Heads up: This project is dead. It fought well, but died on the battlefield of abandoned open source projects. It was a good project. It will be remembered for decades to come.

An event-driven backtesting framework to test stock trading strategies based on fundamental analysis. Preferably this package will be the core of a backend service exposed via a REST API. Currently it has only core functionality.

Basic example:

package main

import (
  "github.com/dirkolbrich/gobacktest"
  "github.com/dirkolbrich/gobacktest/data"
  "github.com/dirkolbrich/gobacktest/strategy"
)

func main() {
  // initiate a new backtester
  test := gobacktest.New()

  // define and load symbols
  symbols := []string{"TEST.DE"}
  test.SetSymbols(symbols)

  // create a data provider and load the data into the backtest
  data := &data.BarEventFromCSVFile{FileDir: "../testdata/test/"}
  data.Load(symbols)
  test.SetData(data)

  // choose a strategy
  strategy := strategy.BuyAndHold()

  // create an asset and append it to the strategy
  strategy.SetChildren(gobacktest.NewAsset("TEST.DE"))
  
  // load the strategy into the backtest
  test.SetStrategy(strategy)

  // run the backtest
  test.Run()

  // print the results of the test
  test.Stats().PrintResult()
}

More example tests are in the /examples folder.

The single parts of the backtester can be set independently:

// initiate new backtester
test := &Backtest{}

// set the portfolio with initial cash and a default size and risk manager
portfolio := &gobacktest.Portfolio{}
portfolio.SetInitialCash(10000)

sizeManager := &gobacktest.Size{DefaultSize: 100, DefaultValue: 1000}
portfolio.SetSizeManager(sizeManager)

riskManager := &gobacktest.Risk{}
portfolio.SetRiskManager(riskManager)

test.SetPortfolio(portfolio)

// create a new strategy with an algo stack
strategy := gobacktest.NewStrategy("basic")
strategy.SetAlgo(
    algo.CreateSignal("buy"), // always create a buy signal on a data event
)

// create an asset and append to strategy
strategy.SetChildren(gobacktest.NewAsset("TEST.DE"))

// load the strategy into the backtest
test.SetStrategy(strategy)

// create an execution provider and load it into the backtest
exchange := &gobacktest.Exchange{
    Symbol:      "TEST",
    Commission:  &FixedCommission{Commission: 0},
    ExchangeFee: &FixedExchangeFee{ExchangeFee: 0},
}
test.SetExchange(exchange)

// choose a statistic and load into it the backtest
statistic := &gobacktest.Statistic{}
test.SetStatistic(statistic)

None so far. Only the standard library.

These are the basic components of an event-driven framework.

1. BackTester - general test case, bundles the following elements into a single test
2. EventHandler - the different types of events, which travel through this system - data event, signal event, order event and fill event
3. DataHandler - interface to a set of data, e.g historical quotes, fundamental data, dividends etc.
4. StrategyHandler - generates a buy/sell signal based on the data
5. PortfolioHandler - generates orders and manages profit & loss
   • (SizeHandler) - manages the size of an order
   • (RiskHandler) - manages the risk allocation of a portfolio
6. ExecutionHandler - sends orders to the broker and receives the “fills” or signals that the stock has been bought or sold
7. StatisticHandler - tracks all events during the backtests and calculates useful statistics like equity return, drawdown or sharp ratio etc., could be used to replay the complete backtest for later reference
   • (ComplianceHandler) - tracks and documents all trades to the portfolio for compliance reasons

An overviev of the infrastructure of a complete backtesting and trading environment. Taken from the production roadmap of QuantRocket.

• General
  • API gateway
  • configuration loader
  • logging service
  • cron service
• Data
  • database backup and download service
  • securities master services
  • historical market data service
  • fundamental data service
  • earnings data service
  • dividend data service
  • real-time market data service
  • exchange calendar service
• Strategy
  • performance analysis service - tearsheet
• Portfolio
  • account and portfolio service
  • risk management service
• Execution
  • trading platform gateway service
  • order management and trade ledger service
  • backtesting and trading engine

These links to articles are a good starting point to understand the intentions and basic functions of an event-driven backtesting framework.

• Initial idea via a blog post Python For Finance: Algorithmic Trading by Karlijn Willems @willems_karlijn.
• Very good explanation of the internals of a backtesting system by Michael Halls-Moore @mhallsmoore in the blog post series Event-Driven-Backtesting-with-Python.

• QuantConnect
• Quantopian
• QuantRocket - in development, available Q2/2018
• Quandl - financial data
• QSTrader - open-source backtesting framework from QuantStart
• bt - Flexible Backtesting for Python - an inspiration for algorithm building blocks and a strategy/assets tree

• Quantocracy - forum for quant news
• QuantStart - articels and tutorials about quant finance