Published
Last week my backtester spat out a result that, a year ago, would have made me reach for the leverage slider. One altcoin — MEGA — came back with a 95.9% win rate, a profit factor of 3.3, and a simulated net of +617 over 1,600 trades. Nine hundred out of every thousand trades green. If you saw that on a screenshot you'd assume someone found the holy grail.
I closed the tab and moved on. Here's why.
The number that kills it
The backtest tracks a column most people never look at: how often a trade only survived because it averaged down. On this MEGA setup, the DCA-rescue rate was between 52% and 68% depending on the config. Read that again. More than half the "wins" were positions that first went against me, got a second (cheaper) buy thrown at them, and only then crawled back to a small green close.
That's not a 96% win rate in any honest sense. It's a 96% eventually-clawed-it-back rate. The strategy is quietly converting losing trades into "wins" by doubling exposure at the worst moment — exactly the behaviour that ends accounts.
Look at the worst case, not the average
The other column I trust more than win rate is maximum adverse excursion — the deepest underwater a trade went before it closed. On every single MEGA config it was the same: −14.2%. With any real leverage on a coin that thin, a −14% drawdown isn't a dip you wait out. It's a liquidation. The 96% only exists in a simulation that always has the margin to add more and wait. In a live account, one of those losing streaks wipes you, and the whole pretty curve goes with it.
So the honest translation of "96% win rate, PF 3.3" is: this thing wins small, often, by sitting through −14% drawdowns it survives only because the simulator never runs out of money. That is not an edge. That's a margin call wearing a green hat.
Why I trust this instinct now
I've run five different bots and thousands of configurations, and the pattern repeats every time: the setups with the prettiest win rates are almost always the ones leaning hardest on averaging down. A high win rate is the easiest statistic in trading to manufacture and the least correlated with actually making money. Profit factor, drawdown, and DCA-dependence tell you the real story. Win rate is the number strategies wear to a job interview.
How to check it yourself
You don't need a backtester to feel this. Open the DCA reality calculator and model a position that averages down through a 14% drop — watch how the "average price" flatters you while the realized number bleeds. Then run the same trade through the liquidation calculator at the leverage you'd actually use, and see where −14% puts you. If the answer is "liquidated," the 96% was never yours to keep.
I'll keep mining these backtests and posting the ones worth talking about — the winners and the traps. This one was a trap.