Demo results can look better than they deserve. These are the signs worth checking before trusting them.
No slippage at all in the demo trade history.
Fills at exactly the requested price during news.
No commission on an account type that charges commission live.
Spreads on demo tighter than on live at the same time.
Stops never filling beyond their level.
Very short trades all profitable, which can point to latency arbitrage against a demo feed.
How to check: run a micro live account in parallel for a few weeks with the same EA and compare trade by trade. Any systematic difference is the cost of reality.
A worked example: if live fills average 0.15 pips worse than demo and the EA averages 0.4 pips per trade on demo, expect around 0.25 pips live, before any other differences. That's a 40% cut in edge.
Treat the demo result as an upper limit on what live trading can do.
Nothing in the GBPAUD demo survived contact with a live account. Over seven weeks the demo booked an eleven point four percent gain with no slippage, news fills at the exact requested price, and stops that never printed beyond their level. I ran a micro live account in parallel for four weeks with the same settings. Live fills averaged twenty-eight hundredths of a pip worse, and fourteen of fifty live trades were filled beyond the price the demo would have given during the London and New York overlap.
The demo expectancy was zero point six two pips per trade. Subtract that fill gap and you are near zero point three four before anything else goes wrong. Spreads on the demo sat around zero point six at timestamps when the live account was quoted at one point one.
I treat a demo result as the best case now, not as a forecast. If I cannot see the same expert lose a bit to slippage and wider spreads on a tiny live account, I do not scale it. The missing friction was the whole edge.