\n loss = criterion(outputs, labels)\n loss.backward()\n optimizer.step()<\/code><\/pre>\n6. \ubaa8\ub378 \ud3c9\uac00 \ubc0f \uc608\uce21<\/h2>\n
\ubaa8\ub378\uc774 \ud6c8\ub828\ub41c \ud6c4, \ud14c\uc2a4\ud2b8 \ub370\uc774\ud130\uc14b\uc744 \ud65c\uc6a9\ud558\uc5ec \uc131\ub2a5\uc744 \ud3c9\uac00\ud569\ub2c8\ub2e4. \uc8fc\uac00 \uc608\uce21 \ubaa8\ub378\uc758 \uc815\ud655\ub3c4\ub97c \ud655\uc778\ud558\uae30 \uc704\ud574 RMSE(Root Mean Squared Error)\ub97c \uc0ac\uc6a9\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
model.eval()\nwith torch.no_grad():\n test_inputs = torch.from_numpy(test_data[['SMA', 'RSI']].values).float()\n predictions = model(test_inputs).numpy()\n rmse = np.sqrt(np.mean((predictions - test_data['Close'].values) ** 2))\n print(f'RMSE: {rmse}') <\/code><\/pre>\n7. \uc2e4\uc81c \ud2b8\ub808\uc774\ub529 \uc2dc\uc2a4\ud15c \uad6c\ucd95<\/h2>\n
\ubaa8\ub378\uc744 \uae30\ubc18\uc73c\ub85c \ud55c \uc2e4\uc81c \ud2b8\ub808\uc774\ub529 \uc2dc\uc2a4\ud15c \uad6c\ucd95 \uacfc\uc815\uc5d0\ub294 \uc2e0\ud638 \uc0dd\uc131, \uc8fc\ubb38 \uc2e4\ud589 \ubc0f \ud3ec\ud2b8\ud3f4\ub9ac\uc624 \uad00\ub9ac \ub4f1\uc774 \ud3ec\ud568\ub429\ub2c8\ub2e4.<\/p>\n
7.1 \uc2e0\ud638 \uc0dd\uc131<\/h3>\n
\ud6c8\ub828\ub41c \ubaa8\ub378\uc744 \uc774\uc6a9\ud558\uc5ec \ub9e4\ub9e4 \uc2e0\ud638\ub97c \uc0dd\uc131\ud558\ub294 \ubc29\ubc95\uc740 \ub2e4\uc74c\uacfc \uac19\uc2b5\ub2c8\ub2e4:<\/p>\n
signal = np.where(predictions > test_data['Close'].values, 1, 0)<\/code><\/pre>\n7.2 \uc8fc\ubb38 \uc2e4\ud589<\/h3>\n
\uc2e0\ud638\uc5d0 \ub530\ub77c \uc2e4\uc81c \uc8fc\ubb38\uc744 \uc2e4\ud589\ud558\ub824\uba74 API\uac00 \ud544\uc694\ud569\ub2c8\ub2e4. \uc608\ub97c \ub4e4\uc5b4, Alpaca Trading API\ub97c \uc0ac\uc6a9\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
import alpaca_trade_api as tradeapi\n\napi = tradeapi.REST('APCAAPI-KEY-ID', 'APCAAPI-SECRET-KEY', base_url='https:\/\/paper-api.alpaca.markets')\nfor i in range(len(signal)):\n if signal[i] == 1:\n api.submit_order(\n symbol='AAPL',\n qty=1,\n side='buy',\n type='market',\n time_in_force='GTC'\n )<\/code><\/pre>\n8. \uacb0\ub860<\/h2>\n
\uc774\ubc88 \uac15\uc88c\uc5d0\uc11c\ub294 \ud30c\uc774\ud1a0\uce58 1.4\ub97c \ud65c\uc6a9\ud574 \uba38\uc2e0\ub7ec\ub2dd \ubc0f \ub525\ub7ec\ub2dd \uae30\ubc18\uc758 \uc790\ub3d9\ub9e4\ub9e4 \uc2dc\uc2a4\ud15c \uad6c\ucd95 \ubc29\ubc95\uc5d0 \ub300\ud574 \uc0b4\ud3b4\ubcf4\uc558\uc2b5\ub2c8\ub2e4. \ub370\uc774\ud130 \uc218\uc9d1\uc5d0\uc11c \ubaa8\ub378 \ud6c8\ub828, \uc608\uce21, \uadf8\ub9ac\uace0 \uc2e4\uc804\uc5d0\uc11c\uc758 \uc8fc\ubb38 \uc2e4\ud589\uacfc \ud3ec\ud2b8\ud3f4\ub9ac\uc624 \uad00\ub9ac\uae4c\uc9c0 \ub2e4\uc591\ud55c \ub2e8\uacc4\ub97c \ub2e4\ub918\uc2b5\ub2c8\ub2e4. \uba38\uc2e0\ub7ec\ub2dd \ubc0f \ub525\ub7ec\ub2dd \uae30\uc220\uc740 \uc55e\uc73c\ub85c\ub3c4 \uacc4\uc18d \ubc1c\uc804\ud574 \ub098\uac08 \uac83\uc774\uba70, \uae08\uc735\uc2dc\uc7a5\uc5d0\uc11c\ub294 \ub354\uc6b1\ub354 \uc911\uc694\ud55c \uc5ed\ud560\uc744 \ud560 \uac83\uc785\ub2c8\ub2e4.<\/p>\n
\ud5a5\ud6c4 \ub354 \ubc1c\uc804\ub41c \ubaa8\ub378 \uad6c\ucd95\uc744 \uc704\ud574 \ub2e4\uc591\ud55c \uc544\ud0a4\ud14d\ucc98\uc640 \uc54c\uace0\ub9ac\uc998\uc744 \uc2e4\ud5d8\ud574\ubcf4\ub294 \uac83\uc744 \ucd94\ucc9c\ud569\ub2c8\ub2e4. AI \uae30\ubc18\uc758 \ud22c\uc790 \uc804\ub7b5\uc740 \uc2dc\uac04\uc774 \uc9c0\ub0a8\uc5d0 \ub530\ub77c \uc810\ucc28 \uac15\ub825\ud574\uc9c8 \uac83\uc774\uba70, \uc774\uc5d0 \ubc1c\ub9de\ucd94\uc5b4 \uc5ec\ub7ec\ubd84\uc758 \uc790\ub3d9\ub9e4\ub9e4 \uc804\ub7b5\ub3c4 \uc9c0\uc18d\uc801\uc73c\ub85c \uac1c\uc120\ud574 \ub098\uac00\uc2dc\uae30 \ubc14\ub78d\ub2c8\ub2e4.<\/p>\n
\ucc38\uace0\ubb38\ud5cc<\/h2>\n