Introducing GA-MSSR for RoboTrading for after-work quants

Takeaway. GA-MSSR does not train a net on raw ticks. It evolves the parameters of technical-rule features so a rule-based FX bot maximises a Sharpe + Sterling (drawdown-aware) fitness—noise-robust search, not a crystal ball.

GA-MSSR: Genetic Algorithm Maximizing Sharpe and Sterling Ratio Method for RoboTrading (Zezheng Zhang, Matloob Khushi, 2020; arXiv:2008.09471 / IJCNN 2020).

Fit for learners: Genetic search of FX rule parameters to maximise a Sharpe–Sterling blend—risk-aware fitness, not raw PnL.

This Hub note is a plain-language intro. Not investment advice.

The problem

FX is huge and noisy. ML on OHLC + canned indicators often overfits because the labels are chaotic and the indicator windows were never tuned to a risk-adjusted goal. Raw profit as fitness invites one huge winner and a crash.

The idea / core intuition

• Build rule features from technical indicators (thresholds, crossovers)—not the raw series alone.
• Use a genetic algorithm (population, crossover, mutation) to search indicator parameters.
• Fitness = combination of Sharpe and Sterling (return vs typical / max drawdown family)—cut variance and the size of underwater periods.
• Intraday tests on six major pairs (2018–2019) in the paper; reported equity curves are backtests.

A search heuristic—not a licensed robot.

Why after-work Python learners care

GA is just lists of candidate parameter dicts plus a scorer. Sterling vs Sharpe is a reminder that the objective is a product decision. Never confuse a 5-minute AUDUSD backtest with a live edge.

Tiny evening exercise: three toy trades pnl = [1, 1, -2]; compute mean/std Sharpe-like ratio and max drawdown from the cumulative sum. Learning only.

One-evening reading checklist

1. What gets encoded in a GA individual (rule parameters).
2. Why SSR fitness beats maximising raw return.
3. How rule-based features differ from dumping raw prices into a net.
4. Two reasons published FX backtests overstate live results (costs, regime).

Download PDF

Download PDF

Landing: arXiv abs

Important disclaimer

Not investment advice. This Hub note is educational only. GA-MSSR is an academic FX backtest study—not a recommendation to run robotic FX trading.

What this is not

• Not CEF / subsidy framing
• Not Quant or Macro course CTA
• Not a forex-robot sales page

Next step

Single course CTA: SoloQuant Academy’s Introduction to Python Programming—random.choice and lists first; a GA is a loop over scored parameter lists.