I ran one of my range-reclaim setups across 386 coins, roughly 220,000 configurations per coin, on a single RTX 3070 chewing through it overnight. After all of that, exactly four symbols cleared the bar I set for a deeper test. Four. Out of three hundred and eighty-six.
The one that caught my eye wasn't the flashy one. It was BILL.
Why BILL looked like the keeper
MEGA had the loud numbers: a 95.9% win rate, profit factor 3.5. But it also carried a −14.2% maximum drawdown on every config. That scared me off fast. BILL looked calmer. Over 1,600 simulated trades it posted a 96.7% win rate, a profit factor of 2.12, and a worst-trade drawdown of just −5.1%. Lower risk, almost the same hit rate. On a screenshot it's the obvious pick.
So I pulled the one column I've learned to check before anything else.
The DCA-rescue rate
My backtester records how often a trade only finished green because it averaged down first — bought a second, cheaper entry after the position went against it, then closed for a small win once price crawled back. On BILL, that rate was 72%.
Seventy-two percent. Nearly three out of four "winning" trades were losing trades that got rescued by adding more size at a worse price. MEGA, the one with the scary drawdown, sat at 65%. So the coin that looked safer was actually leaning on averaging-down harder.
That's the trick the small drawdown hides. BILL's −5.1% isn't proof it's gentle. It's proof it adds to losers early and often, which keeps any single trade's paper loss shallow — right up until a move doesn't come back and you're holding a doubled bag instead of a single one.
| Setup (1,600 trades) | Win rate | Profit factor | DCA-rescued | Worst drawdown |
|---|---|---|---|---|
| BILL — "the safe one" | 96.7% | 2.12 | 72% | −5.1% |
| MEGA — "the scary one" | 95.9% | 3.47 | 65% | −14.2% |
What four survivors out of 386 actually tells you
Here's the part that stuck with me. All four coins that made it through — BILL, MEGA, and two others — were the same family of setup, and every one of them won the same way: by averaging down into red. I didn't find four edges. I found four flavours of the same illusion, and a backtester generous enough to always have margin for one more buy.
A live account doesn't work like that. It has a real balance and a real liquidation price. The simulator survives a −5% or −14% excursion because it can. You survive it only if your size and leverage leave room. And a 96% win rate quietly tempts you to use more leverage, not less, which is exactly backwards.
What I actually do with this
I throw out win rate as a headline number. When I look at a backtest now, the first two columns I read are the DCA-rescue rate and the worst-trade drawdown, in that order. If a strategy needs to average down to win, I assume the win rate is borrowed, not earned, and I price the trade as if that rescue won't be there.
And before any leverage goes near a thin coin like this, I check where the position dies. If you're sizing a leveraged trade, run the real number first — the liquidation calculator shows how little room you actually have, and the position size calculator sizes the trade from what you're willing to lose instead of what a green backtest promised. BILL's 96% didn't change my mind. The 72% did.