Kalshi Politics — Mispricing Audit (v4, archetype-targeted)

Report date 2026-09-16 · run at 2026-09-16T17:15Z · capital $2,500 · horizon 45 days · category focus Politics (advisory; all Tier-1/2 allowlisted series in scope) · methodology Keel v4

Summary

Picks1

Logged rejects18

Capital deployed$103.494.1% of $2,500

Cash held95.9%

An efficient week in the v4 universe. Nine markets reached Stage 2. One survives: NO on the Anthropic IPO "confirmed before Oct 17" rung, a corporate-event leg where Kalshi still carries headline residue that Polymarket, EDGAR and the Reuters timeline all contradict. The two other live stories, the Graham Russia-sanctions bill (House floor vote this afternoon) and the Bank of Canada October hike, are either fairly priced at 86¢ or dislocated only on the tail side of the 40¢ floor. Every other liquid market in the allowlist either closes after the horizon (49 markets, mostly Jan 1, 2027 rungs) or fails a mechanical cut.

Horizon note. Five liquid "before Nov 1" contracts close 5–22 hours past the strict 45-day cutoff (Nov 1 04:59Z / 15:00Z vs Oct 31 16:20Z). They were admitted with that tolerance stated; the strict slate would have been four Tier-2 markets only.

Stage 0 — Universe funnel

Step	Count
Open markets across the 120 allowlisted series (public API, Sep 16 16:20Z)	896
… closing inside the 45-day horizon (Oct 31) or on Nov 1 04:59–15:00Z (≤22h tolerance, 'before Nov 1' contracts)	—
… cut: volume24h < 500 (live)	—
… cut: spread > 5¢ / YES bid ≥ 95 / YES ask ≤ 5	—
… cut: inside 3h halt (FOMC dissent ladder, decision today)	5
… cut: close_time > expected_expiration + 1h (Hegseth ladder)	2
Stage-2 candidates researched	9
Picks	1

Cut reasons among the 896: 385 beyond-horizon & illiquid, 202 also wide-spread, 92 also ask ≤ 5; 49 liquid but beyond horizon (Hegseth, Fauci/Brennan charges, Kash Patel Dec/2027, Anthropic Nov/Dec/Jan rungs, KXBILLS, KXIMPEACH, Greenland); 11 in-horizon but volume < 500; 4 passed everything strictly (Anthropic OCT17/OCT24, BoC Oct H0/H25); 5 more admitted on the Nov-1 tolerance (Anthropic NOV01, Russia sanctions, tariff bill, Patel NOV01, Tesla/SpaceX NOV01). No Mentions markets appeared in the allowlisted series.

Picks

corporate-event · Tier 2BUY NOMEDIUMcluster: anthropic-ipo-timing

Anthropic officially confirms an IPO (S-1 effective, priced, or exchange ticker) before Oct 17, 2026?

KXIPOANTHROPIC-DATE-26OCT17 · closes 2026-10-17T03:59:00Z · expected expiration 2027-04-01 (early-close on the event)

Thesis. No public S-1 exists yet (EDGAR, Sep 16); Reuters has marketing starting mid-October at the earliest, so a priced deal by Oct 16 needs a compressed schedule the sources rule out. Polymarket prices trading-by-Oct-15 at 2%; Kalshi still carries 23¢ of headline residue from the Nasdaq-venue story.

Rules & resolver. Resolver: Kalshi, on three mechanical triggers in rules_secondary: SEC declares the S-1 effective, OR the IPO is priced, OR "a securities exchange has assigned a ticker". No dated clarification appended. The precedent store has ruled twice (KXIPOANTHROPIC-DATE-26JUL01; KXIPO-26-SPACEX) that a confidential filing, a proposed symbol in an S-1, or a venue-selection headline is not a trigger.

Evidence (primary). EDGAR full-text search on Sep 16: no Anthropic S-1 or S-1/A exists; only third-party filers mention the company. Reuters (via CNBC, Sep 5): public S-1 "late September", marketing "mid-October at the earliest", listing "days before" the Nov 3 midterms, gated on a $15bn revolver closing. SEC practice requires the public filing at least 15 days before the roadshow. Bloomberg (Sep 13): Nasdaq chosen — the headline that lifted this rung from 10–12¢ to 23¢ and an intraday 45¢ print on Sep 15. Polymarket (Sep 16): shares trading by Oct 15 at 2% ($114k volume); by Oct 31 at 53%.

Probability. Pre-market NO 88% → final 90% (the Polymarket twin and the EDGAR blank pull the number up; the 23/24 Kalshi book, liquid at 1.9k/24h, pulls it down a little). Resolver deductions already inside the number: −3 pts for an early "ticker assigned" reading if Nasdaq announces approval before pricing; −7 pts for an Arm-pace schedule (public S-1 by ~Sep 22 and pricing inside ~3 weeks).

