TitanForge EA review: a backtest with the answers built in

This TitanForge EA review comes down to one test. The author's backtest shows 1,175 trades, 97.11 % winners, profit
factor 49.69. I kept every gold price exactly as it was and only changed the dates. Over 1,643 trades the profit factor
came out at 0.79, 1.00 and 0.99.

Gold prices don't know what day it is. This EA does.

Tested: TitanForge EA MT5 v1.3, the Market demo, 4 November 2026.

What does it promise?

$599.99 for an "algorithmic impulse swing system" on XAUUSD H1. No grid, no martingale, hard stop loss. Fifteen reviews,
all 5 stars. My own run on real gold with the real dates looks just as clean:

What made me suspicious?

Every trade has TP 5 USD and SL 19 USD. If the losers hit the full stop, the profit factor tops out around 8.8. To reach
49 the EA has to almost never touch its stop. Like it knew which trades would go wrong.

The win rate itself means little. A trailing stop pulls the SL to break-even after 1.40 USD of profit, so a random entry
gets there before losing 19 USD about 93 % of the time.

What did my TitanForge EA review test show?

I fed the demo real gold prices through a custom symbol, then shifted the date labels and left the prices alone.

Data	Win rate	Profit factor
Real prices, real dates	98.7 %	11.1
Every price +0.02 USD	98.5 %	6.8
Same prices, dates moved back 7 days	91 %	0.99
2023-24 prices labelled as 2021-22	90 %	0.79

Nudging the prices didn't break it, unlike Quantum iGold. Moving
the dates broke it every time. The last row is the clearest: the same price path that looked perfect under its real
dates loses money once the calendar says 2021.

So the trades follow the date, not the price.

What happens on a live account?

Live is always a new date, and there's no stored history for tomorrow. Without it, 6.6 to 9 % of trades hit the stop,
each one -19 USD × lot. The author's signal has 27 trades with no stop hit yet, but at its current 2.8 lots one hit costs
about 27 % of the account. 31 % of its profit came from a single NFP trade with 26 USD of lucky slippage. And buyers on
IC Markets say they dont get the trades the signal gets.

Why will the next version look perfect too?

The first week already brought two "urgent re-optimizations", v1.1 and v1.2, and the v1.3 tester repeats the signal's
earlier trades. Each new version brings the table up to its build date. So there will be more versions, and each one
will look perfect up to the day it came out. The days after a release are the ones you actually trade.

Verdict

Don't buy it. The backtest is history the EA already had, and on new dates it breaks even at best, before costs.

Want to check yourself? Copy real gold into a custom symbol, move every date back 7 days, and run the demo on both. If
the result collapses, thats your answer. Next post: Morpheus Algo, from the same author.

The two kinds of problem in this series: MQL5 EAs I wouldn't buy.