Fifteen questions on the things that separate honest backtesting from wishful curve-fitting: validation, overfitting, risk math and market regimes. Pass and you get a public, verifiable certificate to share on LinkedIn or X.

Name on the certificateEmailUsed only to send your certificate link.

Question 1 of 15
What does a backtest fundamentally tell you?
How a strategy will perform in the futureHow a strategy's exact rules would have behaved in a past periodWhether a strategy is profitable in all market conditionsThe maximum amount a strategy can lose
Question 2 of 15
A backtest enters trades at a candle's close based on an indicator computed from that same closed candle. Suspicious?
Yes — that's lookahead biasNo — the candle is closed, so its data was available at decision timeYes — indicators must only use data from a week earlierNo — but only if the timeframe is daily
Question 3 of 15
What is out-of-sample (OOS) data?
Data from a different exchange than the one you tradeData excluded because of gaps or bad qualityHistory the strategy was never tuned on, reserved for honest evaluationSimulated data generated by Monte Carlo
Question 4 of 15
Which is the classic symptom of a curve-fit (overfit) strategy?
It trades too rarelyGreat in-sample results that collapse on data it never sawIt only goes long, never shortIts equity curve is too smooth in live trading
Question 5 of 15
What does walk-forward analysis do?
Repeatedly optimizes on one window and tests on the next unseen window, rolling through timeRuns the same backtest many times to average out randomnessTests a strategy on multiple coins simultaneouslyProjects the equity curve forward using regression
Question 6 of 15
The Sharpe ratio measures…
Total return divided by number of tradesReturn earned per unit of volatility (risk-adjusted return)The percentage of winning tradesHow fast a strategy recovers from drawdowns
Question 7 of 15
Why 'deflate' a Sharpe ratio?
To account for inflation eroding returnsTo convert it from daily to annual unitsBecause the best of many tried configurations has an inflated Sharpe by selection luckBecause exchanges overstate volume
Question 8 of 15
After a 50% drawdown, what return is needed just to get back to break-even?
50%75%100%150%
Question 9 of 15
A strategy wins only 35% of the time. Can it be profitable?
No — you need at least 50%Yes — if average winners are enough larger than average losersOnly in bull marketsOnly with leverage
Question 10 of 15
'Risk 1% per trade' means…
Position size is 1% of account equityThe trade's stop-loss distance implies losing at most 1% of equity if hitYou trade once per day at mostFees must stay under 1% of the position
Question 11 of 15
The main practical use of Monte Carlo simulation on a backtest's trades is…
Predicting next month's returnEstimating the range of outcomes and worst-case drawdowns beyond the single historical pathRemoving losing trades from the recordOptimizing indicator parameters faster
Question 12 of 15
A trend-following strategy 'suddenly stopped working'. The most common boring explanation is…
The exchange changed its APIThe market shifted into a ranging regime, where trend systems bleed by designOther traders copied the strategyThe indicator library has a bug
Question 13 of 15
Moving a strategy from daily to 15-minute bars typically…
Reduces the impact of fees because trades are smallerMultiplies trade count, making fees and slippage a much larger share of returnsHas no effect on costsGuarantees more profit due to more opportunities
Question 14 of 15
Bitcoin bought in January 2018 eventually recovered to strong gains. What does that history hide?
Nothing — the return is the whole storyA multi-year, roughly 80% drawdown the holder had to sit throughThat the data before 2020 is unreliableThat fees consumed most of the gains
Question 15 of 15
You test your held-out (OOS) data, tweak the strategy because results disappointed, and re-test on the same held-out data. What happened?
Good practice — iteration improves strategiesThe held-out data has effectively become in-sample; its evidential value is goneNothing, as long as the tweak was smallThe strategy is now walk-forward validated

Also send me the newsletter (monthly, unsubscribe anytime — requires a confirmation click).

The certificate attests to passing this exam only; it is educational and not a professional or financial qualification.

Crypto Backtesting Fundamentals — certificate exam