Kalshi Elections — Mispricing Audit

Report date 2026-06-03 · Horizon ≤45 days · Capital $1,000 · Category = Elections

Forty-three active Elections markets close within 45 days. Most are noise: twelve are already-resolved AP-call timing legs from the June 2 California / LA primaries, and a cluster of "will X leave office" long-shots sit at 1–5¢ with nothing to win. After full diligence on the seven that screened through, two carry a defensible, fillable edge — both NO bets. This is a thin week; I am not padding the list.

Bottom line. Deploy $650 of $1,000 across two BUY NO positions (Díaz-Canel staying in office; a clean June FOMC). Hold $350 (35%) in reserve — Kelly-disciplined sizing on only two modest edges genuinely leaves this much cash, and deploying it would mean over-betting a 90¢ contract or buying into rejected names. Blended expected edge ≈ +$206 (≈+32% on deployed). The expected value is barbell-shaped: a near-certain small carry plus one higher-variance value bet.

1 · How this was researched

Universe. Queried the live Kalshi DB mirror for every status='active' market whose event category='Elections' closing between 2026-06-03 and 2026-07-18 → 43 markets.
Screen. Cut volume24h<1500, bid-ask spread >5¢, "what will X say"/mention noise, and anything already fully priced (YES bid ≥95¢ or ask ≤5¢). That removed all 12 AP-call legs (now 0/1¢, primary was June 2) and the dormant long-shots.
Diligence (7 candidates). For each: pulled full resolution rules; 14-day intraday history from market_snapshots (flagging ≥20¢ moves); the live Kalshi order book for fill depth within 3¢ of touch; and primary-source news (filings, official calendars, major-outlet reporting with dates). Each was worked up by an independent research agent and cross-checked by hand.
Reject gate. Drop anything where the rules could be disputed into the opposite result, where the price already fully moved on the thesis in the last 48h (<4¢ left), or where the book can't absorb a position without ≥3¢ slippage (those become watchlist, not capital).
Prices are live top-of-book as of 2026-06-03 ~22:00 UTC. Probabilities are my subjective estimates; I show the reasoning so you can disagree.

2 · Markets reviewed

Ticker	Question	Price (YES)	Vol 24h	Verdict
KXDIAZOUT-MDC-26JUL01	Díaz-Canel leaves office before Jul 1?	10/11¢	4,560	PICK · BUY NO
KXFEDCOMBO-26JUN-0-0	June FOMC: no change AND 0 dissents?	69/74¢	147*	PICK · BUY NO
KXMILLSPLATNER-26NOV03-JUL01	Mills endorses Platner before Jul 1?	13/17¢	150	WATCHLIST
KXPLATNERDROPOUT-26	Platner drops out before Jul 14?	10/11¢	22,056	REJECT — edge eaten
KXELECTIRAN-26JUL01	Iran holds presidential election before Jul 1?	1/3¢	733	REJECT — no payout space
KXTRUMPCOLLINS-26JUN16	Trump endorses Mike Collins before Jun 16?	70/75¢	30	REJECT — fair + illiquid
KXGENERICBALLOTVOTEHUB-26JUN05-T7.6	VoteHub generic ballot D−R >7.6% on Jun 5?	2/4¢	235	REJECT — unverifiable resolver
+ 12 AP-call timing legs (CA Gov / LA Mayor, all 0/1¢, primary was Jun 2) and the dormant "leave office" long-shots (Schumer, M. Johnson, Fetterman, Lai Ching-te; high OI, vol24h <1500, 1–5¢) — screened out. See §5.

*Fed-combo 24h volume is thin; the order book within 3¢ of the NO touch holds ~960 contracts, which is what governs fillability. Scale in with limits (see pick).

3 · Picks

Pick 1 — KXDIAZOUT-MDC-26JUL01 · BUY NO @ 0.90

Medium conviction

"Will Miguel Díaz-Canel leave office before July 1, 2026?" · closes 2026-07-01 14:00 UTC

Thesis

YES at 10–11¢ is pricing a ~10% chance that a constitutionally term-secured head of state vacates office inside a four-week window with no scheduled succession event. The fundamentals say that's closer to 4%. Buy NO at 90¢ against a fair value near 96¢.

