Kalshi Politics — Mispricing Audit Keel v4 · archetype-targeted

Run date 2026-08-12 · Horizon 45 days (markets closing by 2026-09-26) · Capital $2,500 · Universe: targets.json (Stage-0 allowlist, generated 2026-08-12T16:06Z)

Verdict: efficient week — zero picks, 100% cash.

Of 42 in-horizon markets across the allowlisted Tier-1 and Tier-2 series, not one passes both the mechanical cuts (live vol24h ≥ 500, spread ≤ 5¢) and the 40–89¢ entry band. The entire slate is priced at extremes — YES ≤ 8¢ or NO ≥ 93¢ — and most books are dead. The v4 rules treat cash as a position; this week the rules earn their keep by refusing 42 temptations, several of which look like "value" only through the lens that lost money in v1–v3 (tails and deep favorites).

Picks

Logged rejects

Deployed of $2,500

100%

Cash

Precedent rows written

What was scanned

Stage 0 read the full allowlist (60+ Tier-1 series, ~45 Tier-2) from targets.json and intersected it with open markets closing by 2026-09-26, using the DB mirror first and then a complete live sweep of every allowlisted series via the public API (to catch mirror lag — it caught the current KXMAMDANIEO weekly and confirmed nothing else was missed). Big-OI series like KXSAVEACT, KXTRUMPADMINLEAVE, KXFEDERALCHARGE, KXSTATEBALLOTMEASURE have no markets closing inside the horizon — their strikes sit at Dec 31 / Nov election dates and will enter future runs as they roll in.

Why the slate is extreme-priced: every in-horizon strike is a "before Sep 1" rung, Congress is in August recess (killing every confirmation/enact path), and the departure/ouster rungs are all quiet-news incumbents. The market has this figured out.

Notable non-picks (the ones worth a paragraph)

KXLUXONOUT-26AUG-26SEP01 departure/ouster · Tier 1

The one live news event in the slate. New Zealand's National Party held an emergency caucus meeting today (Aug 12) on Christopher Luxon's leadership — hence the 5,291-contract 24h volume, the highest of any Tier-1 candidate. Luxon survived the confidence vote ("full support of caucus" — RNZ, 1News). YES sits at 4/5¢, which is post-vote efficient pricing; residual pre-Sep-1 departure risk with an election ~3 months out is a few percent. Prob: 4% pre-market → 4% final. Both sides banned: YES is a <40¢ tail, NO at 96¢ is deep-favorite band. Reject: entry band.

KXMAMDANIEO-26AUG15-T0 confirmation/enact · Tier 1

The only market to pass every mechanical cut (vol24h 3,062, 1¢ spread) — and the closest thing to a real anomaly. Settled history: 2 YES in 10 tracked weeks, both via early settlement on the signing day (heat EO Jun 22, "OPEN for Small Business" EO Jul 20). Naive base rate ~20% vs 8¢ ask. But no August EOs have been signed through Aug 12 and no policy-package event is on the City Hall calendar inside the Aug 9–15 window. Prob: 15% pre-market → 15% final (book is liquid but this is exactly the burst-pattern the market underweights). Still untouchable: YES 8¢ is a banned tail, NO 93¢ a banned deep favorite. Reject: entry band, both sides. Findings persisted to the precedent store.

KXUSAIRANAGREEMENT-27-26SEP scheduled-process · Tier 2

114,527 contracts of 24h volume — the most specialist-dense book in the universe. YES 1/2¢. Precedent store: the June and July rungs both resolved NO because the rulebook demands an executed instrument with a concrete nuclear limit AND sanctions relief; the June 2026 14-point memorandum was explicitly ruled out by a dated resolver clarification in rules_secondary. Talks remain "ongoing" (Al Jazeera, Aug 3) — precisely the state the resolver has twice ruled NO. Prob: 3% pre-market → 2% final (deferred to a very liquid book). Reject: entry band.

KXMARTINDNCOUT-26MAY-26SEP01 departure/ouster · Tier 1

The only in-band entry in the whole slate — NO at 87¢ (thin-favorite band, half-size). But vol24h is 2 contracts and the spread is 6¢ (13/19). This is the exact zero-volume-phantom-edge cohort that had negative realized edge across 98 historical picks. Reject: mechanical (dead book, spread > 5¢).

