Backtested a dead-simple EMA trend system on SOL for 6 months: 35% win rate but still net green. Win rate is a trap metric

Wanted to share a backtest that's a clean reminder of why win rate on its own means almost nothing. Method, numbers, and honest caveats below. Would genuinely love feedback on the exit logic.

The system (kept it dead simple on purpose):

\- SOL, 1H chart

\- Enter long when the 9 EMA crosses above the 21 EMA

\- Exit at +8% target or -4% stop (a hard 2:1), positions can hold across days

\- $1,000 per trade, small slippage assumed

6-month results (Jan to Jul 2026, 40 trades):

\- Net: +7.65%

\- Win rate: 35%

\- Avg win +$27.8 / avg loss -$25.9 (realized \~2:1)

\- Profit factor: 1.07

\- Sharpe 0.63 / Sortino 7.87

\- Max drawdown: -24.5% (worst run: 7 losses in a row)

The takeaway:

A 35% win rate feels awful, but because the winners run roughly 2x the losers, expectancy stays positive. Classic trend-following shape: you bleed small on the chop and catch the occasional big leg. The gap between Sortino (7.87) and Sharpe (0.63) is the interesting part, most of the volatility is upside, it's the losing streaks that test you, not any single trade.

Honest caveats so nobody takes this as gospel:

\- 6 months and one asset is a small sample. This is an illustration, not an edge I'd bet the farm on.

\- Crypto perps carry funding costs I didn't model, so real returns would be lower.

\- Sitting through 7 straight losses and a 24% drawdown live is a lot harder than reading it in a table.

Question for the swing crowd: would you swap the fixed 8% target for a trailing stop to let the big trends run further, or does that give back too much on the reversals? What has actually worked for you on trend systems?

(ran this on RIX, an AI backtesting tool I use, happy to paste the exact strategy code if anyone wants to replicate it or tear it apart.)