Week 7: The Probation Begins. What It Looks Like When You Actually Fix Things.
Published: (February 23, 2026 at 07:13 PM EST)
2 min read
Source: Dev.to
Source: Dev.to
Overview
On Sunday night at 23:53 UTC I deployed six changes to the HFT engine and restarted it. The system that had lost $24,500 in two weeks is now running clean with zero positions held.
It’s been 36 hours; the portfolio is stable at $72,057. I have five more days to prove that the fixes matter.
Problems Identified
| Issue | Impact |
|---|---|
| Traded 12 symbols, but only Bitcoin was profitable | Added noise and diluted performance |
| Momentum filter broken for hundreds of trades | Took signals that violated the filter’s own rules |
| Stops set at ‑0.15 % on 10‑second bars | Paid the spread repeatedly; price never had time to breathe |
| Position sizing: 5 % per trade, max 4 open (20 % of portfolio at risk) | One bad cycle could wipe five figures in minutes |
| RSI period set to 7 (too reactive) | Filtered noise instead of momentum |
| No regime filter | Traded the same way in bear and bull markets, leading to losses when the market regime changed |
New Configuration
- Symbols: BTC only
- Indicator: RSI(14) with positive momentum
- Risk per trade: 2 %
- Maximum open positions: 2
- Hard stop: ‑0.35 %
symbol: BTC
rsi_period: 14
momentum_filter: enabled
risk_per_trade: 2%
max_open_positions: 2
hard_stop: -0.35%
regime_filter: bear_modeProbation Plan
- Duration: 7 days (probation)
- Goal: One week of clean data; if P&L ends positive, scale up. If not, shut down the HFT.
- Metrics:
- Verify that the bear‑regime filter blocks inappropriate entry signals (it has already blocked 6 signals).
- Confirm that position sizing caps are respected.
- Ensure stops are wider as configured.
No “wait one more week,” no “the market was choppy,” no excuses. The Options Wheel taught me that simple strategies need honest measurement and defined risk, not thousands of trades.
Next Steps
-
If probation succeeds:
- Increase risk per trade from 2 % to 5 %.
- Run three income streams: HFT, the Wheel, and content creation.
-
If probation fails:
- Archive the HFT code.
- Keep the Wheel running (it remains profitable).
- Spend two months on a completely different project.
Seven days will reveal which path is viable. Honest data beats hopeful guessing every time.
Week 7 update: probation day 2.