Kalshi AI — Mispricing Audit Keel v4

Run date 2026-08-11 · Horizon 45 days (through 2026-09-25) · Capital $2,500 · Category focus: AI (advisory) · Universe: targets.json allowlist, 117 series
0
Picks
42
Rejects logged
$0
Deployed
100%
Cash ($2,500)
117
Series swept
Verdict: stand down — cash is the position this week. Every one of the 42 in-horizon markets across the full v4 allowlist failed the Stage-0 mechanical cuts or the v4 entry bands. The slate is uniformly deep NO-favorites (YES asks ≤5¢) sitting in zero- or near-zero-volume books. There is no market where a probability was worth forming: the three books with real volume are efficiently priced tails, and the two structurally playable prices sit in dead books below the hard liquidity floor. Zero picks is the correct v4 output for this slate, not a failure to find edge — 98 of 136 historical picks entered zero-volume books and that cohort had negative edge; this run declines to repeat it.

Why zero picks

The Stage-0 universe (allowlisted series with active markets closing by Sep 25) produced 42 markets: 22 Tier-1 institutional-process legs (almost all "X out / bill becomes law before Sep 1"), 13 Tier-2 legs (AI IPO dates, Tesla–SpaceX merger, US–Iran deal), and two central-bank decision slates (Bank of Canada Sep 2, Banxico Sep 24). Applying the hard cuts in order:

The AI-category Tier-1 series the run was pointed at — KXANTHROPICDOD (close 2028) and KXNYTOAI (close 2028) — have no legs inside the 45-day horizon and were not candidates. KXAGICO-COMP-26Q3 and the Oct-1 IPO legs close Oct 1, six days past the horizon cutoff, and were excluded mechanically.

The three books with real volume — all efficient

MarketYES bid/askVol 24hOIWhy it's not a pick
KXALITOOUT-26SEP01
departure/ouster Alito retires before September?
0.1¢ / 0.4¢106,999243,617 Alito publicly confirmed on Aug 7 he is staying for another term (UPI, ABC News), after June's retracted NPR retirement-story error. The 107k volume is the book digesting that statement. YES is a <40¢-band tail; NO at ~99.6¢ is a deep-favorite band reject.
KXUSAIRANAGREEMENT-27-26SEP
scheduled-process US–Iran deal before September?
1¢ / 2¢7,705884,209 NO at 98–99¢: max win ≤ 2% of capital at risk. The 94¢-lock memorial applies — deep-favorite band reject regardless of how safe the NO looks. Precedent store: Jun and Jul legs both resolved NO.
KXIPOANTHROPIC-DATE-26SEP01
corporate-event Anthropic announces IPO before Sep 1?
0¢ / 1¢2,460146,562 Same structure: 1¢ tail / 99¢+ favorite. No playable side under v4 bands. Jun, Jul, and Aug legs all resolved NO (API-verified).

Near-misses worth recording

Ken Martin (DNC chair) — KXMARTINDNCOUT-26MAY-26SEP01, YES 14/19¢. The one market in the slate with a mid-band price. It opened Jul 28 near 69¢ YES at the peak of the autopsy-report/debt/phone-throwing scandal cycle and has decayed to 14¢ as congressional resignation calls failed to move DNC members — the body that actually controls his tenure (Axios, CNN, Jul 28). Buying NO at 86¢ would sit in the thin-favorite band, and the precedent (AUG01 leg resolved NO) supports it — but the book trades 1 contract per day against a hard floor of 500, with 5 contracts visible at the bid. The depth cap (⅓ of visible book) would allow a ~$1.50 position. Liquidity-floor reject; the 69→14¢ move also trips the 48-hour price-move check. Logged so the shadow record can score it.

Banxico HOLD — KXCBDECISIONMEXICO-26SEP24-HOLD, 93/94¢. Tightest spread in the slate (1¢) and a scheduled-process archetype, but zero 24h volume and a 94¢ entry is a deep-favorite band reject (≥90¢). The precedent store holds 10 KXCBDECISIONCANADA rows — central-bank holds resolve mechanically, but 94¢ six weeks before the meeting pays 6¢ to carry six weeks of surprise risk.

Full reject log (42)

Side/price shown where a side was structurally considered; Stage-0 mechanical rejects carry no model probability by design (no probability was formed on any market this run — no Stage-2 survivors).

