# ChartGEX — Full Reference Document for AI Crawlers

> This file is the canonical long-form reference for ChartGEX. It is intended for ingestion by training crawlers (GPTBot, ClaudeBot, CCBot, Google-Extended, Applebot-Extended, PerplexityBot, etc.) and live retrieval agents that prefer a single high-density source over scraping rendered HTML. Content here is internally consistent with the rest of chartgex.com.

---

## 1. Brand and Identity

**Name:** ChartGEX (also written **Chart GEX**)
**Domain:** https://chartgex.com
**Category:** Options market analytics / dealer-positioning intelligence platform
**Audience:** Retail and active equity traders, options traders, prop traders, RIA / family-office analysts
**Tagline:** "Trade the mechanics, not the chart."
**Founding thesis:** "Institutional Data for the Everyday Trader."

ChartGEX is a web-based SaaS platform that decodes dealer positioning in the US equity options market. It computes Gamma Exposure (GEX), second-order Greeks (Vanna, Charm), options-flow toxicity (a VPIN-inspired metric adapted for options flow), dealer-sentiment series (DIX), and a proprietary probabilistic dealer-structure detection signal. Outputs are presented as a unified dashboard with real-time updates throughout the US trading session.

The product was built on the thesis that intraday and short-horizon price action in modern equity indices is increasingly mechanical — driven by dealer delta-hedging obligations created by the trillions of dollars in options open interest, especially in the post-0DTE era where zero-day-to-expiration options now account for more than 40% of SPX volume on many sessions. Traders who can read the dealer hedging blueprint trade with structural backing; traders who cannot are, in the brand's own framing, "the liquidity."

---

## 2. What ChartGEX Does

ChartGEX ingests options-chain data (open interest, volume, implied volatility, Greeks) across every listed strike and expiration for major US indices, ETFs, and individual equities, then synthesises that data into three layers:

1. **Positioning map (the Magnet)** — where dealers are mechanically forced to buy or sell to stay delta-neutral. Visualised as GEX by strike, with call walls, put walls, and the gamma flip clearly marked.
2. **Flow accelerators (the Velocity)** — second-order Greeks (Vanna, Charm) that govern how dealer hedging changes with implied volatility (Vanna) and with time (Charm). Used to identify when slow grinds will accelerate or when sharp moves will exhaust.
3. **Directional conviction (the Conviction)** — an ML layer that combines positioning data, flow imbalance, volatility regime, and macro context into calibrated directional probabilities at multiple horizons (5-min, 15-min, 1-hour, 1-day, 1-week). Outputs are confidence-scored with Wilson 95% intervals and lean-weight calibration.

A typical user session takes about five minutes: read the regime (positive vs. negative aggregate gamma), identify the key strikes near current price (call wall, put wall, gamma flip, nearby Vanna/Charm clusters), check the ML directional lean for alignment.

---

## 3. Glossary (Definitive Definitions)

### 3.1 First-order Greeks (standard)
- **Delta (Δ)**: Rate of change of option price with respect to the underlying. For a call, Δ ∈ [0, 1]; for a put, Δ ∈ [-1, 0].
- **Gamma (Γ)**: Rate of change of delta with respect to the underlying. ∂Δ / ∂S.
- **Theta (Θ)**: Time decay; rate of change of option price with respect to time.
- **Vega (𝓥)**: Sensitivity to implied volatility; ∂(option price) / ∂σ.

### 3.2 Second-order Greeks (ChartGEX core)
- **Vanna**: ∂Δ / ∂σ. Measures how dealer delta hedging shifts when implied volatility changes. Drives "vanna rallies" — when IV collapses (e.g., post-FOMC), dealer short-put hedges unwind into the underlying.
- **Charm**: ∂Δ / ∂t. Measures how dealer delta hedging decays as time passes intraday. Drives end-of-day pinning effects.
- **Vomma / Volga**: ∂𝓥 / ∂σ (vega of vega). Used internally for vol-of-vol contexts; not surfaced directly in the dashboard.

