TL;DR — Analysis of ~150 live trades across 4 accounts showed the strategy has a real edge (55% win rate, 1.46× average win vs loss on the cleanest account), but two things were bleeding it: over-trading (commissions turned gross-winning accounts net-negative) and a low-quality trade bucket (C-grade setups lost money after costs). v4.2.4 makes three targeted fixes to keep more winners and cut avoidable losers, plus adds Fair Value Gap mapping on the chart. No change to the core strategy, the runner logic, or the stepped stop ladder.
What Changed — 4 Items
1 · Higher-quality trades onlyTuning
Raised the minimum trade grade from C to B. The weakest-confluence setups are now skipped entirely.
DATA: C-grade trades won only 30% and lost ~$7.83/contract each after costs — while A and B trades were both net positive. Removing this bucket is the cleanest single improvement in the dataset.
2 · Fewer, better entriesTuning
Tightened the three main entry gates so the bot waits for stronger, cleaner momentum before entering a trade. Fewer trades, higher average quality, lower commission drag.
Setting
Was
Now
Purpose
Min power imbalance
80
85
Require stronger directional dominance
Min power gap
50
55
Wider spread between buyers/sellers
Min trend strength (ADX)
20
23
Stricter "is this actually trending" filter
DATA: A 146-trade account was +$334/ctr gross but −$29 after ~$360 of commissions. A 36-trade account kept +$178 net. The difference was almost entirely trade count. Target: ~30–40% fewer trades while keeping the best setups.
3 · Hard per-trade loss capRisk Control
New $400 maximum intended loss per trade. Before entering, the bot automatically reduces position size so that stop-distance × contracts × point-value cannot exceed $400. Protects against outsized losses on wide-stop, multi-contract positions.
DATA: Full-stop losses were the single biggest point leak (−260 pts total), including one trade that lost approximately $700. This cap prevents that class of loss by construction. Note: caps intended risk — cannot fully prevent gap/slippage beyond the broker’s stop order.
4 · FVG mapping on the chartVisual Only
Fair Value Gaps (3-candle imbalance zones) are now drawn as boxes on the chart — teal for bullish gaps, coral for bearish. Boxes auto-extend and are removed once price fills the gap. The engine already uses FVG data internally for scoring; this simply makes the gaps visible for the operator. Zero effect on trading logic.
What Did NOT Change
The core Power Engine strategy — power-imbalance momentum entries are unchanged in concept. The bot still reads the balance of power, enters when one side decisively dominates, and skips balanced/choppy conditions.
The runner logic — the mechanism that captured +48-point, +34-point, and +30-point winners is completely untouched. Big winners are not capped.
The stepped stop ladder — early breakeven rung (+9 pts), half-lock rung (+18 pts), and the stepped trail are all unchanged.
Confidence-based sizing — higher-conviction trades still get more contracts; lower-conviction trades still size down.
All safety systems — latched daily lockout, blackout windows, pre-halt flatten, broker-side disaster stops, session gates — all unchanged.
What We’re Watching Next Session
This release is a set of pre-registered, data-driven hypotheses. On the next clean session we verify:
Did trade count drop ~30–40%? (selectivity working)
Did after-cost results turn positive across all accounts? (cost drag addressed)
Did the $400 cap trigger on any trade? (sizing cap actively protecting)
Are C-grade entries now essentially zero? (grade floor enforced)
If the metrics confirm, the changes stick and we move to the next refinement (exit-model A/B testing, runner-room calibration). If a metric misses, we adjust one variable and retest. No shotgunning.
Deployment
Load v4.2.4 and confirm the dashboard banner reads POWER v4.2.4.
Recreate both alerts (execution + heartbeat) on v4.2.4. Alerts freeze the version and settings at creation time — stale alerts silently trade the wrong engine.
Confirm the first trade’s extras.reason reads “Power” — the ground truth that the Power Engine is live.
All display defaults (Surf Report, large font, bottom-left, Automated source) and new parameter values are baked in. No manual configuration needed.