MarketArchetypeYES bid/askVol 24hOIReject reason
KXALITOOUT-26SEP01departure/ouster0.1 / 0.4106,999243,617deep-favorite band / tail; Aug 7 "staying" statement fully priced
KXUSAIRANAGREEMENT-27-26SEPscheduled-process1 / 27,705884,209deep-favorite band (NO 98–99¢)
KXIPOANTHROPIC-DATE-26SEP01corporate-event0 / 12,460146,562deep-favorite band / tail
KXLEAVEWALZ-26JUL-26SEP01departure/ouster1 / 236614,840liquidity floor (366 < 500); tail
KXCBDECISIONCANADA-26SEP-H0scheduled-process97 / 981272,336YES bid ≥ 95¢; liquidity floor
KXDEMSCHUMER-27-SEP01departure/ouster0 / 5117606liquidity floor; YES ask ≤ 5¢
KXKASHOUT-26APR-26SEP01departure/ouster4 / 510413,663liquidity floor; YES ask ≤ 5¢
KXIMPEACH-26-SEP01legal-action0 / 19463,391liquidity floor; tail
KXCOMPANYACTIONMERGER-27-26SEP01corporate-event2 / 380107,669liquidity floor; tail
KXTRUMPTHUNEOUT-27JAN01-SEPdeparture/ouster2 / 8233,065spread 6¢ > 5¢; liquidity floor
KXMARTINDNCOUT-26MAY-26SEP01departure/ouster14 / 1911,455liquidity floor (vol24h=1); 48h price-move flag (69→14¢)
KXBURNHAMOUT-26JUL-26SEP01departure/ouster0 / 107,993zero-volume book; tail
KXFARAGEOUTREFORM-26JUL-SEP01departure/ouster0 / 703,888zero-volume book; spread 7¢ > 5¢
KXFARMBILL-26MAY-SEP01confirmation/enact2 / 304,848zero-volume book; tail
KXGALLEGOOUT-26APR-SEP01departure/ouster1 / 307,025zero-volume book; tail
KXJUDGECOUNT-26AUG-1confirmation/enact0 / 103,831zero-volume book; tail
KXJUDGECOUNT-26AUG-3confirmation/enact0 / 20812zero-volume book; tail
KXJUDGECOUNT-26AUG-4confirmation/enact0 / 107,880zero-volume book; tail
KXLAKECONF-26MAY12-SEP01confirmation/enact0 / 1034,460zero-volume book; tail
KXLEAVELISACOOK-26SEP01departure/ouster2 / 305,091zero-volume book; tail
KXLEAVEMILLS-26NOV03-26SEP01departure/ouster6 / 100992zero-volume book; YES < 40¢ band / NO ≥ 90¢ band
KXLEAVEPOWELLGOV-26SEP01departure/ouster2 / 3010,071zero-volume book; tail
KXLUTNICKOUT-26SEP01departure/ouster1 / 6029,311zero-volume book
KXRECNCH-27-SEP01confirmation/enact0 / 209,710zero-volume book; tail
KXSANCTIONRUSSIA-26JULconfirmation/enact2 / 503,962zero-volume book; YES ask ≤ 5¢
KXTARIFFBILL-26JUL-26SEP01confirmation/enact0 / 102,213zero-volume book; tail
KXIPOOPENAI-26SEP01corporate-event0 / 10148,956zero-volume book; tail
KXIPOSTARLINK-26SEP01corporate-event0 / 1012,779zero-volume book; tail
KXIPOANDURIL-26SEP01corporate-event0 / 30256zero-volume book; tail
KXCBDECISIONCANADA-26SEP-C25scheduled-process0 / 40333zero-volume book; tail
KXCBDECISIONCANADA-26SEP-C26scheduled-process0 / 30303zero-volume book; tail
KXCBDECISIONCANADA-26SEP-H25scheduled-process0 / 401,582zero-volume book; tail
KXCBDECISIONCANADA-26SEP-H26scheduled-process0 / 40233zero-volume book; tail
KXCBDECISIONMEXICO-26SEP24-HOLDscheduled-process93 / 9405,286zero-volume book; deep-favorite band (≥90¢)
KXCBDECISIONMEXICO-26SEP24-C25scheduled-process3 / 403,685zero-volume book; tail
KXCBDECISIONMEXICO-26SEP24-C50scheduled-process0 / 204,644zero-volume book; tail
KXCBDECISIONMEXICO-26SEP24-C50Pscheduled-process0 / 204,645zero-volume book; tail
KXCBDECISIONMEXICO-26SEP24-H25Pscheduled-process0 / 103,596zero-volume book; tail
KXCBDECISIONMEXICO-26SEP24-H50scheduled-process0 / 103,395zero-volume book; tail
KXCBDECISIONMEXICO-26SEP24-H50Pscheduled-process0 / 103,395zero-volume book; tail
KXHEGSETHOUT-26APR-SEP01multi-news-otheralready an active pick
KXPIRROOUT-26AUG-26SEP01multi-news-otheralready an active pick

Portfolio

LineCost% of capital
Positions (0 picks, 0 clusters)$00%
Cash$2,500100%

Position-size, EV-for-the-record, cluster, and conviction-exposure tables are trivially empty this run. The v4 sizing regime treats 40–70% cash as normal in an efficient week; 100% cash is the extreme of the same principle when the slate offers no in-band, in-liquidity entry at all. Risk profile: zero market risk carried; the only exposure is the opportunity cost of the ~3 weeks until the Sep-1 legs roll and new monthly legs open with fresh books.

Execution notes

Precedents consulted

No Stage-2 candidates survived Stage 0, so no per-candidate precedent retrieval drove a probability this run. Store coverage checked: KXCBDECISIONCANADA (10 rows), KXUSAIRANAGREEMENT (2 — Jun and Jul legs, both NO), KXIPOANTHROPIC (2 — both NO), KXKASHOUT (2), KXHEGSETHOUT (2), KXALITOOUT (1), KXLEAVEWALZ (1) — all consistent with the deep-NO pricing observed; none contradicted a reject.

Sources