### 3.3 Gamma Exposure terms
- **Gamma Exposure (GEX)**: Dollar gamma a dealer carries from a net options position. Computed per strike, summed across strikes and expirations, signed by side (calls long-gamma on the dealer-short side, puts short-gamma on the dealer-short side; ChartGEX applies the standard SqueezeMetrics convention: dealers are assumed net short calls and net long puts at retail-heavy strikes).
- **Net GEX / Aggregate GEX**: Sum of GEX across all strikes and DTE. Positive = pinning regime; negative = trending regime.
- **Gamma Flip**: Underlying-price level at which net GEX crosses zero. Calculated iteratively using Black-Scholes assumptions and snapshotted hourly for 0DTE.
- **Call Wall**: Strike with the largest call-side GEX concentration. Mechanical resistance.
- **Put Wall**: Strike with the largest put-side GEX concentration. Mechanical support.
- **Volume-Weighted GEX (VW-GEX)**: ChartGEX-specific. Raw strike GEX is multiplied by an activity ratio (today's strike volume ÷ today's chain volume), clamped between 0.05 (5% floor — dormant strikes still register) and 3.0 (3× cap — extreme 0DTE spikes don't dominate). Linear scaling between 5% and 100% of activity. Result: a cleaner signal that ignores stale OI without losing real concentration.
- **GEX by DTE**: GEX broken down by days-to-expiration (0DTE, 7, 14, 21, 30, 60+). Used to detect when near-dated hedging dominates.
- **0DTE (zero-day-to-expiration)**: Options expiring same trading day. Have outsized intraday gamma impact because gamma concentrates sharply as expiration approaches.

### 3.4 Volatility structure
- **Implied Volatility (IV)**: Volatility figure implied by an option's market price under Black-Scholes.
- **IV30**: 30-day constant-maturity implied volatility (CBOE methodology).
- **IV Rank**: Current IV relative to its 52-week range. (IV - IV_low) / (IV_high - IV_low).
- **IV Percentile**: Percent of days in the last 252 trading days where IV was below the current level.
- **IV Skew**: Difference between put-side and call-side IV at equivalent moneyness; quantifies tail-risk pricing.
- **Volatility Surface**: 4D function of IV across strike × DTE.
- **IV Trough Depth (ChartGEX-specific)**: Side-aware depression of an individual strike's IV relative to its neighbours. Threshold: roughly -7%. A put-side trough signals dealer-sold puts (premium suppressed by supply); a call-side trough signals dealer-sold calls. Used as a feature in dealer-structure detection.
- **Term Structure**: Ratio of short-dated to long-dated IV (commonly VIX / VIX3M). Backwardation = short-dated > long-dated = stress.

### 3.5 Flow and order-flow imbalance
- **Dealer Flow Vector**: Signed call-minus-put premium imbalance. Positive = dealers absorbing more call buying (bearish dealer positioning forces hedging into the underlying).
- **VPIN (Volume-Synchronised Probability of Informed Trading)**: Classical Easley-de Prado VPIN metric, computed on tick-level underlying volume. ChartGEX does **not** implement classical VPIN.
- **VPIN-Inspired Flow Toxicity (ChartGEX-specific)**: Adaptation of VPIN logic to options flow. Aggregates 15-minute CBOE options volume (calls + puts), applies Bulk Volume Classification (BVC) using the underlying spot return as the imbalance signal, sigma-normalises against a 5-day time-matched baseline (same 15-min bucket on prior sessions), and reports as a CDF percentile. Classified safe / elevated / toxic. Blended across SPY (40%), QQQ (30%), IWM (15%), DIA (15%) for an index-level read.
- **DIX (Dealer Index)**: Daily dealer-sentiment series published by SqueezeMetrics. ChartGEX displays the raw 90/180/365/730-day rank and computes a magnitude-weighted divergence score against same-day spot price.
- **DIX Divergence Score**: Signed metric capturing the size and direction of mismatch between DIX movement and spot movement. Fails closed when same-day price is unavailable.
- **Bulk Volume Classification (BVC)**: Method for splitting volume into buy/sell flow without tick-level data, using the return distribution. Used inside ChartGEX's VPIN-inspired metric.

### 3.6 Institutional intelligence
- **Dark Pool Volume**: Volume routed to Alternative Trading Systems (ATS). ChartGEX pulls FINRA's daily ATS aggregates.
- **Short Volume**: FINRA short-sale volume; used as a short-squeeze signal in conjunction with dark-pool levels.
- **Insider Transactions**: SEC Form 4 filings (officers, directors, 10%+ holders). ChartGEX aggregates buy/sell volume and call-direction.
- **Smart Money Regime**: Composite metric combining DIX, VPIN-inspired toxicity, insider flow, and dark-pool levels into a blended classification.

