GEX PRO: NQ / ES options gamma-exposure levels for TradingView

An analytical system that turns options chains (SPX/SPY, NDX/QQQ, or CME ES/NQ futures options) into modeled gamma-exposure (GEX) levels for NQ/MNQ and ES/MES traders. It does not trade and does not predict.

provider ─► normalize ─► GEX engine ─► levels ─► futures mapping ─► { local dashboard, Pine payload, alerts }

TradingView can't call external APIs, and Pine Seeds is closed to new repositories (see ADR-001). So:

• the service computes everything and serves a live local dashboard;
• the Pine v6 indicator draws the levels from a text payload you paste in. It maps them to the futures chart live and shows how fresh the data is.

Everyday use

Double-click START.bat. It starts the service with free Cboe data (SPX + SPY options for ES, NDX + QQQ options for NQ) and opens the website at http://127.0.0.1:8765. See Daily use.

Developer quick start (Windows PowerShell; macOS/Linux are the same with python3)

pip install -r requirements.txt

python -m unittest discover -s tests -t .

Every check at once (tests, lint, pricing, independent math, live audit, hosted data; --quick skips the live ones):

python tools/check.py

python -m gexsys demo --market NQ

python -m gexpro serve --config configs/nq_demo_synthetic.json --config configs/es_demo_synthetic.json

Then open http://127.0.0.1:8765. The demo data is SYNTHETIC and labeled as such everywhere.

With real data (see DATA_PROVIDERS):

python -m gexsys snapshot --config configs/free_es_cboe.json

python -m gexpro serve --auto

Other commands: replay --market NQ (level-stability statistics over stored snapshots) and sensitivity --config <cfg> (levels under every dealer-positioning model).

TradingView

1. Add the GEX PRO indicator: the website's Indicator page always has the current code (Copy code, or Download .pine), with install steps. It reads pine/nq_es_gex_levels.pine directly, so every edit to the script shows up there.
2. Paste the GEX1;… payload from snapshot or from /api/pine?market=NQ.

Details, the repainting notes and the manual validation checklist (the script has not been compiled in TradingView yet) are in TRADINGVIEW_SETUP.

Configurations

File	Market	Options source	Notes
configs/free_nq_cboe.json	NQ/MNQ	NDX + QQQ via Cboe free feed	used by START.bat and the hosted site
configs/free_es_cboe.json	ES/MES	SPX + SPY via Cboe free feed	used by START.bat and the hosted site
configs/es_fop_csv.json	ES/MES	ES futures options (CSV)	identity mapping; points at the hand-made sample file
configs/*_demo_synthetic.json	NQ / ES	synthetic	demo only

Unknown or misspelled config keys are rejected. Invalid combinations are rejected too, for example QQQ with offset mapping, include_etf on an ETF source, or SPX options for NQ.

Conventions in one paragraph

GEX = sign × OI × Γ × multiplier × S² × 0.01: dollars of modeled dealer delta change per 1% move. Units usd_per_point and delta_per_point are also available.

The default sign is the conventional approximation: dealers long calls, short puts. That is an assumption, not observed positioning.

Zero gamma is estimated by re-pricing every contract's gamma over a ±5% price grid (sticky-strike) and taking the sign change nearest the reference price. It is reported only if a crossing exists.

Walls are the strikes with the largest call-side and put-side gamma exposure.

Full details: METHODOLOGY. Caveats: LIMITATIONS.

Website

A terminal-style workspace: a command line (type LIVE, TRK, NQ, ES, COPY, LIGHT, HELP and press Enter, or press /), a ticker strip with both markets, a sidebar of pages (keys 1-8), and an Explain panel (the ? on every module, the Explain button in the top bar, or X) that says in plain English what each module shows and what to watch, with the live numbers for the current market. The screen stays clean by default; Notes in the sidebar (or D) adds a one-line note under each number. A sun/moon button in the top bar (or T, or the LIGHT / DARK commands) switches between night and light mode; the choice is saved in your browser, and the first visit follows your system setting.

