xtraderbacktest

This project is a trading system for trading algo. It includes backtest, market scan and live trading features. Star it if you like and keep tracking the updates.

Requirements

python 3.6.8 in 64bit

The tools might need: Microsoft Visual C++ Build Tools(https://visualstudio.microsoft.com/visual-cpp-build-tools/)

Features

• Data feeds from local files.

- Safe fake ticks generate - Real ticks supported. Multilevel price and volume.

• Multiple instruments supported, stocks,cfd,index,commodity, cryto and custom symbols.
• Custom market scanner
• Reverse logic by one click.
• Custom trailing stop loss.
• Multiple take profit and stop loss.
• Upload backtest result to remote storage.
• Two level caching.
• Backtest evaluation.
• Tick driven backtest.
• Gap detection and avoid invaid orders.
• Slippage, commission, swaps simulation.
• Parameters Optimization.
• Custom untradable periods.
• Tablib supported.
• Multiple and custom timeframe supported(seconds,mins,days).
• Absolute no future data usage.
• Live data feed and trading with.

- Alpaca broker support(US Stocks and ETFs).

• Notification in live trading.

- Telegram - Mail

• Live order record.

- MongoDB. - Mysql(ongoing).

• Draw custom chart for analysis usage.

- Line Chart - Bar Chart(ongoing) - Line Segment(ongoing)

• Market Event supported.

- Calendar Event - News Event(ongoing)

• Distribute nodes for optimizing parameters.(ongoing)
• Indicator calculate accelerate.(ongoing)
• Tools

- Analysers. - Web crawler for calendar events. - ...

How to Start

Quick Start

Here is how to run the demo strategy.

• Copy /configuration/sys/systemconftemplate.yaml to /configurations/sys/system_conf.yaml
• Modify the content of /configuration/sys/system_conf.yaml as you need
• pip3 install -r requirement.txt
• Modify the demo strategy's configuration in /configurations/strategy/single/demo/AAPL_demo.json
• Cd /bots
• python demo_strategy.py

Basic Start

Step 1

Put the 1min price data in the folder /data/price/SYMBOL_NAME.csv. The 1min price data should be in this format: date, open,high,low,close,volume,open_interest(optional)

2019-02-15 04:40:00,169.31,169.31,169.31,169.31,100
2019-02-15 04:41:00,169.31,169.31,169.31,169.31,300

Optional: If you want to use real ticks, please place the real ticks data into /data/ticks The file name should be in this format: (symbol).csv The data should be in this format in every line:

symbol,date,lastprice,openinterest,volume,ask1,ask1volume,ask2,ask2volume,ask3,ask3volume,ask4,ask4volume,ask5,ask5volume,bid1,bid1volume,bid2,bid2volume,bid3,bid3volume,bid4,bid4volume,bid5,bid5volume,is_gap

Note: 'is_gap' should always be true when it is true tick, which means there is gap between this tick and last one.

Step 2

Save the symbols' configurations in the folder /configurations/symbols_conf/ Here is the stock AAPL template.

---
name: AAPL                              # symbol name
point: 0.01                             # point value
ticksize: 0.01                         # ticksize
commission: 0.1                         # commission in currency
slippage: 0                             # slippage in points
margin_rate: 1                          # margin rate 0.05 = 5%
minimumtpsl: 3                        # minimum gap in points between entry price and tpsl price
contract_size: 1                        # contract size
spread: 5                               # the spread(between ask and bid) in points for backtest
type: stock                             # symbol type
exchange: None                          # the trading exchange
t+0: true                               # whether is t+0
swaps:                                  # the swaps in points
  long: -2
  short: -3
  triple_day: 0                         # 0-Monday 6-Sunday
minimum_lots: 1                         # minimum lot
trade_session:                          # tradable session
  sunday: []
  monday:                               # the template of multiple tradable session in one day
  - - '01:00:00'
    - '10:59:59'
  - - '10:59:59'
    - '23:59:59'
  tuesday:
  - - '01:00:00'
    - '23:59:59'
  wednesday:
  - - '01:00:00'
    - '23:59:59'
  thursday:
  - - '01:00:00'
    - '23:59:59'
  friday:
  - - '01:00:00'
    - '23:59:59'
  saturday: []

Step 3

Put the strategy's parameters in the folder /configurations/strategy/single/(strategyname)/(symbolname).json in json format. The template is as below.