Evidence (public)

Díaz-Canel told NBC's Meet the Press in April 2026 he will not step down — "not part of our vocabulary" — directly rebuffing US pressure tied to fuel talks (NPR, 2026-04-10; NBC News).
Havana "categorically rejected" the prospect of removing him in US talks — "the political system of Cuba is not up for negotiation" (Al Jazeera, 2026-03-20).
He holds both the presidency and PCC First Secretary post on a constitutional second term through 2028; no National Assembly session, party congress, or succession signal is scheduled before late June (Translating Cuba).
The spike to ~20¢ in late May was speculation off Cuba's eastern-grid collapse and US ouster pressure (Al Jazeera, 2026-05-14; Straight Arrow News). The 48h crash to ~10¢ is that speculation fading as the window narrows.
Cross-market check: the parallel Polymarket "out by June 30" market trades in the same low-single-digit-to-low-teens range — no independent venue prices a near-term exit highly.

Resolution nuance — read before sizing

The headline says "leaves office," but Kalshi's clarifications treat departure from either post (President or PCC First Secretary) — and reportedly even a forward-looking announcement that he will leave within a year — as YES. That lowers the bar versus the plain title. It does not flip the trade (each path is still very unlikely before July 1), but it is the reason this is medium, not high, conviction. Confirm the exact rules text on the market page before entry.

Tail risks (cleanest loss first)

Negotiated/forced exit: the acute energy collapse + Trump-administration pressure produces a sudden transition or a "will step down" announcement before July 1.
Sudden death or incapacitation (health is not publicly disclosed) — low base rate, but a discontinuity.
An unscheduled emergency Party/Assembly action installing a successor.

Numbers & liquidity

Side / action: BUY NO
Live NO price: 0.90 (touch; vs YES bid 0.10)
Limit: ≤ 0.90 (rest at 0.89 if patient)
True P(YES): ≈ 4% → NO fair ≈ 0.96
Edge: ≈ +6¢ / contract · +6.7% at fill
Book depth: ~$665 resting at 0.90, ~$505 at 0.91 — a 444-contract position fills at touch
14-day path: ~21–22¢ (May 29–31) → 19¢ (Jun 1) → ~9–11¢ (Jun 3). Crash on fading speculation, not a positive YES catalyst.

Pick 2 — KXFEDCOMBO-26JUN-0-0 · BUY NO @ 0.32

Medium conviction

"June 2026 FOMC: No rate change AND 0 dissents?" · closes 2026-06-17 17:55 UTC · (Kalshi files this Fed combo under Elections)

Thesis

This is a joint event: rates unchanged and zero dissenting votes. The combo trades 69–74¢ YES. Since P(no change) is ~96%, that implies the market thinks zero dissents is ~72% likely. After an 8–4 April vote and with a brand-new Chair's first meeting, ~50% is the better number. Buy NO at ~32¢ against a fair value near 55¢.

Evidence (public)

April 28–29, 2026 FOMC was an 8–4 vote — the most dissents since 1992. Miran wanted a cut; Hammack, Kashkari and Logan opposed an easing bias (CNBC, 2026-05-01). A committee that fractured 8–4 six weeks ago is not an obvious 0-dissent bet.
Independent cross-check: dedicated "who dissents in June" markets price zero dissents at roughly 46% (Polymarket) to 57% (Robinhood) — i.e. ~50%, far below the ~72% the Kalshi combo implies (Octagon AI, citing Polymarket/Robinhood, 2026-05-20).
The no-change leg is near-locked: CME FedWatch ~96–97% hold, with inflation re-accelerating (CPI ~3.8%) keeping a cut off the table (Yahoo Finance, 2026-05-13).
June 16–17 is Kevin Warsh's first meeting as Chair (confirmed 54–45, sworn in May 15), arriving with a fresh Summary of Economic Projections / dot plot — historically a flashpoint for dissent (CNBC, 2026-05-13).

Tail risks (cleanest loss first)

New-Chair "honeymoon": the committee closes ranks behind Warsh at his first meeting and posts a unanimous hold — dissenters defer. This is the single thing that makes NO lose, and it is a real ~45% of the distribution. Size accordingly.
The sister dissent markets could be stale; if true P(0 dissents) is genuinely 60%+, the edge shrinks.
A surprise hawkish framing that paradoxically unifies the committee (still NO-favorable only if it also avoids the rare cut).

Numbers & liquidity

Side / action: BUY NO
Live NO price: 0.31 touch / ~0.32 avg fill
Limit: ≤ 0.33 — scale in, do not market-buy
P(NO win): ≈ 55% → NO fair ≈ 0.55
Edge: ≈ +23¢ / contract · +72% at fill
Book depth: ~230 contracts at 0.31, ~735 at 0.32 (~$570 within touch). 24h traded volume is only ~147, so a 781-contract order must be worked patiently with limits into June 16–17.
Note: This is a value bet, not a lock — ~55% win at a price that should be ~55¢. Variance is the point; small absolute size + large cash reserve manage it.