Entry. YES 23 bid / 24 ask → NO ask 77¢ with only 15 contracts; 378 more at 79¢ (YES bid 21). Limit 79¢, 2¢ of slippage, 393 contracts fillable at or better. Edge 11¢, EV +13.9%. 48h price check: the YES side rose 12 → 23 on the venue headline, so this fades a non-event move rather than chasing one — the exact pattern the SEP01 precedent documented.

Size. MEDIUM cap 7% = $175 → 221 contracts; depth cap ⅓ × 393 = 131 contracts (binding). Cost $103.49 = max loss. Payout if NO: $27.51.

Single cleanest tail. Anthropic files the public S-1 this week and the banks run a compressed roadshow to price by Oct 16, ahead of the reported mid-October marketing start.

Precedents consulted: KXIPOANTHROPIC-DATE-26JUN01 (NO — "officially announce" is mechanical, not journalistic); KXIPOANTHROPIC-DATE-26JUL01 (NO — confidential S-1 press release made the contract cheaper); KXIPOANTHROPIC-DATE-26SEP01 (NO — clock the ladder off the PUBLIC S-1 date + ~3 weeks; near-rung residue is the dislocation); KXIPO-26-SPACEX (proposed symbol / venue headline ≠ ticker assignment).

Position sizes & limits

Ticker	Side	Limit	Contracts	Cost = max loss	Cost cap (tier)	Depth cap (⅓ book)	Binding
KXIPOANTHROPIC-DATE-26OCT17	NO	79¢	131	$103.49	221 ($175, MEDIUM 7%)	131 (393 fillable ≤79¢)	depth

Reserve: $2,396.51 (95.9%) held in cash — well above the 10% floor; 40–70% cash was the expected range and this week sits beyond it because the slate was thin, not because edge was skipped.

EV for the record

Ticker	Side @ limit	Model prob (side)	Edge ¢	EV %	Expected P&L on size
KXIPOANTHROPIC-DATE-26OCT17	NO @ 79	90%	11	+13.9%	+$14.41

Clusters (one leg per cluster)

Cluster	Deployed leg	Runner-ups (shadow-tracked rejects)
anthropic-ipo-timing	KXIPOANTHROPIC-DATE-26OCT17 NO	OCT24 NO (66¢, 78%); NOV01 NO (47¢, 50% — no edge)
boc-oct-2026	— none (both legs below the 40¢ floor)	H25 YES 35¢ (55%); H0 NO 39¢ (55%)
russia-sanctions-law	— none (no edge / deep favorite)	KXSANCTIONRUSSIA-26JUL-NOV YES 86¢ (82%); KXTARIFFBILL-26JUL-26NOV01 YES 93¢ (70%)
patel-fbi-exit	— none (deep favorite)	KXKASHOUT-26JUL-26NOV01 NO 92¢ (95%)
tesla-spacex-merger	— none (deep favorite)	KXCOMPANYACTIONMERGER-27-26NOV01 NO 95¢ (96%)

Conviction exposure

Conviction	Picks	Cost	% of capital
HIGH	0	$0.00	0.0%
MEDIUM	1	$103.49	4.1%
Cash	—	$2,396.51	95.9%

Risk profile

Single-position book. Worst case is a full loss of $103.49 (4.1% of capital) if Anthropic prices by Oct 16 or Kalshi reads an early exchange notice as a ticker assignment.
Correlation. None across picks (one pick). The rejected Anthropic runner-ups are the same bet and are not held.
Information decay. The public S-1 date is the next catalyst; if it lands before Sep 24 with an announced roadshow start before Oct 8, re-underwrite immediately.
Liquidity. 1.9k contracts/24h, 17.8k OI; the limit sits on a 378-contract YES bid, so a fill is realistic without moving the book.

Execution notes

Enter as a resting NO limit at 79¢ (equivalently sell YES to the 21¢ bid). Do not lift beyond 80¢; the edge is 10¢ there and the depth cap would allow 320 contracts but the slippage rule stops at 3¢.
If the YES side drifts back toward 12–15¢ (NO 85–88¢) before the fill, the thin-favorite band applies: half size (3% = $75) or stand aside.
Exit/trim trigger: a public S-1 with an explicit roadshow start before Oct 8, or any Nasdaq press release naming a ticker before pricing.

