Backtra is an advanced futures backtesting framework where we loop through kline (ohlcv) data on a specific way to acheive more realistic results.
This is a futures backtesting framework but it can also be used for spot backtesting.
More in-depth documentation is coming soon!
git clone https://github.com/AlgoQ/backtra
cd backtra
pip install -r requirements
We will be backtesting a simple golden cross strategy with a stoploss and take profit based on the ATR. The leverage is also based on the ATR and the max percentage you are willing to lose per trade.
:max_bytes(150000):strip_icc():format(webp)/GoldenCross-5c6592b646e0fb0001a91e29.png)
- Config
Rename the exampleConfig.json file to config.json and update the parameters if wanted.
{
"capital": 10000,
"makerFee": -0.0025,
"takerFee": 0.0075,
"reduceAmount": 0
}- Strategy
Create a strategy named GoldenCross.py and add it into the strategies folder.
More info about creating a strategy will be added soon.
from strategies.BaseStrategy import BaseStrategy
import numpy as np
import pandas as pd
# TODO: Import needed indicators
from talib import SMA
class GoldenCross(BaseStrategy):
def __init__(self, strategyName, symbol, params, timeFrames, ohlcvs, tradeLogs=True):
# Changeable values
self.strategyName = strategyName
self.symbol = symbol
self.params = params
self.timeFrames = timeFrames
self.ohlcvs = ohlcvs
self.tradeLogs = tradeLogs
# Fixed values
self.configData = self._getConfigData()
self.capital = self.configData['capital']
self.makerFee = self.configData['makerFee']
self.takerFee = self.configData['takerFee']
self.reduceAmount = self.configData['reduceAmount']
self.capitalFollowup = [[self.ohlcvs[self.timeFrames[0]].index[1], self.capital]]
self.openTradesL = {}
self.closedTradesL = []
def run(self):
print('Trading started...')
minKlines = max(201, self._calcMinKlines())
for i in range(minKlines, len(self.ohlcvs[self.timeFrames[0]])):
tempDf = self.ohlcvs[self.timeFrames[0]].iloc[i-minKlines:i+1].copy()
# Indicators
sma1 = SMA(tempDf['close'], self.params['sma1'])
sma2 = SMA(tempDf['close'], self.params['sma2'])
if sma1[-2] < sma2[-2]: # Long condition 1
prevClose = tempDf['close'][-1]
tempDf.loc[tempDf.index[-1], 'close'] = tempDf['high'][-1]
sma1 = SMA(tempDf['close'], self.params['sma1'])
sma2 = SMA(tempDf['close'], self.params['sma2'])
if sma1[-1] > sma2[-1]: # Long condition 2
tempDf.loc[tempDf.index[-1], 'close'] = prevClose
for i in np.arange(tempDf['open'][-1], tempDf['high'][-1] + self.ohlcvs['pip'], self.ohlcvs['pip']):
openPrice = round(i, self.ohlcvs['precision'])
tempDf.loc[tempDf.index[-1], 'close'] = openPrice
# Indicators
sma1 = SMA(tempDf['close'], self.params['sma1'])
sma2 = SMA(tempDf['close'], self.params['sma2'])
if sma1[-1] > sma2[-1]: # Long condition 2
if len(self.openTradesL) != 0:
self.closeTrade(
id = '1',
time = tempDf.index[-1],
tradeType = 'market',
closePrice = openPrice)
self.openTrade(
id = '1',
time = tempDf.index[-1],
side = 'long',
tradeType = 'market',
leverage = self.params['leverage'],
amount = self.capital,
openPrice = openPrice)
break
elif sma1[-2] > sma2[-2]: # Short condition 1
prevClose = tempDf['close'][-1]
tempDf.loc[tempDf.index[-1], 'close'] = tempDf['low'][-1]
sma1 = SMA(tempDf['close'], self.params['sma1'])
sma2 = SMA(tempDf['close'], self.params['sma2'])
if sma1[-1] < sma2[-1]: # Short condition 2
tempDf.loc[tempDf.index[-1], 'close'] = prevClose
for i in np.arange(tempDf['open'][-1], tempDf['low'][-1] - self.ohlcvs['pip'], self.ohlcvs['pip'] * -1):
openPrice = round(i, self.ohlcvs['precision'])
tempDf.loc[tempDf.index[-1], 'close'] = openPrice
# Indicators
sma1 = SMA(tempDf['close'], self.params['sma1'])
sma2 = SMA(tempDf['close'], self.params['sma2'])
if sma1[-1] < sma2[-1]: # Short condition 2
if len(self.openTradesL) != 0:
self.closeTrade(
id='1',
time = tempDf.index[-1],
tradeType = 'market',
closePrice = openPrice)
self.openTrade(
id = '1',
time = tempDf.index[-1],
side = 'short',
tradeType = 'market',
leverage = self.params['leverage'],
amount = self.capital,
openPrice = openPrice)
break- Main File
Now we will create a main file that connects/runs our strategy.
I this file we send all variable values through the params, select timeframe and link your data.
Create mainGoldenCross.py in the main folder:
from strategies.GoldenCross import GoldenCross
from utils import jsonToOhlcv
import quantstats as qs
params = {'sma1': 50, 'sma2': 200, 'leverage': 1}
timeframe = '4h'
ohlcv = jsonToOhlcv(r'/media/kobe/D/feda/data/ohlcv_ftx_BTCUSD_784days.json', timeframe)
# precision = round(ohlcv['close'].apply(lambda x: len(str(x).split('.')[-1])).mean())
precision = 0
if precision > 0:
pip = float('0.' + ('0' * (precision - 1)) + '1')
else:
pip = 1
ohlcvs = {}
ohlcvs['precision'] = precision
ohlcvs['pip'] = pip
ohlcvs[timeframe] = ohlcv
goldenCross = GoldenCross(
strategyName = 'Golden Cross',
symbol = 'BTC/USD',
params = params,
ohlcvs = ohlcvs,
timeFrames = [timeframe],
tradeLogs = True
)
goldenCross.run()
results, percChange = goldenCross.calcResults()
goldenCross.showResults(results)
qs.plots.snapshot(percChange, title=f'Results', savefig=f'Results')- Output
Strategy Golden Cross
Symbol BTC/USD
Timeframes ['4h']
Parameters {'sma1': 50, 'sma2': 200, 'leverage': 1}
Start 2019-08-20 16:00:00
End 2021-10-12 12:00:00
Duration (days) 783
Equity Start [$] 10000
Equity Final [$] 58406.2285
Equity Max [$] 89791.4633
Return [%] 484.06
Max. Drawdown [%] -39.86
Win rate [%] 54.17
Buy & Hold [%] 424.69
Total trades 24
Avg. trade [%] 12.52
Avg. winning trade [%] 31.36
Avg. losing trade [%] -9.74
Avg. long trade [%] 24.64
Avg. short trade [%] 0.4
Snapshot:
If you don't have any ohlcv/kline data you can always fetch crypto kline data with FEDA.
- More in-depth documentation
- Slippage parameter in the config
- Optimization
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent