Demo accounts lie in specific, predictable ways

[Fairman]

Demo accounts lie in specific, predictable ways

A strategy makes money on demo for six weeks. It loses money on live. The first reaction is to blame psychology. Often, the real explanation is mechanical.

Demo fills are usually instant. There is no queue, no slippage, no last look. Spreads may be tighter or more stable. Stops fill at the exact price. Large orders fill in full, regardless of size.

Some demo servers also use different data feeds from the live ones, with fewer spikes and cleaner candles.

Live trading adds all the friction at once. The result is a gap in performance that has nothing to do with emotions.

It helps to treat demo as a test of logic, not of results. Does the code work? Do the orders go where they should? Are the signals correct?

Live results with minimal size then serve as the real test. The gap between demo and live, measured over a few hundred trades, gives a realistic estimate of friction.

How large was the gap between demo and live performance on the last strategy, and how much of it came from fills?