Quantitative Research and Development Backtesting Framework

A Python-based signal generation and analysis framework to create and test quantitative trading strategies based on public time-series data of financial assets.

Overview Mathematical Framework GitHub Python Package Changelog

Current Status

Overview

Latest Version:Version Equity Curves
Current winning strategy:

Description: Preset AC/correlation/volatility regime filter based implementation of cross asset mv, cross asset momentum, time series momentum + Adam optimization of strategy, regime filter parameters

Results:

          Avg return  Avg Sharpe  Avg Annualized Alpha  Avg DD   Passed
          19.76%      1.720       0.168                 -4.03%    4/5

Implemented Strategies

Cointegration
Mean reversion Time based
Mean reversion Cross sectional
Momentum trending Time based
Momentum trending Cross sectional

Implemented Approaches

Neutrality filters

Beta neutrality
Dollar neutrality
PC neutrality

Regime based strategy execution

Autocorrelation, dispersion, average variance, average pair wise correlation, liquidity, trend

Factor models
Learning based approaches

Logistic regression, elastic net, neural networks, transformers, local kernel methods, SVM, trees, forests

Assets/Data

460 elements time series data from the SP at daily frequency.

Metrics

Sharpe over time period
Alpha over time period
Total return over time period
Max drawdown over time period

Methodology/Model constraints

Training, validation, held out periods vary depending on setup
Rolling window runs
Transaction costs, slippage, risk-free rate vary depending on setup