See what the research feels like in your inbox.

Executive finding

The 5-minute version is not broadly broken; it is regime-fragile. Its small overall edge (+24,885 across 1,356 trades, PF 1.05) disappeared in the latest chronological out-of-sample segment (−4,060; PF 0.97). The clearest validated drag is shorts taken in NORMAL volatility with positive GEX: negative in both development (−63/trade, PF 0.82, n=157) and out-of-sample (−92/trade, PF 0.67, n=55).

The first apparent explanation—shorts lacking Delta confirmation—was real-looking but failed to explain the curve: it contained only seven trades. It is a warning, not a strategy-level conclusion.

What the research found

Across all shorts, high-volatility trades were the source of meaningful performance: +40,910 total P&L, 50.6% win rate, PF 1.94. In contrast, LOW volatility shorts lost 11,310 (PF 0.67) and NORMAL volatility shorts lost 14,465 (PF 0.91). This rejects a blanket “shorts are the problem” conclusion.

Positive GEX is also unfavorable for shorts historically: −19,075 across 318 trades, PF 0.84, versus +32,765 under negative GEX, PF 1.32. GEX is not a directional command, but here it identifies a market structure in which this strategy’s short entries have not converted.

What survived scrutiny

The robust finding is narrow but actionable for research:

• Shorts in NORMAL volatility with positive GEX were weak in both chronological segments.
• The weakness is not simply a lower win rate. In the out-of-sample segment, it won only 30.9% of the time and produced PF 0.67.
• Path evidence suggests both entry/path quality and exit giveback matter. Among 57 reconstructed low/normal-volatility, positive-GEX shorts, 41 losers (71.9%) first moved favorably, then reached adverse excursion. Losers reached favorable excursion quickly—median 6.5 minutes—but later deteriorated, with median giveback of 46.5 points.
• Winners followed the opposite path: all 16 first experienced adverse excursion, then developed into sustained moves; their median time to Maximum Favorable Excursion (MFE) was 45.6 minutes.

So a tighter exit alone is insufficient: many losing shorts initially look workable. The more fundamental issue is that these entries frequently fail to develop after an early favorable move.

What did not hold up

• Delta confirmation: short × UNCONFIRMED_30S lost 4,895 across seven trades and its single out-of-sample trade lost 1,465. That is too little evidence to establish a durable rule or explain a 1,356-trade equity curve.
• LOW-volatility short filter: it was deeply negative in development but positive out-of-sample. The reversal means volatility by itself is not a reliable exclusion.
• Trend Meter as a standalone gate: strict 4-of-4 confirmation helped shorts historically (+92.8/trade, PF 1.28) but hurt longs (−23.4/trade, PF 0.94). More confirmation is not universally better.
• Adaptive Trend State Engine alignment alone: aligned longs lost money while conflicting longs made money; the relationship is direction-dependent and not a standalone fix.

Working hypothesis

The 5-minute strategy appears to need expansion and follow-through after entry. It performs best when conditions support larger movement—particularly high volatility and negative GEX. Its flat curve arises when it continues taking short setups in more contained positive-GEX, normal-volatility structure, where early movement often reverses before the trade can mature.

This is an entry-selection problem first, with an exit-management component second. The path evidence does not support assuming that all early favorable movement should trigger a mechanical early exit: profitable trades in the weak regime often required an initial adverse phase before extending.

Risk / mitigation experiment

Keep the existing strategy as the control. Treat this as a shadow research layer:

• Flag, rather than automatically block, short entries when volatility is NORMAL and GEX is positive.
• Compare flagged trades against all other shorts separately in future out-of-sample observation.
• Track whether the failure signature remains: early Maximum Favorable Excursion (MFE), later Maximum Adverse Excursion (MAE), and negative peak capture.

Do not promote the rule yet. The validated sample is 212 trades across development and out-of-sample, but the overall strategy itself remains unstable out-of-sample.

Things to try next

In your strategy backtester, test these separately before combining them:

1. Exclude only short entries in NORMAL-volatility, positive-GEX conditions. Compare trade count, net P&L, profit factor, average trade, drawdown, and out-of-sample behavior against the unchanged control.
2. Add a short-entry follow-through rule rather than an immediate exit rule. Test a narrowly defined condition requiring the short setup to retain bearish continuation after its initial favorable move; do not use a generic tighter stop. Compare Maximum Adverse Excursion (MAE), Maximum Favorable Excursion (MFE), peak capture, average trade, and drawdown.
3. Test Trend Meter gating by side, not globally. Run strict 4-of-4 confirmation on shorts only, then separately test removing it from longs. Compare trade count, net P&L, profit factor, average trade, drawdown, and out-of-sample behavior.