Reject log — all 42

Reason codes: BAND = outside 40–89¢ entry band (tail <40¢ or deep favorite ≥90¢) · MECH = fails liquidity/spread floor (vol24h < 500, spread > 5¢, or bid/ask extreme) · CONT = continuous-number thesis · ACTIVE = already a live pick. Price shown is the YES ask (or noted side).

Market	Archetype	YES ask ¢	vol24h	Reject
KXLUXONOUT-26AUG-26SEP01	departure/ouster	5	5,291	BAND — post-vote efficient; prob 4%
KXMAMDANIEO-26AUG15-T0	confirmation/enact	8	3,062	BAND both sides — prob 15%; passes all mech cuts
KXUSAIRANAGREEMENT-27-26SEP	scheduled-process	2	114,527	BAND — precedent: 2 prior rungs NO
KXHEGSETHOUT-26APR-SEP01	multi-news-other	—	—	ACTIVE — already a live pick
KXALITOOUT-26SEP01	departure/ouster	0.4	1,003	BAND — precedent: JUL01 rung NO
KXKASHOUT-26APR-26SEP01	departure/ouster	4	27	MECH + BAND
KXMARTINDNCOUT-26MAY-26SEP01	departure/ouster	19	2	MECH — dead book, 6¢ spread; NO 87¢ was in band
KXLEAVELISACOOK-26SEP01	departure/ouster	3	335	MECH + BAND
KXLEAVEPOWELLGOV-26SEP01	departure/ouster	3	0	MECH + BAND — governor seat ≠ chair (precedent)
KXLEAVEWALZ-26JUL-26SEP01	departure/ouster	2	242	MECH + BAND
KXLUTNICKOUT-26SEP01	departure/ouster	6	0	MECH + BAND
KXBURNHAMOUT-26JUL-26SEP01	departure/ouster	1	0	MECH + BAND
KXFARAGEOUTREFORM-26JUL-SEP01	departure/ouster	7	0	MECH (7¢ spread) + BAND
KXGALLEGOOUT-26APR-SEP01	departure/ouster	3	0	MECH + BAND
KXLEAVEMILLS-26NOV03-26SEP01	departure/ouster	12	0	MECH (7¢ spread, dead book)
KXTRUMPTHUNEOUT-27JAN01-SEP	departure/ouster	24	0	MECH (22¢ spread — unquotable)
KXDEMSCHUMER-27-SEP01	departure/ouster	4	0	MECH + BAND
KXIMPEACH-26-SEP01	legal-action	1	76	MECH + BAND
KXLAKECONF-26MAY12-SEP01	confirmation/enact	1	1	MECH + BAND — Senate recess; 1¢ correct
KXFARMBILL-26MAY-SEP01	confirmation/enact	2	50	MECH + BAND
KXRECNCH-27-SEP01	confirmation/enact	2	0	MECH + BAND — FY27 rung, no upstream action
KXSANCTIONRUSSIA-26JUL	confirmation/enact	6	469	MECH (just under vol floor) + BAND
KXTARIFFBILL-26JUL-26SEP01	confirmation/enact	1	0	MECH + BAND — tariffs are executive action
KXJUDGECOUNT-26AUG-1	confirmation/enact	2	0	CONT (count pace) + MECH
KXJUDGECOUNT-26AUG-3	confirmation/enact	2	0	CONT + cluster runner-up
KXJUDGECOUNT-26AUG-4	confirmation/enact	1	0	CONT + cluster runner-up
KXCBDECISIONCANADA-26SEP-H0	scheduled-process	100	2	BAND (bid 98) + MECH — hold-leg carry (precedent)
KXCBDECISIONCANADA-26SEP-C25	scheduled-process	4	5	MECH + BAND
KXCBDECISIONCANADA-26SEP-C26	scheduled-process	1	0	MECH + BAND — outer rung (precedent: all NO)
KXCBDECISIONCANADA-26SEP-H25	scheduled-process	1	0	MECH + BAND
KXCBDECISIONMEXICO-26SEP24-HOLD	scheduled-process	96	3	BAND (≥90¢) + MECH
KXCBDECISIONMEXICO-26SEP24-C25	scheduled-process	5	0	MECH + BAND
KXCBDECISIONMEXICO-26SEP24-C50	scheduled-process	2	0	MECH + BAND
KXCBDECISIONMEXICO-26SEP24-C50P	scheduled-process	2	0	MECH + BAND
KXCBDECISIONMEXICO-26SEP24-H25P	scheduled-process	4	0	MECH + BAND
KXCBDECISIONMEXICO-26SEP24-H50	scheduled-process	1	0	MECH + BAND
KXCBDECISIONMEXICO-26SEP24-H50P	scheduled-process	1	0	MECH + BAND
KXCOMPANYACTIONMERGER-27-26SEP01	corporate-event	2	0	MECH + BAND
KXIPOANTHROPIC-DATE-26SEP01	corporate-event	1	19,210	BAND — mechanical IPO definition (precedent)
KXIPOOPENAI-26SEP01	corporate-event	1	0	MECH + BAND
KXIPOSTARLINK-26SEP01	corporate-event	1	0	MECH + BAND
KXIPOANDURIL-26SEP01	corporate-event	3	0	MECH + BAND