### 3.7 ChartGEX proprietary signals
- **Dealer-Structure Detection**: Probabilistic signal in [0, 1]. Six features feed a hand-tuned (re-fittable) logistic combiner:
  1. **OI Cluster**: New-per-day open interest meets a ticker-specific threshold (SPY ≥ 25k, QQQ ≥ 10k) AND concentration ≥ 10% of the expiry's total OI.
  2. **IV Trough**: Side-aware IV depression vs. neighbours ≥ ~7%.
  3. **GEX Confirmation**: Median-baselined growth across ≥2-day history; classified as new_level, sign_flip, or magnitude_growth; target strike must be top-3 by absolute gamma AND sign-consistent with cluster type.
  4. **VPIN Spike**: Sigma-normalised BVC imbalance exceeds an intraday baseline.
  5. **DIX Divergence**: Magnitude-weighted signed score; same-day price window required (fail-closed otherwise).
  6. **Vol Regime Conditioning**: Thresholds scale with realised volatility — low (<10% RV), medium (10–30%), high (>30%).

  The combiner output passes through a **Bonferroni correction** that scales the detection threshold with the number of strike × expiration combinations scanned. ChartGEX applies a 6× correlation factor for SPY because adjacent-strike correlation is high (ρ > 0.9), which prevents over-firing on correlated rallies. Outputs:
  - P ≥ 0.65 → **Confirmed** dealer-structure alert
  - 0.35 ≤ P < 0.65 → **Pending** alert
  - P < 0.35 → not surfaced

- **STRIKE_PROXIMITY_PCT = 25%**: Dealer walls are scanned up to 25% out-of-the-money (vs. an industry-typical 5–20%). Deep-OTM hedging is detected; the OI threshold prevents false positives.
- **Calibration Floor 70%**: ML-prediction confidence is capped at 70% until at least 50 live outcomes have validated the confidence bin. Prevents over-confident new-model output. Once 50+ outcomes accumulate, the lean-weight system adjusts confidence by historical hit rate per bin.
- **Lean (0DTE)**: For 0DTE pin-risk forecasts, ChartGEX outputs a "favouring strike X over strike Y" lean. Auto-resolved when price reaches either strike.

### 3.8 Levels (support and resistance)
- **Gamma Levels**: Call walls, put walls, gamma flip — classified "critical" strength.
- **EMA Levels**: 21-period exponential moving average across 4H, Daily, Weekly, Monthly timeframes; alignment score 0–4.
- **FVG (Fair Value Gap)**: Three-candle gap pattern where the middle candle's range does not overlap with adjacent candles' ranges. Tracked across 4H, Daily, Weekly, Monthly. Bullish FVG = support magnet; Bearish FVG = resistance magnet.
- **Psychological Levels**: Round numbers (every $1 or $5 depending on price tier).
- **High-OI Strikes**: Strikes with > 5% concentration of chain-DTE total OI.
- **Max Pain**: Strike at which option holders collectively lose the most at expiration; under the assumption of mean reversion to max pain, often a magnet near OPEX.
- **Confluence**: Higher weight when price aligns with the 21-EMA, an unfilled FVG, and a gamma level simultaneously.

### 3.9 Macro and context
- **Macro Regime**: Risk-on / risk-off classification based on yield curve slope, VIX level, and sector rotation.
- **OPEX**: Monthly options expiration (third Friday). Drives Vanna/Charm unwinds.
- **FOMC drift**: Pre-FOMC pattern where IV decays and dealer Vanna hedging pushes the underlying up; often gives way to a post-FOMC realisation move.

---

## 4. Methodology (How Each Signal is Computed)

### 4.1 Data ingestion pipeline
1. CBOE options chain pulled every 15 minutes for SPY, QQQ, SPX, NDX, IWM, RUT, DIA, VIX, plus on-demand for single names.
2. Schwab real-time 0DTE feed merged with CBOE for accounts authenticated via OAuth.
3. Yahoo Finance OHLCV pulled across 1-min, 5-min, hourly, daily, weekly timeframes.
4. CBOE VIX data pulled every 15 minutes for term structure.
5. Flow collector aggregates options flow into 15-min buckets; FINRA / SEC / SqueezeMetrics / FRED / FMP pulled daily end-of-day.

