Tested a free BTCUSD EA — 90% wins on 118 trades. Real edge or cu

Hi everyone,

Posting this for discussion because the numbers look great but the sample size makes me want a second opinion from people who trade crypto bots.

It's a free BTCUSD scalper EA called IS New Project, MT4 only, M1 timeframe, trades the 21-23 GMT window. Strategy combines grid placement with a bar-size filter that tries to skip flat-market periods.

Backtest stats:

— Win rate: 90% (106 wins / 118 trades) — Profit factor: 9.99 — Max drawdown: 6.65% — Gain: +787% — Min deposit: $100 — Recovery method: grid with wide spacing + fixed TP and trailing stop

What's making me cautious:

118 trades is a tiny sample for any conclusion. Bitcoin spent 2024 mostly trending up; an M1 scalper backtested only on that regime is going to look brilliant. I'd want to see it through Q1 2022 (the crash) and Q3-Q4 2023 (chop) before trusting the 90%.

A 9.99 profit factor in live trading almost always degrades 30-50% from backtest due to slippage, broker spreads on crypto, and execution delays at M1.

Grid recovery on BTCUSD is fine until it isn't. Volatility spikes during news or weekend gaps can break the grid in a single session.

How I'm running it on demo:

— 0.5% risk per trade for the first month — Demo for 4-6 weeks across mixed bull/bear/sideways days — Tracking live performance vs backtest pattern — Will scale up only if live matches backtest within 70%

Full review with all parameters and a setup walkthrough is on ForexCracked: https://www.forexcracked.com/forex-e...free-download/

Asking the community:

— Anyone here actually trading a BTCUSD grid EA live? What's your real-world spread budget? — What's your minimum live-trade count before you trust a backtest? — Anyone tested the bar-size filter approach on alt pairs (ETH, SOL)?

Disclaimer: not financial advice. Free download, but free doesn't mean safe. Demo before live. Crypto especially.

Cheers.

Quick follow-up to the opening post.
For anyone who wants to look at the EA itself before commenting — the .ex4 file and default .set presets are in the article. The interesting bit is the bar-size filter; it skips trades when the average bar size over the lookback window falls below a threshold. That's the mechanism that's supposed to keep it out of flat ranges.
My read after looking at the code: the filter does help, but it's tuned to bitcoin volatility patterns from a specific period. Run it on demo through a low-vol weekend and you'll see the filter pause entries for hours at a time, which is the desired behaviour. The risk is that the threshold doesn't adapt if BTC's baseline volatility shifts.
Interested if anyone here has run similar bar-size or ATR filters and how you handle the threshold drift.