TradeLab research
Learn how to write clear rules, run a backtest, and spot weak results. Each guide shows the steps, the limits, and the risk.
Start with the question in front of you
Start with clear rules. Then check the test. Try paper trading next. No step can tell you what the live market will do.
Define
Name the market, time frame, entry, exit, size, and risk before you look at the score.
Write the rulesValidate
Check the rules, data, trades, costs, and test size before you trust the score.
Use the check listForward test
Learn what a past test can show. Then see what a paper test can add.
Compare the two stepsStress test
Check fees, spread, price slip, trade count, and trade size. Do not hide them in notes.
Check trade costsWe check each guide against the TradeLab app and public site. Updated July 29, 2026.
Research library
Check the final status, error record, trade output, and saved settings before you interpret the score.
A no-trade result is a signal to inspect the rules, data, warm-up period, timeframe, and eligibility.
Review position sizing, leverage, concentration, and combined open risk without hiding assumptions inside the headline.
Separate development from evaluation, freeze the rules, and protect reserved history from repeated tuning.
Stress commissions, spreads, slippage, market impact, and turnover before interpreting the result.
Audit when data existed, when the signal became final, and which price could actually have been available for execution.
There is no universal minimum. Judge the count alongside market coverage, independence, concentration, and stability.
Validate the specification, implementation, and evidence, then classify the result as pass, investigate, or reject.
Learn what each test checks, which one comes first, and what to do when the results do not match.
Use a six-part rule template to set the market, timing, entry, exit, size, and proof needed for a clear test.
A result that rejects a weak idea, catches a configuration error, or prevents false confidence has done useful work.
A 20/50 SMA idea produced zero daily trades in a seven-day sample and 51 noisy trades on one-minute data. The failure was the lesson.