GEX

GEX outcomes

Returns validation data comparing our pinning predictions against actual settlement prices. Each expiry event generates 4 rows across different lookback windows (1h, 6h, 12h, 24h before expiry). Contains the predicted magnetic strike, max pain, stability/persistence scores at prediction time, and the actual settlement outcome with distance metrics and accuracy classification. Enables hit rate analysis by stability bucket, dominance type, expiry type, and direct magnetic strike vs max pain comparison. This is the credibility foundation of the entire GEX system. Indefinite retention.

GET/v1/gex/outcomescoreproalpha

Query Parameters

coinstringrequired

Underlying asset. Required.

e.g.BTCETH

lookback_hoursintegeroptional

Filter by lookback window: 1, 6, 12, or 24.

e.g.246

expiry_typestringoptional

Filter: DAILY, WEEKLY, MONTHLY, QUARTERLY.

e.g.WEEKLY

outcome_accuracystringoptional

Filter: HIT_EXACT, HIT_CLOSE, HIT_ZONE, MISS_NEAR, MISS_FAR.

e.g.HIT_CLOSE

fromstringoptional

Expiries from this date (ISO 8601).

e.g.2026-02-01T00:00:00Z

tostringoptional

Expiries until this date (ISO 8601).

e.g.2026-03-10T00:00:00Z

limitintegeroptional

Max rows returned. Default 500. The effective ceiling is your tier row limit, see Tiers and Limits.

e.g.200

Response Schema

Fields marked pro require a Pro subscription. Fields marked alpha require Alpha.

Field	Type	Tier	Description
`coin`	`string`	core	Underlying asset.
`expiration_timestamp`	`datetime`	core	Option expiration time in UTC.
`lookback_hours`	`integer`	core	How far before expiry the prediction was captured: 1, 6, 12, or 24 hours. Shorter lookback = fresher prediction = should be more accurate.
`expiry_type`	`string`	core	Expiry classification: DAILY, WEEKLY, MONTHLY, QUARTERLY.
`prediction_timestamp`	`datetime`	core	When the prediction was captured (expiry minus lookback_hours).
`predicted_magnetic_strike`	`float`	core	Magnetic strike at prediction time.
`predicted_magnetic_gex_usd`	`float`	core	Net GEX at the magnetic strike at prediction time.
`predicted_concentration_pct`	`float`	core	Gamma concentration at prediction time. Higher concentration = stronger magnet.
`predicted_magnet_dominance`	`string`	core	DOMINANT, CONTESTED, or DISPERSED at prediction time.
`predicted_pinning_zone_lower`	`float`	core	Lower pinning zone bound at prediction time.
`predicted_pinning_zone_upper`	`float`	core	Upper pinning zone bound at prediction time.
`predicted_stability_score`	`integer`	alpha	Stability score (0-100) at prediction time.
`predicted_persistence_score`	`integer`	alpha	Persistence score (0-100) at prediction time.
`predicted_confidence`	`integer`	alpha	Prediction confidence (0-100) at prediction time.
`predicted_pin_score`	`float`	alpha	Pin score (0-100, uncalibrated) at prediction time.
`predicted_max_pain_strike`	`float`	core	Max pain strike at prediction time. The benchmark we compare against.
`spot_at_prediction`	`float`	core	Spot price when the prediction was captured.
`spot_at_expiry`	`float`	core	Actual spot price at expiry settlement. Source: deribit_spot_rates.
`distance_from_magnetic_pct`	`float`	core	Distance from settlement to predicted magnetic strike as pct. The core accuracy metric.
`distance_to_max_pain_pct`	`float`	core	Distance from settlement to max pain as pct. For direct comparison: was our magnet or max pain closer?
`closed_in_pinning_zone`	`boolean`	core	TRUE if settlement fell inside the predicted pinning zone.
`spot_at_prediction_distance_pct`	`float`	core	Distance from spot at prediction time to the magnetic strike. Shows how far price had to travel.
`price_converged_to_magnet`	`boolean`	core	TRUE if price moved closer to the magnet between prediction and expiry.
`magnet_closer_than_max_pain`	`boolean`	core	TRUE if our magnetic strike was closer to settlement than max pain. The head-to-head comparison.
`total_oi_at_expiry_usd`	`float`	core	Total OI in USD at expiry. Higher OI = stronger pinning mechanics.
`filled_at`	`datetime`	alpha	When this outcome row was computed.
`outcome_accuracy`	`string`	pro	How close did settlement land to our predicted magnetic strike? See detail ↓

Derived Fields

FieldTypeTiercollapse all

▶outcome_accuracystringprocollapse

How close did settlement land to our predicted magnetic strike?