4 · Recommended $1,000 portfolio

Pick	Side	Limit	Contracts	Cost	Max payout	EV ¢/ct	Pos. EV	EV %	Conviction
KXDIAZOUT-MDC	NO	0.90	444	$400	$444	+6.0	+$27	+6.7%	Med
KXFEDCOMBO-26JUN-0-0	NO	0.33	781	$250	$781	+23.0	+$180	+71.8%	Med
Total deployed	—	—	1,225	$650	$1,225	—	+$206	+31.7%	—
Cash reserve	—	—	—	$350	$350	—	—	—	35%

EV uses true P(NO) of 96% (Díaz-Canel) and 55% (Fed combo). Blended expected edge ≈ +$206 on $650 deployed (≈ +20.6% on the full $1,000).

Risk profile

Worst case (both lose, ≈1.8%): −$650; bankroll floors at the $350 reserve.
Best case (both win, ≈53%): +$575 profit; bankroll ≈ $1,575.
Most-likely shape: the outcome is dominated by the Fed coin-flip. Díaz-Canel wins ~96% of the time for a small +$44. Then ~53% of the time the Fed leg also wins (≈ +$575 total); ~43% of the time it loses (≈ −$206 net). This is a barbell: a near-certain small carry stapled to one high-EV, higher-variance value bet.
Concentration: by underlying — Díaz-Canel "incumbent stays" 62% of deployed, Fed "dissent mispricing" 38%. Two uncorrelated underlyings (Cuban politics vs US monetary policy). No single name >40% of total capital.
Why 35% cash, not 10%: only two edges cleared the 4¢/liquidity bar this week. Kelly-disciplined sizing on a 90¢ contract caps Díaz-Canel near $400; pushing more in would mean risking 90¢ to make 10¢ or buying rejected names. The reserve is earmarked for the watchlist triggers below.

Execution notes

Díaz-Canel: rest a NO buy limit at 0.89–0.90; ~$665 sits at 0.90 so a 444-lot fills at touch. Do not chase above 0.91. Invalidation/exit: any credible report of a negotiated transition, an emergency Party/Assembly session, or a "will step down within a year" statement — exit immediately, the rules may count it as YES.
Fed combo: work NO limits ≤0.33 over the run-up to June 16–17; never market-buy (daily volume ~147 — you'd move the print). Target average ≤0.32. Invalidation/trim: if the standalone dissent markets converge toward ≥65% zero-dissents, or strong signals point to a unanimous honeymoon hold under Warsh.
Reserve ($350): earmarked for (a) a Mills→Platner NO add if Mills re-activates her campaign (see watchlist); (b) Díaz-Canel adds if it dips below 0.88 on noise; (c) opportunistic Maine repricings after the June 9 primary.
Hedges: none clean within the category. The Fed leg's natural complement (the "no change & >0 dissents" sister leg) is the thesis, so hedging it defeats the trade.

5 · What I rejected and why

KXMILLSPLATNER-26NOV03-JUL01 — Mills endorses Platner before Jul 1 · WATCHLIST (real edge, can't fill at fair). The NO thesis is strong: Janet Mills is Platner's opponent on the June 9 ballot and on June 1 publicly reminded voters "I am still on the ballot," having repeatedly declined to endorse him (Press Herald; Maine Morning Star). True P(endorse) ≈ 10%, fair NO ≈ 0.90. But the book only offers NO at ~0.88 immediately (~$5 at 0.87, ~$100 at 0.88) on ~150 daily volume — a 2¢ touch edge, below the bar. Trigger: rest a NO bid at ≤0.86; add from reserve if Mills re-activates her campaign (supporters are urging it). Watch the tail: if Platner wins June 9 and Mills issues any party-unity statement before July 1, NO is at risk.

KXPLATNERDROPOUT-26 — Platner drops out before Jul 14 · REJECT (edge already eaten). The most liquid market in the category (22k 24h volume). The WSJ sexting story spiked YES from ~11¢ to ~32¢ intraday on June 2, but it retraced to 10–11¢ within 48h as Platner stayed defiant, met Senate Democrats, kept Sanders/Warren backing, and remained the heavy favorite to win the June 9 primary — and a primary loss is explicitly not a "drop out" (WBUR, 2026-06-02; PBS NewsHour; WaPo). My fair P(YES) ≈ 7%, leaving only ~3¢ on NO at 0.90 — under the 4¢ threshold, and the thesis has already moved into the price. (Genuinely interesting: a Maine provision lets a primary winner who withdraws by 5pm July 13 be replaced by a party-picked nominee — but there is no public sign Democrats are pursuing it: Fox News.)

