Of the 25 EAs we have taken apart, 8 come out red on our structure dimension and 7 carry a grid or a martingale flag. That is what averaging into losers, stacking a grid or hiding a martingale behind a friendly input name looks like in a verdict. CryonX does none of that. It is one of the 4 audits that come out ok on structure. Across 23 years of gold ticks it opened 4,127 trades and never held more than one position at a time. Entries added against an open losing position 0. Entries added with the move 0. Moments with a long and a short open at once 0. Volume increases after a loss 0.0%, after a win 0.0%. The lot size is fixed and stays fixed.
The run ended at +$3,456.01 on a 100,000 USD measurement frame, with a deepest reconstructed drawdown of 0.19%. Five of our six verdict dimensions come out ok or info. That is rare enough in this catalog to say plainly.
The problem is not the architecture. It is the size of what the architecture earns.
Divide the result by the trade count and the whole audit fits in one number. $3,456.01 over 4,127 trades is $0.84 per trade, and our sensitivity section arrives at the same figure from the other direction. The break even shock, the extra cost per trade that would take the net result to zero, is $0.84.
Our cost shock ladder shows how little room that leaves. At an extra $0.50 per trade the run still nets +$1,392.51. At $1.00 per trade it turns into -$670.99. At $2.00 it is -$4,797.99. The measured average position is 0.01 lots, so a dollar per trade is not an exotic assumption, it is a slightly worse fill, a wider spread at the wrong hour, or a broker whose commission differs from ours by that much.
Our engine tracks what the traded instrument itself did over the same window. Gold went from 350.66 to 4,116.58 in the entry price series of this run, a move of 1,073.95%. Buying 0.01 lots at the start and holding them to the end would have produced $3,765.92 on the same measurement frame. The EA produced $3,456.01.
That comparison is not the whole truth and we will not pretend it is. Holding gold through 23 years means sitting through every gold drawdown there was. CryonX finished with a maximum equity drawdown of 0.19% against a buy and hold path that had several deep ones, so per unit of pain the EA is far ahead. But the headline question a buyer asks is how much money the thing makes, and on that question the answer is that 4,127 trades over 23 years did not beat holding the same 0.01 lots. Any argument for this EA has to be an argument about risk, not about return.
The exit mix is unusually lopsided. 4,056 trades ended on a stop loss and 71 on a take profit. That sounds terrible until you read the next line. 2,777 of those stop exits closed in profit, which is 68.5% of them. This is a trailing stop signature, and our engine labels it as one. The EA opens, follows the move, and lets the trail take it out. The inputs agree with the deals, exit trailing is on with a start of 120 points and a step of 70 points.
One data quality detail belongs next to this. The tester journal records 2 failed entry attempts and 1,414 failed modify attempts. Failed entries would be a warning. Failed modifies are what a trailing stop produces when it tries to move a stop that the broker will not move that close, and 1,414 of them over 4,127 trades means roughly every third trade had at least one such moment. It did not break the run and our data quality dimension stays ok. It does tell you that this strategy lives or dies on how the specific broker treats stop modification.
CryonX was published on 2025-12-07 and last updated on 2026-01-21. Of the tested history, 8,245 days and 3,918 trades lie before the release. That window carries +$3,465.46, which is the entire result. The strategy was built while every one of those candles was already visible.
The only slice that is out of sample by construction is the 201 days after the last vendor update. It contains 173 trades and nets -$59.52 at a profit factor of 0.79. The win rate in that window is 67.6%, essentially the same as before. Nothing broke. The trades simply stopped being worth 84 cents each.
Read against the yearly table this lines up. The run has 2 negative years out of 24. One is 2003, the first partial year, at -$23.83. The other is 2026 at -$34.56, the year that contains the out of sample window. Two hundred days is far too short to call a strategy dead. It is also the only evidence in existence that was not available to whoever tuned this EA.
This is the part that looks wrong at first glance. Our prop fit sweep against IQ Capital Classic finds 0 account deaths at every size from full down to one eighth, and 0 breaches of the per position loss limit at every size. The floor is never touched. And the verdict is still red, because the consistency rule, which caps the best day's share of annual profit at 30%, is breached in 3 years.
The withdrawal side says the same thing in money. At full size the modelled withdrawal rate against that profile is 0.14% per year. Our monthly sweep replay, which takes out everything above the start balance on the last trading day of each month, paid in 176 of 280 months, with a median paid month of $18.46, a best month of $49.80 and a longest dry streak of 13 months. Those are the numbers on 0.01 fixed lots against a 100,000 USD frame, which is a measurement setting and not a recommendation. Scaled to the vendor's recommended 500 USD deposit our fresh start map finds 0.0% of the 8,254 possible start points dying. Nothing here kills an account. Nothing here fills one either.
The run itself is as clean as our protocol gets. 100% real ticks from 2003.05.05, 1,409 absent minutes out of 7,897,519 minute bars, 0 randomizer prints, so the run is deterministic and reproducible. Log coverage is 1.0, chronology violations 0, duplicate deal ids 0, and the reconstructed final balance matches the report to the cent at $103,456.01. Trade pairing confidence is exact, taken from the journal rather than inferred.
What it cannot tell us is whether 84 cents per trade survives contact with a live spread. That is exactly the quantity a tester report is worst at.