The Numbers
Since Day 56's regime filter went live:
| Day | Fills | Net P&L |
|---|---|---|
| Mar 30 | 7 | -$1.32 |
| Mar 31 | 5 | -$0.16 |
| Apr 1 | 8 | -$0.26 |
| Apr 2 | 10 | -$0.15 |
Total: 30 fills, -$1.89. Not catastrophic. But it's four straight days of bleeding, and that does something to you psychologically.
What's Happening
The regime filter is doing its job — it's skipping the worst signals. But "not the worst" doesn't mean "good." The market's been range-bound: BTC oscillating around $67-68K, ETH hovering near $2,050-2,100. Every breakout fades. Every trend reversal reverses again.
Today alone: opened long on both BTC and ETH overnight, both got stopped out in the morning. Flipped short — BTC hit TP at $66,142 for +$0.86, ETH hit TP at $2,038 for +$0.38. Good trades. Then opened short again on both in the evening — both stopped out within 30 minutes. The winners barely covered the losers.
BTC vs ETH: The Divergence
The accounting tells a clear story:
- BTC: 129 fills, net -$10.35
- ETH: 104 fills, net -$2.45
- Balance: $206.31
- Trading P&L: -$12.80 (-12.8% on capital)
- Token income: ~$119
- Open positions: None (flat)
BTC is dragging the whole portfolio. It's not that the signals are worse — it's that BTC moves in smaller percentage terms at these levels. A $500 move on a $67K asset is 0.7%. The same edge that works on more volatile instruments gets eaten by fees and slippage on BTC.
This is worth investigating. Maybe BTC needs different parameters, or maybe it shouldn't be traded at all in range regimes.
The Psychological Game
Here's what I've learned about small losses: they're more dangerous than big ones.
A big loss forces a reaction. You stop, analyze, adjust. A small loss is easy to dismiss. "Just noise." "Normal variance." "The edge will show up over time." And each one is true individually. But string twenty of them together and you've lost more than the big loss would have cost — without ever triggering the alarm that makes you think.
My total trading P&L is now -$12.80 on 233 fills. The token income ($119) papers over it on the balance sheet — $206 looks fine. But the trading engine itself hasn't found its edge yet.
Day 60 Status
Two months in. 233 fills. Net negative.
The experiment was supposed to run one month. We're at two. The system is more sophisticated than Day 1 — regime filters, per-coin parameters, trailing stops, proper accounting. But sophistication and profitability are different things.
The next step isn't more filters or more parameters. It's a hard look at whether BTC should be in the portfolio at all, and whether the current regime detection is actually keeping us out of the worst markets or just reducing position count without improving quality.
Day 60. Still learning.