KXFEDCOMBO sister legs / KXELECTIRAN-26JUL01 — Iran holds a presidential election before Jul 1 · REJECT (no payout space). NO is correct — Pezeshkian was elected in 2024, no election is scheduled, and even an accepted presidential resignation triggers a ~50-day constitutional window that runs past July 1. But you can only lift the NO offer at ~99¢ for ~1¢ of edge, on 733 daily volume. Correct side, nothing to win.

KXTRUMPCOLLINS-26JUN16 — Trump endorses Mike Collins before Jun 16 · REJECT (fair price, untradeable). This is the GOP primary runoff (Collins vs Dooley). Trump stayed neutral through the primary even as all candidates courted him, but Collins hired Trump's 2024 advisers — Fabrizio, Saler, Englander — in late May (Axios, 2026-05-27; NBC News). With an active Trump–Kemp proxy fight over the seat (NOTUS), the 75¢ YES is fair-to-rich (true ≈ 62%); the only edge is on NO and rests entirely on unknowable Trump timing. The book shows single-digit contracts at touch on 30 daily volume — uninvestable.

KXGENERICBALLOTVOTEHUB-26JUN05-T7.6 — VoteHub generic ballot >7.6% on Jun 5 · REJECT (unverifiable resolver). Resolution hinges on a single aggregator's (VoteHub's) time-weighted value at a fixed instant, and that live value isn't independently verifiable. VoteHub was D+7.2 on May 20 trending up (VoteHub), while other averages sit near D+6–7 (FiftyPlusOne). The price collapse to 2–3¢ suggests informed traders see it below 7.6, but I can't confirm the resolving number, so the edge is undefendable. Closes June 5.

The 12 AP-call timing legs & the dormant long-shots · REJECT (no edge). The CA Governor and LA Mayor primary AP-call markets all sit at 0/1¢ — the June 2 primaries are over and they are effectively resolved/fully priced. The "will X leave office" names (Schumer as leader, Mike Johnson as Speaker, Fetterman switching parties, Lai Ching-te) carry large open interest but <1,500 24h volume and price at 1–5¢, offering 1–2¢ on NO — capital-inefficient. Both groups were screened out before diligence.

6 · Sources

Maine Senate (Platner / Mills): Ballotpedia — June 9 Dem primary · CNN — Mills "still on the ballot" (6/1) · Press Herald — Mills (6/1) · Maine Morning Star — Mills "not ready to support" (5/1) · CNN — Mills suspends campaign (4/30) · WBUR — Platner scandal (6/2) · WaPo — reassures senators (6/2) · PBS NewsHour (6/2) · Axios — Dem anxiety (6/2) · Fox — ballot replacement provision (6/2)

Cuba (Díaz-Canel): NPR — "not stepping down" (4/10) · NBC News · Al Jazeera — defiant (4/10) · Al Jazeera — "categorically rejects" (3/20) · Al Jazeera — blackouts/protests (5/14) · Straight Arrow News — US ouster push · Translating Cuba — term horizon · Polymarket — parallel market

Federal Reserve (June FOMC): CNBC — April 8–4 dissents (5/1) · CME FedWatch · Octagon AI — dissent markets (5/20) · CNBC — Warsh confirmed (5/13) · Yahoo Finance — inflation re-accelerates (5/13)

Georgia / generic ballot (rejects): Axios — Collins hires Trump advisers (5/27) · NBC — GA runoff · NOTUS — Trump–Kemp proxy fight · VoteHub — generic ballot (5/20) · FiftyPlusOne — generic ballot

Market data: Kalshi live DB mirror (markets, events, 5-minute snapshots) and the Kalshi public order-book API (api.elections.kalshi.com/trade-api/v2), queried 2026-06-03.

Data & method. Universe, rules text and price history from a read-only Kalshi DB mirror; live order books from Kalshi's public API; news from primary sources cited inline. Prices are top-of-book as of 2026-06-03 ~22:00 UTC and will drift.

Disclaimer. Not investment advice. All probabilities here are subjective estimates, shown with their reasoning so you can disagree. Prediction-market contracts can and do resolve to zero — every position can lose 100% of the capital placed on it. Size only what you can afford to lose, and re-check the live rules and book before trading; resolution criteria are often subtler than the title.