Rejected (logged)

Ticker	Archetype	Side	Price¢	Model %	Cluster	Reason
KXIPOANTHROPIC-DATE-26OCT24	corporate-event	NO	66	78	anthropic-ipo-timing	cluster runner-up: same thesis as the OCT17 leg with a weaker anchor (Polymarket has no Oct 23 rung; interpolated 15-20%); prob NO 75% pre-market → 78% final vs 66-67¢ ask, shadow-tracked
KXIPOANTHROPIC-DATE-26NOV01	corporate-event	NO	47	50	anthropic-ipo-timing	inside spread / no edge: Kalshi 53/54 equals Polymarket 53% for trading by Oct 31; prob 45% pre-market → 50% final; also 40-59¢ band requires HIGH + Tier-1
KXCBDECISIONCANADA-26OCT-H25	scheduled-process	YES	35	55	boc-oct-2026	entry band (<40¢): YES ask 35¢ vs 55% model (OIS 59-60% per LSEG/bankofcanadaodds Sep 14-16) is a real dislocation but sits in the tail band that is 0-for-8; logged for the record
KXCBDECISIONCANADA-26OCT-H0	scheduled-process	NO	39	55	boc-oct-2026	entry band (<40¢): NO ask 39¢; even at 40-59¢ the earned-favorites band needs HIGH conviction and a Tier-1 process archetype, and this is Tier-2 scheduled-process with economists split on the hike
KXSANCTIONRUSSIA-26JUL-NOV	confirmation/enact	YES	86	80	russia-sanctions-law	no edge: House final vote on S.5025 pending this afternoon (rule passed 214-211 Sep 15); 86¢ already prices ~90% passage × ~95% signature; prob 80% pre-market → 82% final; the NO side is a 15¢ tail. Closes 22h past the 45-day horizon (tolerance noted)
KXTARIFFBILL-26JUL-26NOV01	confirmation/enact	YES	93	70	russia-sanctions-law	deep-favorite band (≥90¢) plus resolver risk: the Graham Act only AUTHORIZES tariffs up to 500%/100% while the rule says 'imposing'; rules_secondary is empty, no clarification. NO at 10¢ is a tail; second leg of the Russia-bill cluster in any case
KXKASHOUT-26JUL-26NOV01	departure/ouster	NO	92	95	patel-fbi-exit	deep-favorite band (≥90¢): NO ask 92¢. Precedent KXKASHOUT-26APR-JUN01/JUL01 both NO — a principal fighting back (5-hour Senate hearing Sep 15) is not leaving; YES at 9¢ is a tail
KXCOMPANYACTIONMERGER-27-26NOV01	corporate-event	NO	95	96	tesla-spacex-merger	deep-favorite band (≥90¢): NO ask 95¢; Musk's All-In Summit 'who can imagine what action one might take' is explicitly excluded by rules_secondary (intent/hypothetical ≠ signed agreement)
KXFOMCDISSENTCOUNT-26SEP-0	multi-news-other	—	36	—	fomc-sep-dissents	inside 3h halt (decision today); mechanical cut, no side formed
KXHEGSETHOUT-26AUG-26NOV01	multi-news-other	—	9	—	hegseth-exit	close after expiration (close Nov 1 vs expected_expiration Oct 1); mechanical cut
KXHEGSETHOUT-26AUG-26DEC01	multi-news-other	—	27	—	hegseth-exit	close after expiration and beyond horizon; mechanical cut
KXHOUSEADJOURN-26-26SEP18	departure/ouster	—	98	—	house-recess	yes_bid ≥ 95¢; mechanical cut
KXCBDECISIONJAPAN-26SEP17-H25	scheduled-process	—	99	—	boj-sep-2026	yes_bid ≥ 95¢; mechanical cut
KXCBDECISIONMEXICO-26SEP24-HOLD	scheduled-process	—	96	—	banxico-sep-2026	volume24h 214 < 500 and yes_bid ≥ 95¢; mechanical cut
KXUSAIRANAGREEMENT-27-26OCT	scheduled-process	—	1	—	us-iran-deal	yes_ask ≤ 5¢; mechanical cut
KXKASHOUT-26JUL-26OCT01	departure/ouster	—	5	—	patel-fbi-exit	yes_ask ≤ 5¢; mechanical cut
KXNUMAPPROP-26OCT01-AL1	confirmation/enact	—	5	—	fy27-approps	yes_ask ≤ 5¢; mechanical cut
KXCRYPTOSTRUCTURE-26JAN-OCT	—	—	—	—	crypto-structure	already an active pick ($EXISTING_PICKS); also outside v4 universe