Classification of prediction accuracy based on distance from settlement to magnetic strike. HIT_EXACT: within 1 pct (bull's eye). HIT_CLOSE: within 2 pct (strong hit). HIT_ZONE: inside the pinning zone but more than 2 pct from magnet. MISS_NEAR: within 5 pct (close miss). MISS_FAR: more than 5 pct away (clear miss). The 24h lookback window is the hardest test; 1h lookback is easiest.

Accuracy classification (d = distance from settlement to magnetic strike)

\text{outcome\_accuracy} = \begin{cases} \text{HIT\_EXACT} & \text{if } |d| \leq 1\% \\ \text{HIT\_CLOSE} & \text{if } |d| \leq 2\% \\ \text{HIT\_ZONE} & \text{if in zone AND } |d| > 2\% \\ \text{MISS\_NEAR} & \text{if } |d| \leq 5\% \\ \text{MISS\_FAR} & \text{if } |d| > 5\% \end{cases}

RangeHIT_EXACT, HIT_CLOSE, HIT_ZONE, MISS_NEAR, MISS_FAR

Suggested Calculations

Not included in the API response. Compute these client-side from the fields above. Formulas and context provided.

FieldTypeInputsexpand all

▶hit_rate_by_lookbackfloatclient-sideWhat is our hit rate? Compute it yourself.expand

What is our hit rate? Compute it yourself.

Group outcomes by lookback_hours, count HIT_EXACT + HIT_CLOSE as hits, divide by total. The 1h lookback should have the highest hit rate (freshest data). The 24h lookback is the hardest test and the most impressive when it hits. This is the most important number for evaluating our system. We expose it so you can verify it yourself.

Inputsoutcome_accuracylookback_hours

Hit rate per lookback window

\text{hit\_rate}(w) = \frac{\text{COUNT}(\text{HIT\_EXACT} + \text{HIT\_CLOSE})}{\text{COUNT}(*)} \quad \text{WHERE lookback\_hours} = w

▶magnet_vs_max_pain_win_ratefloatclient-sideHow often does our magnet beat max pain? The ultimate test.expand

How often does our magnet beat max pain? The ultimate test.

Percentage of expiries where our gamma-weighted magnetic strike was closer to settlement than the OI-weighted max pain. If this is consistently above 50 pct, we have a measurable edge over the industry standard. Compute it by lookback window. If we beat max pain at 24h lookback, the prediction has real lead time. All fields needed for this calculation are Pro tier.

Inputsmagnet_closer_than_max_painlookback_hours

Magnet vs max pain win rate

\text{win\_rate} = \frac{\text{COUNT}(\text{magnet\_closer} = \text{TRUE})}{\text{COUNT}(*)}

▶accuracy_by_expiry_typefloatclient-sideWhich expiry types pin best?expand

Which expiry types pin best?

Group by expiry_type and compute hit rates. WEEKLY and MONTHLY expiries typically have higher OI and stronger pinning mechanics than DAILY. QUARTERLY expiries are complex (multiple competing effects). This tells you which expiries to watch most closely with the pinning system.

Inputsoutcome_accuracyexpiry_type

▶convergence_analysisstringclient-sideDoes price actually move toward the magnet over time?expand

Does price actually move toward the magnet over time?

Filter to price_converged_to_magnet = TRUE, compute the percentage. If convergence rate is high (say > 60 pct), the magnet is genuinely attracting price. Cross-reference with spot_at_prediction_distance_pct: does price converge more reliably when it starts close (< 2 pct) vs far (> 5 pct)?

Inputsprice_converged_to_magnetspot_at_prediction_distance_pctlookback_hours

▶oi_vs_accuracystringclient-sideDo higher-OI expiries pin more reliably?expand

Do higher-OI expiries pin more reliably?

Bucket expiries by OI magnitude and compute hit rates per bucket. Higher OI should produce stronger pinning (more gamma = more dealer hedging). If hit rates do not improve with OI, something else dominates.

Inputstotal_oi_at_expiry_usdoutcome_accuracy

▶hit_rate_by_stability_bucketfloatclient-sideDo high-stability predictions hit more often?expand

Do high-stability predictions hit more often?

Bucket outcomes by predicted_stability_score (0-30 = low, 30-70 = medium, 70-100 = high) and compute hit rate per bucket. If stability above 70 has a meaningfully higher hit rate, the Stability Score is a validated signal and belongs in the calibrated Pin Score model.

Inputsoutcome_accuracypredicted_stability_score

▶hit_rate_by_dominancefloatclient-sideDo DOMINANT magnets pin more reliably than CONTESTED ones?expand

Do DOMINANT magnets pin more reliably than CONTESTED ones?

Group by predicted_magnet_dominance (DOMINANT, CONTESTED, DISPERSED) and compute hit rates. DOMINANT magnets should pin more reliably. This validates whether the dominance classification adds predictive power beyond raw concentration.

Inputsoutcome_accuracypredicted_magnet_dominance