← Home

WorldQuant BRAIN

I built a Python pipeline to research trading signals, backtest them on WorldQuant BRAIN, tune variables with a genetic algorithm, and submit the top performers.

Ranked #1 in Singapore on the platform's Genius leaderboard in Q2 2026.

Research pipeline

  1. Research ideasForm alpha expressions
  2. BacktestWorldQuant BRAIN API
  3. Fit variablesGenetic algorithm
  4. SubmitTop performers

Each iteration feeds tuned candidates back into the next backtest cycle.

Ideas simulated
228,549
Signals submitted
304

Live throughput

Fresh backtests per hour
68.8

The bottleneck is BRAIN's backtest and simulation time. Throughput varies with server load, region, and universe size.

Median Sharpe retention

OOS / IS
47.9%

Retention is the ratio of out-of-sample to in-sample Sharpe.

Best alpha

Analyst CAPEX forecast alpha

OOS Sharpe
2.21
OOS fitness
2.05