{
    "accountid":"demoaccount",
    "period":["10m"],
    "backtest_graininess":"5m",
    "symbols":["AAPL"],
    "platform":"IB",
    "start_date": "2019-10-25 08:47:00",
    "end_date": "2019-10-29 19:59:00",
    "strategynamecode": "DM",
    "strategy_name": "demo",
    "reverse_mode":"enable",
    "calendar_event":"enable",
    "cash":10000,
    "untradable_period":[
        {
            "start":"23:59:59",
            "end":"23:59:59"
        }
    ],
    "tag":"demo",
    "custom":{
        "ma_fast":10,
        "ma_slow":21,
        "lots":1
    }
}

The parameters that the strategy used is under the key custom, while the other keys are compulsory.

Step 4

Write your own strategy and put it in the folder /bots/. Here is a double ma strategy as demo, which buy stocks when fast ma > slow ma and vice versa.

import os
import sys
sys.path.append(os.path.join(os.getcwd().split('xtraderbacktest')[0],'xtraderbacktest'))
import modules.other.logg
import logging 
import modules.common.strategy
import modules.other.sysconfloader as sysconfloader
import modules.common.technical_indicators as ti
class Bot(modules.common.strategy.Strategy):
    def init(self,pars):
        super(Bot,self).init(pars)
# Handle Tick     def handle_tick(self, tick):                  pass
# Handle Bar     def handle_bar(self, bar,period):         #logging.info("new bar "+bar["date"])         #logging.info("currenttime " + self.currenttime)         df = self.get_bars(bar["symbol"],30,period)         mafast = ti.MA(df,self.pars["mafast"]).iloc[-1]         maslow = ti.MA(df,self.pars["maslow"]).iloc[-1]         if mafast > maslow:             if len(self.getcurrentposition(direction="long")) ==0:                 self.open_order(bar["symbol"],"market",self.pars["lots"],"long")             self.closeallposition(direction="short")         elif mafast < maslow:             if len(self.getcurrentposition(direction="short")) ==0:                 self.open_order(bar["symbol"],"market",self.pars["lots"],"short")             self.closeallposition(direction="long")         pass
if name == "main":     pars = sysconfloader.readconfigsjson("AAPLdemo.json","/configurations/strategy/single/demostrategy/")     backtest = Bot(pars)     import modules.backtest.scheduler      scheduler = modules.backtest.scheduler.Scheduler("backtest")     scheduler.register_strategy(backtest)     scheduler.start()

Then run this (strategy).py in the folder /bots/. And you can find the backtest result in the folder /data/backtest_results/ after finish backtesting.

Step 5

Analyse the backtest result. Open /modules/backtest/evaluation/evaluation_dashboard.html and choose the backtest result as the file to open, after that the result should be shown on page.

Template Result Template Result of multiple Symbols(AMD + AAPL)

Scanner Start

Important note

• Scanner can operate all the actions which normal backtest can, which means you can place order and get the backtest result.
• The data in price folder should be matched with symbolsconf and symbolsreport configurations as well.
• The symbols_rules function is necessary to be implemented in custom scanner as the system would use it filter the symbols first.
• The pricedatamode in /configuration/sys/system_conf.yaml should be turn to "disk" if you are scanning extrem large amount of symbols. Otherwise it might blow up the memory.
• The scanner result would be stored in /data/scanner_results.

The template of scanner is shown in /scanners/demo_scanner.py

Live Strategy(Alpaca)

Step 1

Fill the redis and alpaca configurations correctly.

Step 2

cd ./modules/brokers/alpaca/
chmod 777 run_api.sh
./run_api.sh

Step 3

Run the bot

cd ./bots/
python demostrategylive.py

Existing Problems

• Takes too much time in generating fake ticks.
• Fake ticks take too much memory.