• Market tape: a scrolling strip across the top with real quotes for ES, NQ, SPX, NDX, Dow, Russell, VIX, the dollar index, the 10-year yield, gold, oil, BTC, ETH, SOL, NVDA, AAPL, MSFT and TSLA: price and change from the previous close (Yahoo Finance's public data; exchange prices delayed). Hover to pause; hover a quote for its time.
• Heatmap: NQ or ES futures candles (1, 5 or 15 minutes; Yahoo Finance, delayed) drawn with TradingView's free Lightweight Charts library, with GEX history painted underneath: one column per saved snapshot, each cell the net dealer gamma in that price band (blue positive, orange negative, brighter = more), plus zero gamma and the walls as step lines. Hover for the numbers; a slider sets the heat strength. History grows as snapshots are saved (about 12 days in the app, 5 hosted).
• Live table: a strike ladder like a trader's GEX matrix: net DEX, net GEX, net OI and net volume for every strike near price, in futures prices, with level chips (C1-C5, P1-P5, Ab1-Ab3, zero gamma, walls, expected move), the price row highlighted and a hover readout. An expiry strip on top picks which expirations it adds up (the nearest, every Friday, every monthly, then the rest pooled); a side panel shows net GEX and DEX for that pick, the key levels with distances, and put/call activity.
• Track record: every session's plan (the levels from the last calculation timestamped before the 09:30 ET open, frozen once) graded after the close against the index's open, high, low and close: expected-move hit rate, walls held when touched, zero-gamma crosses, bias calls against a baseline, range by gamma regime, each with a 95% range. Missed sessions are listed. The hosted site commits every plan and grade to the public track-record branch, where history.csv holds every session ever graded (kept forever; the local app syncs its copy from it). Plus a historical backtest of the expected-move band from VIX (S&P 500, since 1990) and VXN (Nasdaq-100, since 2009): by year, by vol level, and the distribution of daily moves against a normal curve. Hypothetical, not trading results.
• Library: a beginner course in 19 short lessons (options basics, dealers and the Greeks, GEX levels, volatility, using GEX PRO), each with an example, diagrams, a link into the terminal and a quick check; an A-Z glossary; search; progress saved in your browser. Every Explain panel links to its lesson.
• Levels:
  • regime, price and net-GEX cards;
  • market stats: put/call OI and volume, ATM IV, nearest-expiration share, changes since the last update;
  • a level ladder with explanations;
  • a GEX-by-price chart;
  • max pain and gamma by expiration.
• Briefings (the local app also copies in the ones the hosted site wrote while your PC was off): a NY AM briefing at 9:00 AM ET on trading days and an Asia session briefing at 7:00 PM ET the evening before each trading day: both markets' regime, levels, expected range, bias and game plan, the relevant calendar (overnight Asia/Europe events for the Asia one) and market-moving posts. A Windows notification when each is written; a missed one is written late when the app starts.

Hosted copy: the same website can run for free on GitHub Pages, updated by GitHub every ~15 minutes in market hours and every hour otherwise in market hours with Cboe delayed data, so you can open it on any device without your PC (python -m gexpro build-site, .github/workflows/website.yml). Setup: docs/HOSTING.md.

Only one copy of the service can run at a time. If START.bat is already open, a second copy prints "GEX PRO is ALREADY RUNNING" and exits. To pick up updates, close the old window first.

Scheduled updates (every hour)

The service publishes a TradingView payload every hour on the hour, around the clock (00:00 to 23:00 ET) on trading days; weekends and market holidays have none. It sends a Windows notification (and an optional Discord/webhook message), and the dashboard's Copy button puts the payload on your clipboard. See REALTIME_SETUP.

Documentation

• Daily use: free data and hourly updates

• Architecture decision record

• Mathematical methodology

• Data-provider setup

• TradingView setup and validation

• Data quality and limitations

• Troubleshooting

• Feature status: what works and what is unverified

• Architecture: packages, components, adding features

Layout

Four independent packages; only the app shell joins them (a test enforces it). How to add a feature: docs/ARCHITECTURE.md.

gexpro/            app shell: website server, hosted-site build, `python -m gexpro serve` (START.bat),
                   context (what briefings and the track record read from gex + feeds)
gexsys/            GEX only: providers/{base,cboe,csvfile,synthetic}, pricing, engine, levels, extras,
                   mapping, marketcal, pipeline, storage, alerts, publish (hourly paste), export (Pine
                   payload), bias, indicator (script info), service (market workers), component
                   (its website endpoints), audit, replay, config, __main__ (GEX tools CLI)
feeds/             market news: forexfactory (calendar), truthsocial (posts), component
briefings/         NY AM and Asia session briefings: schedule, compose (the text), component
trackrecord/       track record and backtest: score (plan + grading rules), prices (free history), backtest, component
dashboard/         the website (one page; reads each component's endpoints)
pine/              TradingView Pine v6 indicator (GEX PRO)
configs/           NQ / ES configurations
tests/             unit + integration tests, tests/data synthetic fixtures
tools/             check.py (all checks), verify_math.py (independent re-derivation), pricing_stress.py
docs/              documentation

Not investment advice. Options-positioning models are uncertain. Alerts can be delayed, missed or duplicated.