Loading
Market Intelligence › Basic Terminologies
Curve fitting, look-ahead bias, and the questions to ask with every claim
Published by The Retail Trader · JUL 26, 2026 · Advisory & Research
Producing a result on past data is easy. Repeating that result live is a completely different matter, and in between lie many mistakes that creep in unnoticed.
The most common problem is curve fitting — tweaking the parameters so much that they fit the past data perfectly. That model memorises history and learns nothing.
The second problem is look-ahead bias, where the model gets information that was not actually available at that time.
These four questions can be asked of any performance claim:
• Was this live or a backtest — and since when has it been live.
• Does it include brokerage, taxes and slippage or not.
• What was the maximum drawdown, not just the return.
• How many trades is this result based on — the result of ten trades is not data.
A claim that becomes uncomfortable at these questions is itself the answer. An honest track record shows its own worst period, because that is its most trustworthy part.
This article is educational only — it is not a buy or sell recommendation, it makes no claim of guaranteed returns, and it is not advice. Rules, rates and mechanics are changed from time to time by the exchanges, SEBI and RBI, so verify every number yourself with your broker or an official source.