NQ Delta-Inertia Strategy

Win rates computed separately per tier from each tier's P&L stream. The same 250 entry signals fire across all three tiers — what changes is contract sizing and pyramid mechanics. P1 = Locked (74.0%) because pyramid layers only add to trades already winning at +50pt MFE. P3 drops to 73.6% because the +25pt layer occasionally flips a borderline winner into a small loss.

Executive Summary

Strategy fires a base entry from a delta-inertia signal, gated by CVD confluence and directional bias filters, and exits at structural targets (session high/low or prior 1-hour high/low). The same entry signal is used across all three sizing tiers — only trade-management and contract-scaling rules change.

Across 4 years (2022 bear → 2024 bull → 2025 strong bull → 2026 partial), every tier is profitable in every year tested. Recovery Factor ranges from 30.76 (Conservative) to 33.79 (Aggressive, using the conservative max-DD estimate). The Balanced tier is recommended as the cleanest implementation with strong risk-adjusted returns and manageable single-trade variance.

How the Indicator Looks in Practice

Simplified three-panel indicator layout as it would appear in ATAS/X. Signal bar in amber. Stop and target lines auto-draw on entry. CVD divergence marker triggers A+ sizing.

Top panel: price action with three levels — entry, stop (−100pt), and structural target (+200pt). Yellow line = VWAP. Middle panel: inertia oscillator; signal fires when the bar enters the green ±1.70–3.00 zone. Bottom panel: cumulative session delta confirming buyers are in control. Sizing rules (1× / 5× A+), pyramid adds, half_50 stop tightening, and reversal-add are applied via the rules below — kept off the chart for clarity.

Validation Scorecard

Approximation flag: pyramid contracts have BE stops that can exit independently of the base contract. The post-process simulator assumes shared exits, which slightly overstates pyramid wins. Conservative realized values: P1 ~$420K (vs computed $517K), P3 ~$540K (vs $693K).

Entry Conditions (all must be true)

• Inertia signal: 1.70 ≤ |inertia_signed| ≤ 3.00 — sign of inertia = trade direction. Range cap filters out exhaustion entries (|inertia| ≥ 2.5 was a net loser).
• CVD confluence: at least 2 of 4 CVDs (cvd_10, cvd_30, cvd_90, cvd_session) must agree in sign with trade direction.
• Long bias filter: close > VWAP AND pre-RTH delta > 0 at signal bar.
• Short bias filter: close < 5-day MA AND pre-RTH delta < 0 at signal bar.
• Timing: entry bar between 09:30 – 11:29 ET. Max 3 entries per session. Stop trading after first winning trade.

A+ Detection (sets sizing for Tier 2 and 3)

• A+ marker (price/CVD divergence):
  • Long: close just made new 10-bar HIGH AND cvd_session is BELOW 10-bar peak by ≥ 100 contracts
  • Short: close just made new 10-bar LOW AND cvd_session is ABOVE 10-bar trough by ≥ 100 contracts
  If A+ fires → size base = 5 contracts. Otherwise → size base = 1 contract.

Trade Management (shared across all tiers)

• Initial stop: 100 points (= 400 ticks = $2,000 risk per contract)
• Stop management: at +50 pt favorable progress, cut stop distance to 50 pt (half_50)
• Target type: structural — session_high for longs, prev_h_low for shorts
• Target bounds: minimum 30 pt, maximum 200 pt distance
• Slippage: 2 ticks (0.5 pt) round-trip per contract
• Commission: $9 round-trip per contract

How Trade Management & Sizing Stack Up

Each row is the cumulative effect of adding the next management/sizing rule. Recovery Factor column shows risk-adjusted return; bigger is better.

Per-Year Stability (every tier, every year)

* P3 max DD estimate: the simulator's computed P3 DD equals P1's (−$15,021) because the worst 2025 losing streak didn't include trades that reached the +25 pyramid threshold. In a future losing streak that DOES include +25-then-reverse trades, each one would add ≈ $2,500 to drawdown. We've inflated P3's DD to −$20,500 as a conservative max estimate (P1 DD + 2 such trades) so the headline Recovery Factor reflects realistic worst-case risk, not historical-best-case.

Trade Frequency & Composition

Slippage Stress Test

How total P&L holds up if real-life slippage is 2× or 4× the backtest baseline (0.5 pt round-trip = ~2 ticks).

Slippage adjustment: each contract pays an additional 0.5 pt ($10) at 1 pt RT, or 1.5 pt ($30) at 2 pt RT. All three tiers stay strongly profitable even at 4× backtest slippage.

Sample Trades — Manual Validation (20 trades)

All P&L shown is combined_pnl (base contract + any reversal-add or pyramid contracts). Entry is next bar open after signal. Stop = 100 pts from entry. Structural target = session_high (long) or prev_h_low (short), bounded 30–200 pt.

Column guide: Inertia = signed inertia score at signal bar (positive = long signal, negative = short signal). MFE = max favorable excursion (pts). MAE = max adverse excursion (pts). Pyr = pyramid layers fired. All losses cap at −100.5 pts (stop) or less; all winners cap near 200 pts (200 pt structural target).

Honest Caveats

Strategy Insights — What the Data Taught Us

1. Catch order flow EARLY, not after confirmation

Every "confirmation" filter we tested REDUCED win rate: visible bursts at entry, fast tape, daily MA aligned with trade direction, inertia magnitude in extreme range (≥2.5). The trades that win are the ones where signals are still in transition, not yet visibly confirmed.

2. Wait for the trade to prove itself before scaling

Adding more contracts AT entry (layered 2-contract base) or AT very early progress (+25 pt with BE stop) consistently lost money. Adding AFTER the trade has demonstrated momentum (+50 pt) or resilience (near-stop reversal) consistently wins.

3. Structural targets beat fixed targets

Using session_high (long) or prev_h_low (short) as the target — bounded between 30 pt min and 200 pt max — outperformed every fixed-target alternative. The market structure provides asymmetric reward sized to volatility.

4. Price/CVD divergence at entry is the strongest A+ marker

When price made a new 10-bar extreme but cumulative session delta did NOT confirm, win rate jumped from 66% (non-A+ trades) to 91% (train) / 82% (test) — combined ~88% across the full dataset. Counterintuitive but consistent: divergence = early momentum reset, not a contrarian signal.

Next Step — Indicator Visual Design

The strategy is locked. The next step is designing how it should LOOK on a live chart so a trader can read the conditions and execute consistently. Goals:

• Glanceable: trader sees in 2 seconds whether to trade and at what size.
• Not overlay-heavy: the chart shouldn't be cluttered. Maybe 2 main panels max.
• Stage-aware: indicator highlights the active management stage (initial, half_50 active, pyramid armed, reversal armed).
• Sizing tag: when a signal fires, indicator should clearly say "1x" or "5x A+" to avoid sizing mistakes in the heat of the moment.

Discussion to follow with concept sketches.

Glossary — Every Term Used in This Report