VolRocObservatory

Calibration

Kalshi \u00b7 Crypto \u00b7 15-minute
Layer A · immutable vintage

Mark date 2026-06-19 · sealed 2026-06-25T15:09:55Z UTC · frozen · write-once · citable by date

Brier

0.1819

lower is better

No-skill baseline

0.2498

climatology · p(1−p)

Coverage

73.6%

legs with a price trajectory

Scored legs

10,601

of 14,410 resolved

As of

2026-06-19

compute as-of date

Skill is the gap between the two leftmost facts: a Brier below the no-skill baseline is the only claim made here. No per-market call is derived above the fact line.

Reliability — measured

0.00 0.00 0.25 0.25 0.50 0.50 0.75 0.75 1.00 1.00 15-60m · pred 0.069 · obs 0.060 · n=714 15-60m · pred 0.201 · obs 0.191 · n=659 15-60m · pred 0.301 · obs 0.276 · n=714 15-60m · pred 0.386 · obs 0.342 · n=663 15-60m · pred 0.461 · obs 0.430 · n=646 15-60m · pred 0.528 · obs 0.483 · n=635 15-60m · pred 0.605 · obs 0.609 · n=706 15-60m · pred 0.690 · obs 0.668 · n=704 15-60m · pred 0.788 · obs 0.809 · n=724 15-60m · pred 0.923 · obs 0.920 · n=750 5-15m · pred 0.072 · obs 0.040 · n=378 5-15m · pred 0.203 · obs 0.204 · n=372 5-15m · pred 0.300 · obs 0.335 · n=418 5-15m · pred 0.386 · obs 0.399 · n=406 5-15m · pred 0.459 · obs 0.476 · n=370 5-15m · pred 0.530 · obs 0.521 · n=384 5-15m · pred 0.604 · obs 0.585 · n=381 5-15m · pred 0.686 · obs 0.714 · n=329 5-15m · pred 0.786 · obs 0.800 · n=340 5-15m · pred 0.919 · obs 0.932 · n=293 predicted (market price) realized frequency τ band 5-15m 15-60m

Each point is a (τ band × price bin) bucket, radius ∝n. On the dashed diagonal = perfectly calibrated. Above = the market under-priced the outcome; below = over-priced. Kalshi-native.

Calibration buckets — kept

τ bandprice binnpredictedrealizedgapbrier
5-15m0-0.143780.0720.040-0.03230.0386
5-15m0.14-0.253720.2030.204+0.00120.1611
5-15m0.25-0.344180.3000.335+0.03480.2244
5-15m0.34-0.424060.3860.399+0.01330.2385
5-15m0.42-0.493700.4590.476+0.01650.2480
5-15m0.49-0.563840.5300.521-0.00910.2500
5-15m0.56-0.643810.6040.585-0.01830.2437
5-15m0.64-0.733290.6860.714+0.02790.2038
5-15m0.73-0.843400.7860.800+0.01360.1598
5-15m0.84-0.9992930.9190.932+0.01250.0603
15-60m0-0.147140.0690.060-0.00860.0550
15-60m0.14-0.256590.2010.191-0.01020.1516
15-60m0.25-0.347140.3010.276-0.02560.1991
15-60m0.34-0.426630.3860.342-0.04360.2264
15-60m0.42-0.496460.4610.430-0.03050.2450
15-60m0.49-0.566350.5280.483-0.04490.2524
15-60m0.56-0.647060.6050.609+0.00410.2374
15-60m0.64-0.737040.6900.668-0.02220.2216
15-60m0.73-0.847240.7880.809+0.02140.1544
15-60m0.84-0.9997500.9230.920-0.00310.0711

Suppressed buckets — disclosed

11 bucket(s) held fewer than min_bucket_n=30 observations. They are omitted from the curve and the kept table — self-omitted, never plotted as fabricated reliability points. dash ≠ zero.

60-240m · 0.34-0.42 · n=160-240m · 0.42-0.49 · n=160-240m · 0.49-0.56 · n=160-240m · 0.73-0.84 · n=1240-1440m · 0.14-0.25 · n=1240-1440m · 0.25-0.34 · n=2240-1440m · 0.34-0.42 · n=1240-1440m · 0.42-0.49 · n=1240-1440m · 0.49-0.56 · n=2240-1440m · 0.56-0.64 · n=2240-1440m · 0.64-0.73 · n=2

Provenance — write-once

Content SHA-25647060163f97ae00167eaa5da188f66629fe5ffa299cb39309d8e337dbbb4c0c4
Outcome SHA-2567ba201d8bf5833b76156ade11dfe7d8b2d420cc4f7807bc3c9f3533f327bee90
Resolver records SHA-25614555b32f47951d651a4762f8212216b8b24b7af3cbbf0a3e9efb1f3324197a4
Settled yes / no7,043 / 7,367
Resolved legs14,410
Resolution basissettlement_ts
Settlement sourcehttps://api.elections.kalshi.com/trade-api/v2
Window2026-05-20T00:00:00Z → 2026-06-19
Resolver coderesolve_outcomes/v1.0.0
Compute versionv1
Resolver run ID20260619T152156Z
Backfilledyes
Sealed at2026-06-25T15:09:55Z

Methodology

A reliability measurement over 14,410 prediction-market legs, binned by time-to-resolution (τ, basis realized) and contemporaneous market price, then compared against realized settlement. The Brier score is the mean squared error of the market price as a probability; the no-skill baseline is climatology base_rate*(1-base_rate) = 0.2498. Lower Brier than baseline is the only skill claim made.

  • Venue-native. Kalshi only; not canonical-leg-bound and never blended with another venue.
  • Suppression. Buckets with fewer than min_bucket_n=30 observations are disclosed but omitted from the curve.
  • Coverage = 73.6%. The miss is capture density against a short market lifetime: a leg snapshotted outside its active window has no price trajectory to score. Per-family join rate: KXBTC15M 80.2%, KXDOGE15M 63.4%, KXETH15M 80.0%, KXSOL15M 73.8%, KXXRP15M 70.4%. The bias is published, not hidden.
  • Time-arrow. The Mark carries no as_of and only legs resolved on/before the as-of date — no record depends on its own future.

Archive — sealed vintages

Each dated page is a Layer-A immutable vintage, citable by its content SHA-256. This latest view (Layer C) tracks the newest row.

The archive is the product; this page is a projection. Calibration measures the price, not the platform. No call is made above the fact line.

About VolRoc Latest calibration Machine-readable Methodology API Observational market statistics. Not a broker.