HFT-Backtesting

This repository contains implementation of a framework for hft (tick-data) backtesting.
You may find a lot of examples in notebooks and tests, it will be described later.

Data Source

Backtesting operates on top of historical data of specific format.
Example of such data can be found in tests/resources/orderbook/.
If you wish to collect own data, please refer to hft.dataloader.loader.py

• This module allows to download and store data in Clickhouse db
• Required data format is simple CSV extracted from database (more details in fields initialization)

Example

In order to run a smallest example, you must initialize reader, strategy, backtester, and your metrics.
output is optional, it allows to store results during backtest evaluation and visualize its performance.

reader = TimeLimitedReader('orderbook_10_03_20.csv.gz', 
                         limit_time='30 min',
                         trades_file='trades_10_03_20.csv.gz',
                         nrows=500000)

delay = 1000 # millisec
target_contracts = 3000
filter_depth=4

strategy = GatlingMM(target_contracts, filter_depth=filter_depth)
backtester = backtest.Backtest(reader, strategy, delay=delay)