Portfolio tables

Position sizes & limits

No positions. $2,500 (100%) held as cash; reserve requirement (≥10%) trivially satisfied.

Cluster table

No deployed clusters. Shadow-tracked cluster runner-ups this run: judges-aug2026 (2 runner-up legs logged).

Conviction exposure

Conviction	Positions	Cost	% of capital
HIGH	0	$0	0%
MEDIUM	0	$0	0%
Cash	—	$2,500	100%

Risk profile & execution notes

Max loss this run: $0. No orders to place.
The 40–70% expected-cash guidance is exceeded (100%) — consistent with an all-recess horizon where every allowlisted strike is a Sep-1 rung on a quiet incumbent or an idle legislative process.
Watch items for the next run (as strikes roll or news lands): KXMAMDANIEO weeklies if a City Hall policy-package event is scheduled inside a window; KXMARTINDNCOUT if the book comes alive (NO was in band); KXLUXONOUT post-election rungs; the Sep-onward KXJUDGECOUNT rungs once the Senate returns (still CONT-screened); Dec-31 strikes on KXSAVEACT / KXTRUMPADMINLEAVE / KXFEDERALCHARGE as they enter the 45-day horizon in mid-November.

Precedents consulted

KXALITOOUT-26JUL01 → NO — retirement risk concentrates in June opinion days; term closed silent.
KXKASHOUT-26APR-JUN01 / -JUL01 → NO — "matter of time" sourcing never pays; only an actual vacancy resolves YES.
KXLEAVEPOWELLGOV-26JUL01 → NO — chairmanship news ≠ Governor-seat contract.
KXLEAVEWALZ-27JAN01 (JUL01 rung) → NO — fixed-term incumbents: only resignation, completed removal, or death.
KXUSAIRANAGREEMENT-27-26JUN / -26JUL → NO — executed instrument with nuclear limit AND sanctions relief required; June MOU explicitly ruled out.
KXIPOANTHROPIC-DATE-26JUN01 / -26JUL01 → NO — mechanical IPO definition (S-1 effectiveness / priced offering / assigned ticker).
KXCBDECISIONCANADA 26JUN + 26JUL ladders → hold legs YES at 1–3¢ carry; all outer rungs NO across five ladders.
KXRECNCH FY26 ladder → rungs ahead of upstream-chamber action are structural NOs.
New this run (written back): KXMAMDANIEO-26JUN27-T0 and -26JUL25-T0 — EO signings arrive in policy-launch bursts; base rate 2/10 weeks; early settlement on occurrence.

Method notes (v4 discipline check)

Probability before price: honored where diligence ran (Luxon 4%, Mamdani-EO 15%, Iran 3% pre-market) — though Stage-0 filter prices were unavoidably seen, no orderbook was studied before committing numbers.
Universe discipline: no market outside targets.json was researched; no exceptions made for tempting off-list books.
Liquidity floor: enforced on live numbers, not file snapshots — it alone killed 30 of 42 candidates, validating the zero-volume-cohort finding.
Band discipline: the only genuine anomaly found (Mamdani-EO YES, ~15% vs 8¢) sits in the banned <40¢ tail — the band rule forfeits ~$0.07/contract of modeled edge to avoid the cohort that went 0-for-8 lifetime. Logged for shadow-tracking; if this reject class starts resolving YES at meaningful rates, the record will say so.