I promised I'd report back when I had real data. Here it is.
971 trades. 1.5% win rate. Losses totaling $24,500 from a $97,231 starting balance. The account sits at $72,600 now.
That is not bad luck. That is a broken system running at full speed.
The Audit
I ran a full post-mortem this week. Not a "let me check a few charts" review — an actual line-by-line teardown of every parameter that touched the strategy.
Six problems. All of them my fault.
Problem 1: Spraying across 12 symbols
The engine was trading BTC, ETH, SOL, AVAX, DOGE, LTC, and six others simultaneously. When I broke out P&L by symbol, one thing became clear immediately: only BTC was consistently profitable. Everything else was noise with a cost attached.
I'd assumed diversification would smooth returns. What it actually did was dilute the one edge I had while multiplying my exposure to symbols I had no real read on.
Problem 2: Stop loss at -0.15% on 10-second ticks
On a 10-second chart, -0.15% is not a stop loss. It's a guarantee. Crypto moves 0.1% on a bad sneeze. I was stopping out of positions before they had any chance to breathe, paying the spread and slippage on every single exit.
This alone explains most of the loss count.
Problem 3: RSI period of 7
RSI(7) on a 10-second chart is basically a noise meter. It oscillates constantly, generating false signals every few bars. I switched to RSI(14) — still fast, but stable enough to mean something.
Problem 4: Broken momentum filter
The filter was supposed to exclude entries during weak momentum. The threshold was -0.3%, which was so permissive it let almost everything through. In practice: no filter.
Reset to mom3 >= 0 — positive momentum required to enter.
Problem 5: Position size at 5%
5% per trade with a max of 4 simultaneous positions = 20% of capital exposed at once. For a strategy with a 1.5% win rate. The math on that is not subtle.
Cut to 2% per trade, max 2 positions.
Problem 6: No hard stop-loss floor
This one is self-explanatory. No floor meant losses could compound freely during drawdowns. Hard stop now at -0.35%.
What I Changed
The new parameters are blunt and conservative:
- BTC-only (drop everything else)
- RSI(14), not RSI(7)
- Momentum filter:
mom3 >= 0 - Position size: 2% (was 5%)
- Max open positions: 2 (was 4)
- Hard stop-loss: -0.35% (was -0.15%)
Engine restarted Sunday night. It's on paper-only probation until it shows positive P&L over a full week.
The Harder Part
The system had been running for weeks. It was generating activity — hundreds of trades, logs, dashboard metrics. It felt like progress.
It was not progress. It was a busy system destroying capital on every cycle.
The lesson I'm sitting with this week: a broken system running is worse than no system. At least nothing doesn't cost you money.
I should have run this audit after day three, not week four. The signals were there — win rate tracking below 30% in the first 200 trades. I explained it away as warmup, market conditions, parameters needing time to settle.
It wasn't any of that. The edge wasn't there, and I was funding the discovery with real losses.
What Week 6 Will Show
Either the new params work or they don't. I'll know by end of next week. If the win rate holds above 45% on BTC with the tighter filters, there's something to build on.
If not, I stop trading this strategy and go back to the drawing board. No more "let it run a bit longer."
The Options Wheel is still running in parallel — slower, less exciting, but showing positive returns. That story is for another post.
Week 1 → Week 5: I started this series saying I'd report honestly. 971 trades at 1.5% felt like the worst entry I could write. But this is exactly what the series is for.
Top comments (0)