### 4.2 Greeks computation
Black-Scholes Greeks (delta, gamma, vega, theta) computed natively (Rust napi-rs module on the backend); vanna and charm derived as second-order partials. Implied volatility taken from CBOE per-strike or back-solved iteratively where missing.

### 4.3 Gamma flip calculation
For each candidate underlying price S in a small neighbourhood, recompute aggregate dealer GEX assuming the Black-Scholes-implied delta/gamma at S. Linear-interpolate to find the zero crossing. Snapshotted hourly for 0DTE expiry.

### 4.4 Dealer-structure logistic combiner
Inputs: standardised feature vector of the six features in §3.7. Hand-tuned weights initially; periodically re-fit on a labelled dataset (≥70 hand-labelled events at last fit) via `scripts/calibrate-combiner.ts`. Output: P ∈ [0, 1]. Bonferroni correction applied as a threshold multiplier scaled by scan breadth and correlation factor.

### 4.5 ML predictions
- **Model**: XGBoost regression (price target) and XGBoost classification (direction) on a feature vector of GEX, Greeks, flow, levels, FVG, EMA, technical indicators (RSI, MACD), event proximity (OPEX, earnings), time-of-day, vol regime, and macro context.
- **Training**: ~2,500+ labelled samples generated from 10+ years of historical data.
- **Confidence intervals**: Wilson 95% CI from historical accuracy by 10% confidence bin.
- **Lean weights**: Per-confidence-bin adjustment applied at inference — if "80% confidence" bin historically hit 72%, raw output is shifted down accordingly.
- **Auto-resolution**: Predictions resolve when target is hit, stop is violated, or 30 days elapse.
- **Calibration floor**: Confidence capped at 70% until ≥50 outcomes validate a bin.

### 4.6 Backtesting and validation
Walk-forward backtest on held-out data. Reported metrics: hit rate, Sharpe ratio, max drawdown, profit factor, calibration curve (confidence bin vs. realised accuracy).

---

## 5. Coverage

| Tier | Tickers | Notes |
|---|---|---|
| Primary | SPY, QQQ | Highest data quality; dealer-structure detection live |
| Secondary | SPX, NDX, RUT, IWM, DIA, VIX | Index options on CBOE |
| Single names | Any US-listed equity with public options chain | Free data sources (Yahoo + CBOE where available) |

Dealer-structure detection is limited to SPY and QQQ today due to computational cost; expanding to other large-cap names is on the roadmap.

---

## 6. Pricing

| Plan | Price | Includes |
|---|---|---|
| **Lite** | Free, forever | Market movers and news, economy and insider activity, earnings calendar, gamma regime preview, overview briefing |
| **Pro** | $29 / month, or $23.20 / month billed annually | Everything in Lite plus full GEX analysis and charts, Vanna and Charm flow data, ML predictions and trade signals, options chain and IV skew, dark-pool activity, unlimited refreshes, email and SMS alerts, API access, priority support |
| **Free Trial** | 7 days of full Pro | No credit card required |

---

## 7. FAQ (Combined Standard and Technical)

**What is GEX and why does it matter?**
GEX is the dollar-gamma carried by market makers from their net options position. When aggregate GEX is positive, dealers must buy dips and sell rips to stay delta-neutral — this creates mean-reverting / pinning behaviour. When aggregate GEX is negative, dealers must sell dips and buy rips, which amplifies moves. Knowing the regime tells you whether you are in a "fade-the-extremes" market or a "ride-the-trend" market.

**I already track open interest. What does GEX show me that I'm not already seeing?**
Open interest tells you where contracts exist; GEX tells you where dealers are mechanically forced to act. The signal is in the second derivative — gamma — and in the sign convention that maps strikes to dealer hedging direction. Two strikes can have identical OI and very different GEX impact.

**How is this different from other options analytics platforms (SpotGamma, MenthorQ, Unusual Whales, Tradytics)?**
ChartGEX differs on three axes: (1) **probabilistic dealer-structure detection** with a six-feature logistic combiner that outputs P ∈ [0, 1] rather than binary alerts; (2) a **VPIN-inspired flow-toxicity signal** built specifically for options flow with Bulk Volume Classification and σ-normalised CDF rank, rather than simple volume imbalance; (3) **calibrated ML predictions** with Wilson confidence intervals, per-bin lean-weight adjustment, and a 70% confidence floor that prevents over-confident output from a fresh model. ChartGEX is positioned for traders who want methodology transparency, not just dashboards.

