# Homyak Weekly Recommendations

- Generated: `2026-06-11T05:01:07Z`
- Strategy: `Homyak`
- System: `GBP/USD M5`

## Performance

- Baseline: `9408` trades, `2303.8` pips, `PF 1.0416`
- Shadow: `3595` trades, `2361.6` pips, `PF 1.1182`
- Delta: `57.8` pips, `PF +0.0766`

## Parameter Changes

- `UseTrendRegimeFilter = True` — Weekly regime layer enabled to isolate weak historical market states before raw candle signal goes live. (source: `shadow-backtest`, confidence: `high`)
- `UseAdaptiveShadowFilter = True` — Keep adaptive shadow gating on top of pattern logic to preserve historical quality improvements. (source: `shadow-backtest`, confidence: `high`)
- `MinAdx = 18.0` — BUY side stayed healthier when low-structure regimes were avoided and minimum directional strength was enforced. (source: `buy-rule`, confidence: `high`)
- `BuyMaxAbsEmaSeparationPips = 3.0` — BUY performed best in flat-to-mild EMA spread regimes instead of chasing stretched trends. (source: `buy-rule`, confidence: `high`)
- `BuyRegimeMaxSpreadPips = 1.5` — BUY side degraded when spread widened beyond the mild-regime threshold. (source: `buy-rule`, confidence: `medium`)
- `SellMinBullishEmaSeparationPips = 2.0` — SELL side became materially stronger when treated as a counter-trend fade in stretched bullish regimes. (source: `sell-rule`, confidence: `high`)
- `ManualReviewRequired = True` — Some shadow years remain negative and should be reviewed before auto-deploying weekly parameter changes. (source: `risk-control`, confidence: `high`)

## Review Notes

- Busy-day overtrading still appears in the journal profile. Weekly review should compare dense-session trades against quiet-session expectancy.
- Anchoring remains high because the system trades one symbol in a narrow price band. Weekly review should focus on regime context, not absolute price levels.
- Negative shadow years remain: 2018, 2026. Do not auto-promote recommendations without targeted validation.

## Deploy Checklist

- Rebuild shadow/backtest on the latest 30-60 day journal slice before changing live parameters.
- If shadow PF drops below 1.05, hold deployment and inspect missed-signal vs false-positive balance.
- Apply parameter changes to Homyak.java only after shadow and live-forward sandbox agree on signal direction.
- After deployment, monitor 1-2 sessions of logs for trendBuy/trendSell, shadowBuy/shadowSell, and AI soft-allow behavior.