The numbers behind the levels.

Gamma is an estimate

Open interest shows outstanding contracts. It does not identify their owners. A call-positive, put-negative sign convention is a model assumption, not a direct measurement of dealer inventory.

Three useful reference points

Call and put walls

Concentrations of modeled exposure across strikes. They can provide context, but do not guarantee support or resistance.

Gamma flip

A modeled price at which aggregate exposure changes sign. It may be absent or have multiple crossings.

Gamma at strike

The distribution of estimated exposure across strikes. This is distinct from a volume profile or an order book.

Every source has a clock

Underlying prices, candle bars and open interest can arrive at different times. Prism keeps each timestamp and delay visible. A fresh page does not make old market data current.

Simple and Pro use the same data

Simple reduces what is shown. Pro adds expiry breakdowns, units and model context. Changing the view does not change the calculation.

How the level scorecard is computed

The scorecard asks whether session price actually interacted with a published level. It is not a backtest and not a trade signal.

Tolerance

A test occurs when a candle’s high–low range intersects the band around a level: max(1 tick, 0.05% of the level price). The default tick is $0.01. Both the tick and the fraction are configurable. A close is a break only if it prints more than that tolerance beyond the level in the breakout direction: call wall above, put wall below, flip through to the other side of the session open.

Session-only

A level is scored only against candles from the same America/New_York cash session it was built for. Current-session levels are never scored against prior-session candles. Overlaying today's profile on an older session is current-GEX-on-past-price, not a historical scorecard. Only 09:30–16:00 ET bars at or after the latest level publication and model build time are eligible. Missing timestamps, mixed sessions and books built after the final candle are unavailable, so an end-of-day book cannot claim it held earlier that day.

Provisional (“so far”)

Every current-session outcome stays provisional and is labeled with the last candle time, for example “held so far (as of 15:55 ET)”. A completed prior day can drop that qualifier only when that session's own walls were published — this source does not currently publish prior-session call/put/flip levels.

This is not a backtest: it does not replay historical GEX books, does not simulate fills, and does not forecast the next move. Untested levels are counted separately from held and broke. A gap through can break a level without a range test. Touches count approaches separated by at least five non-testing bars. A flip uses the opening side of the eligible candle window; if that open is inside the tolerance band, a close outside either side breaks it. Demo examples use fictional candles without publication provenance and provide no evidence of market performance.

Simulated and hypothetical results

Any simulated, backtested or out-of-sample result GEXfreak shows carries the statement below.

These figures are gross of costs. Each playbook passed its gate on a Brier skill score against a base-rate forecast over VALIDATE 2025-H1, the same sessions it was picked on, so the figures are not held out. The other figures and ranges come from those same calls. None of this is trading profit: no fills, commissions or slippage are modeled.

Desk analysis is general, not tailored to you

Grokbot desk notes are drafted from the same market data and model output Prism displays, and they publish only after desk review. Every member who opens an instrument sees the same note.

Grokbot reads market data and GEXfreak model output only; it never receives your positions, orders or account and does not comment on your trades.