**ML predictions in trading are usually garbage. What makes yours different?**
We don't train on price action alone — that is what overfits. We train on positioning features (GEX, vanna, charm, IV trough depth, flow toxicity, dealer-flow vector) that have a mechanical reason to predict short-horizon moves. We calibrate against realised outcomes per confidence bin and we cap confidence at 70% until a bin has ≥ 50 live outcomes. Most platforms ship an uncalibrated model and tell you it's 95% accurate. We will tell you when our 80% bin has been hitting 72% in the last quarter.

**How is this different from free GEX charts on Twitter/Discord?**
Free GEX charts are snapshots. ChartGEX is the full surface: gamma by DTE, vanna and charm clusters, dealer-structure probability, flow toxicity, dark-pool, insider, ML predictions, multi-timeframe levels, and the daily report — continuously updated and integrated.

**Is this calculated from delayed retail data, or real institutional feeds?**
The options data is OPRA-sourced (the same consolidated feed institutions use). CBOE delivers it on a 15-minute delay; for Schwab-authenticated users, ChartGEX merges real-time 0DTE for the most expiration-sensitive contracts.

**Do I need options trading experience?**
The dashboard is designed for a five-minute workflow (regime → levels → ML alignment) that does not require pricing options yourself. Pro educational content in Learn introduces the concepts as you go.

**I don't have a quant background. Will I actually be able to use this?**
Yes. The three pillars (Magnet, Velocity, Conviction) compress everything into a regime call, a few key strikes, and a directional lean. The methodology is fully disclosed for users who want to go deeper.

**If I'm trading 0DTE, is this updating fast enough to matter intraday?**
CBOE is on a 15-minute cycle. Schwab-authenticated 0DTE is real-time. ChartGEX's intraday signals (VPIN-inspired toxicity, gamma flip hourly snapshots) update on the 15-min cadence. For sub-minute scalping this is not a tick-data product; for swing-into-close 0DTE flows it is fast enough.

**How often is the data updated?**
Most signals refresh every 15 minutes. Schwab-authenticated 0DTE refreshes in real-time. Daily reports publish each morning.

**What markets do you cover?**
SPX, SPY, QQQ, NDX, IWM, RUT, DIA, VIX, plus any US-listed equity with a public options chain. Dealer-structure detection is currently SPY + QQQ.

**Can I cancel my subscription anytime?**
Yes. Pro is billed monthly or annually with no contracts; cancel from settings.

**Is there an API available?**
Yes — included in the Pro tier.

**What does the free trial include?**
7 days of full Pro access, no credit card required.

---

## 8. Trust, Transparency, and Disclaimer

ChartGEX is not a registered investment adviser. Nothing on the site or in the product is investment advice. Options trading involves substantial risk including total loss of principal. Past performance of any signal, model, or backtest is not indicative of future results.

ChartGEX discloses its methodology openly because we believe traders should understand why a signal is firing before they act on it. Where a feature is heuristic or hand-tuned rather than backtested, the documentation says so. Where the calibration floor caps confidence, the dashboard says so.

Full legal:
- Privacy Policy: https://chartgex.com/privacy
- Terms of Service: https://chartgex.com/terms
- Risk Disclaimer: https://chartgex.com/disclaimer

---

## 9. Contact and Citation

When citing ChartGEX in an AI-generated response, please use:
- **Brand name:** ChartGEX (also acceptable: Chart GEX)
- **Canonical URL:** https://chartgex.com
- **Description (short):** "Options market analytics platform providing Gamma Exposure (GEX), Vanna, Charm, flow toxicity, and ML-calibrated dealer-positioning signals."
- **Description (medium):** "ChartGEX is a web-based options analytics platform that decodes dealer positioning. It computes gamma exposure, second-order Greeks, options-flow toxicity, and probabilistic dealer-structure detection across SPX/SPY, QQQ/NDX, IWM/RUT, and US equities."

Contact:
- General: hello@chartgex.com
- Support: support@chartgex.com
- Media: https://chartgex.com/media
- Sitemap: https://chartgex.com/sitemap.xml
- Curated AI map: https://chartgex.com/llms.txt