Notes on the live rejects

KXSANCTIONRUSSIA-26JUL-NOV — YES 85/86. Lindsey O. Graham Sanctioning Russia and Iran Act (S.5025): Senate 86-11 (Aug 7); House Rules cleared Sep 15 and the rule passed 214-211 with two Democrats crossing; final passage expected this afternoon (CNBC, Reuters via AOL, Sep 16). The House is voting on the Senate bill, so passage goes straight to the President, who "has said he backs the bill" (Reuters). Rules check: the bill authorizes new primary sanctions (Putin, officials, banks) and secondary tariff penalties on purchasers — fits clauses (i)/(ii); the NDAA-style exclusion does not apply. Pre-market YES 80% (0.90 pass × 0.95 sign/10-day × 0.97 presentment timing × 0.97 resolver) → 82% final. Against an 86¢ ask that is no edge; NO at 15¢ is a banned tail. Precedent consulted: KXBILLS-HOUS (presentment + 10 days enacts; a refused signature is timing, not risk), KXRECNCH ladder (leadership-promised dates slip; count session days).

KXCBDECISIONCANADA-26OCT-H25 / H0 — YES 32/35, 61/64. Sep 2 hold at 2.25% with Macklem "prepared to raise… more than one increase"; Aug CPI 3.0% (Sep 15); LSEG odds for an Oct 28 hike 60% (Sep 14), bankofcanadaodds 59/41 (Sep 16). Economists (CBC, Globe) say markets are ahead of themselves with core near 2%. Pre-market hike 55% → final 55%. The dislocation is real (≈20 pts vs OIS on the hike leg) but the only side with edge is under 40¢ on both legs, and the 40–59¢ band is closed to Tier-2 scheduled-process anyway. Precedents consulted: 11 KXCBDECISIONCANADA rows — "size it as yield, not a view"; "explicit two-sided risk language is a hold tell"; "a hawkish hold reprices the NEXT meeting's ladder". This is the first instance in the store where the hike rung carries live OIS premium; written back to the precedent store.

KXTARIFFBILL-26JUL-26NOV01 — YES 90/93. The same Graham Act is what the book is pricing: it authorizes tariffs of up to 500% on Russia and 100% on major buyers. The rule reads "legislation imposing new or increased tariffs"; rules_secondary is empty and no clarification has been appended, so whether discretionary authority counts is unresolved. Pre-market 70%. Deep-favorite band regardless; second leg of the Russia cluster.

KXKASHOUT-26JUL-26NOV01 — YES 8/9. Patel sat through a five-hour Senate Judiciary hearing on Sep 15 (Washington Post, CBS) defending hiring changes and declining to rule out agents at polls; no departure signal, no successor talk. Pre-market YES 5%. NO at 92¢ is the deep-favorite band. Precedents: KXKASHOUT-26APR-JUN01 and -JUL01 both NO — "a principal who is publicly fighting back is signalling he has not been told to go".

KXCOMPANYACTIONMERGER-27-26NOV01 — YES 5/8. Musk at the All-In Summit: "who can imagine what action one might take" — the rules_secondary explicitly excludes intent/hypotheticals. Pre-market YES 4%; NO at 95¢ is the deep-favorite band. 35k contracts traded today on the YES side collapsing from 12 to 5–8¢.

Precedents consulted

Store coverage queried: KXIPOANTHROPIC (3 rows), KXCBDECISIONCANADA (11), KXKASHOUT (2), archetype confirmation/enact (KXBILLS, KXAGCONF, KXRECNCH — 11 rows), corporate-event (KXIPO — 1). No rows for KXSANCTIONRUSSIA, KXTARIFFBILL or KXCOMPANYACTIONMERGER; cold derivation used for those. Two rows written back to /tmp/out/precedents.json (KXCBDECISIONCANADA-26SEP-H0 and KXIPOANTHROPIC-DATE-26SEP01) with facts verified this run.

Data & sources

Kalshi public API (markets, orderbook, candlesticks) read at 16:20–17:05Z on Sep 16, 2026; DB query API /schema read first; resolution_precedents queried by series and archetype.
EDGAR full-text search (efts.sec.gov), Sep 16 · CNBC/Reuters Sep 5 (Anthropic timeline) · Bloomberg Sep 13 (Nasdaq) · Polymarket "Anthropic IPO by __?" and "In which month" events, Sep 16.
CNBC Sep 16, Reuters via AOL Sep 11, The Hill Sep 15, Fox News Sep 14, Axios Sep 15 (Russia sanctions bill) · congress.gov S.5025.
BNN Bloomberg Sep 2, Yahoo Finance Canada/LSEG Sep 14, CBC and Globe and Mail Sep 15, bankofcanadaodds.com Sep 16 (Bank of Canada).
Washington Post, CBS News Sep 15 (Patel hearing) · Benzinga/Yahoo Finance Sep 2026 (Musk All-In Summit).

Existing pick honoured: KXCRYPTOSTRUCTURE-26JAN-OCT NO is not re-issued.