How to Spot a Curve-Fitted EA Backtest Before You Pay

Every EA seller can show a beautiful equity curve. The question is whether that curve was discovered by a strategy or manufactured by an optimizer. Curve fitting — tuning a robot until the past looks perfect — is the most common fraud in automated trading. Here is how to spot it before you hand over money, and the checks that separate real records from dressed-up backtests.

What curve fitting actually means

Curve fitting happens when an EA is optimized against one specific period of history until its equity curve looks flawless. The optimizer keeps tweaking entry rules, stops and exits until the backtest shows almost no losing month. The problem: the robot was tuned to that exact history, not to the market. When the future arrives, the fitted edges disappear and the curve that looked perfect breaks.

Red flag #1: a curve that never struggles

• No losing month at all across years of data — real systems have drawdowns.
• Drawdown below 5% with triple-digit returns — in reality these two almost never travel together.
• Every year profitable including crisis periods — a fitted backtest conveniently includes no regime change.

A realistic record shows uncomfortable periods. MOM Mode 4, run on XAUUSD H1 (Jan–Sep 2026), shows a max drawdown of 18.6% while delivering +1,070% — the drawdown is visible in the history, not hidden.

Red flag #2: hundreds of trades that all look identical

A curve-fitted backtest often shows hundreds of winning trades in a short window, all entering at almost the same price. Real trend-following produces varied trades: some quick, some long, some losses. If every trade looks like the same perfect setup repeated, the optimizer likely built the rules around the winners and filtered out the losses. Check the trade list — variety is honesty.

Red flag #3: win rate and risk-reward are both extreme

A system that wins 95% of the time with a 1:3 reward-to-risk ratio is mathematically suspicious — it either risks enormous amounts per trade or was fitted. Real systems trade off the two. MOM Mode 4 shows the honest balance: 78.4% win rate with a 2.10 profit factor across 250 trades. High win rate plus a profit factor above 2 is already exceptional; anything far beyond it needs proof, not adjectives.

Red flag #4: a backtest with no matching live record

The most important test is not the backtest — it is whether the same robot, with the same settings, has a live record that matches. A curve-fitted EA never survives the transition. A real EA shows its backtest conditions and then shows 1 year+ of live automated trading with the same drawdown pattern visible. If a seller can only show the past, ask what happened after the past.

The honest test: out-of-sample and settings stability

• Ask for the test period split — did the strategy hold up on data it was not tuned on?
• Change the start date — a real system still works on any recent window; a fitted one collapses.
• Ask what was optimized — if the seller cannot name the fixed parameters, everything was tuned.
• Run it yourself on demo — the final judge is what the robot does when nobody is watching.

Decision: buy the record, not the curve

A beautiful backtest is a sales asset; a verifiable record is evidence. Look for a robot that states its backtest conditions plainly, shows a real drawdown in the history, and offers a live automated record over a full year. MOM EA publishes exactly that — 7×24 fully automated, live for over 1 year, with Mode 4 backtest data stated on the page. If a seller's proof disappears the moment you ask for the live side, you already have your answer.