Kalshi Politics — Mispricing Audit
Report date 2026-07-27 · Horizon 45 days · Capital $2,500 · Methodology v3.3 (conviction-weighted, EV-agnostic sizing; resolver risk priced; sub-35¢ tails and LOW conviction never deploy) · Picks in this report count toward the public track record and are cleared for auto-copy.
1. How this was researched
Mode: category-match. trading_events.category = 'Politics' matched 352 active markets closing within 45 days on the live Kalshi mirror (no theme-keyword fallback needed). Stage-1 mechanical screens — 24h volume ≥ 1,500, bid–ask spread ≤ 5¢, not already priced out (YES bid < 95¢, YES ask > 5¢), no announcer/behavior-noise markets, and no ticker already live in users' feeds — cut 352 → 25. Cutting pure Trump-behavior counting series (Truth Social post-count buckets, posting-hour markets, endorsement counts) as noise left 17 candidates for full diligence.
For each candidate I pulled the resolution rules from the mirror, 14-day daily and 48-hour hourly candlesticks plus the live orderbook from the Kalshi public API, and ran primary-source news diligence (RCP's own JSON data endpoints, the IMF PortWatch ArcGIS feature service, the Senate Democratic Caucus floor schedule, DOJ Pardon Attorney clemency lists, nyc.gov mayoral orders, congress.gov nominations). Every candidate got a resolver profile (Stage 2.5): who prints the resolving number, by what mechanism, and what can change the print between now and close — with each resolver risk priced as an explicit point deduction. Note: the mirror's schema no longer exposes a market_snapshots table; all price history in this report comes from Kalshi's candlestick API instead.
2. Markets reviewed
| Ticker | Question (short) | YES bid/ask | Vol 24h | Disposition |
|---|---|---|---|---|
| KXHORMUZNORM-26MAR17-B260901 | Hormuz 7dMA transit calls >60 before Sep 1 | 8/9 | 56,478 | PICK — BUY NO |
| KXAPRPOTUS-26JUL31-40.2 | RCP approval <40.2 on Jul 31 | 6/7 | 7,083 | PICK — BUY NO |
| KXCLARITYVOTE-26JUL-AUG08 | Any recorded Senate vote on CLARITY Act by Aug 8 | 60/61 | 10,374 | PICK — BUY YES |
| KXELECTIONEMERGENCY-26SEP01 | Midterm national-emergency executive action by Sep 1 | 10/11 | 1,927 | PICK — BUY NO |
| KXMAMDANIEO-26AUG01-T0 | Mamdani signs >0 non-emergency EOs this week | 12/13 | 3,025 | REJECT — banned tail (real edge) |
| KXHORMUZWEEKLY-26JUL26-T30 | Hormuz weekly transits >30, Jul 20–26 | 88/89 | 5,090 | REJECT — edge ≤ spread |
| KXHORMUZWEEKLY-26JUL26-T50 | Hormuz weekly transits >50, Jul 20–26 | 14/15 | 14,910 | REJECT — tail + informed flow |
| KXHORMUZWEEKLY-26JUL26-T60 | Hormuz weekly transits >60, Jul 20–26 | 5/6 | 11,108 | Stage-1 adjacent; folded into T50 analysis |
| KXHORMUZPEAK-26JUL26-T10 | Hormuz max single day >10, Jul 20–26 | 66/70 | 1,665 | REJECT — depth/slippage |
| KXHEGSETHOUT-26APR-SEP01 | Hegseth out as SecDef before Sep 1 | 8/8 | 3,822 | REJECT — duplicates live cluster |
| KXPARDONSTRUMP-26JUL-17 | 10–24 people pardoned in July | 91/92 | 2,225 | REJECT — no edge at ask |
| KXLAKECONF-26MAY12-SEP01 | Kari Lake confirmed (Jamaica) before Sep 1 | 5/8 | 2,196 | REJECT — no edge, live tail |
| KXTRUMPNUMSTATES-26AUG01-E10 | Exactly 10 states visited in July | 71/73 | 2,645 | REJECT — resolver opacity |
| KXTRUMPNUMSTATES-26AUG01-E9 | Exactly 9 states visited in July | 11/12 | 21,328 | REJECT — resolver opacity |
| KXTRUMPUFC-26AUG-DJT | Trump attends UFC 330 | 14/15 | 2,000 | REJECT — tail, no signal |
| KXAPRPOTUS-26JUL31-40.9 | RCP approval 40.8–41.0 on Jul 31 | 31/32 | 3,091 | REJECT — tail, composition risk |
| KXAPRPOTUS-26JUL31-40.6 | RCP approval 40.5–40.7 on Jul 31 | 17/18 | 2,123 | REJECT — fairly priced |
3. Picks — detail and thesis
Pick 1 — KXHORMUZNORM-26MAR17-B260901 · BUY NO @ 0.92 · HIGH CONVICTION
Will the 7-day moving average of Strait of Hormuz transit calls (IMF PortWatch) be above 60 before Sep 1, 2026?
Thesis. YES requires roughly a 5× jump in traffic sustained for a week within 35 days, from a strait that is effectively closed. PortWatch's latest published 7-day average (through Jul 19) is 12.1 transits/day; the threshold is 60. When the June US–Iran deal briefly reopened traffic, the single best day hit 51 and the 7dMA peaked around 34 — a full-scale diplomatic breakthrough got barely halfway to this strike. Since then: the Jul 8 truce breakdown, the Jul 12 IRGC closure declaration, and three more tankers struck on Jul 20. War-risk premiums are ~8× and ~355 vessels are stranded.
Resolver: IMF PortWatch daily chokepoint dataset (Hormuz = chokepoint6 in the ArcGIS feed), first occurrence of 7dMA > 60 on published data; publishes weekly Tuesdays ~13:00 UTC with ~2-day lag. Priced risks: −2 pts (reopening queue-flush: 355 stranded vessels could briefly spike daily counts far above normal), −1 pt (PortWatch revision risk — it restated the chokepoint6 boundary in March 2026), −1 pt (faster-than-June deal implementation). Evidence-based ~99% → 95%.
- PortWatch daily prints Jul 13–19: 9–17/day; weekly sum 85 (prior week 107) — live ArcGIS query, Jul 27.
- Strait "effectively closed for 149 days"; Jul 20: Greek tankers Kavomaleas & Acheloos and Kuwaiti Kaifan struck (straits.live; Wikipedia crisis timeline).
- June analog: deal ~Jun 19 → peak day 51 (Jun 24), 7dMA topped out ~34 — never approached 60.
Tail risk (cleanest loss): a genuine ceasefire-plus-reopening in early August followed by a queue release that pushes a full week above ~60/day before Sep 1. That is the one scenario that flips this; everything short of it loses YES the race against the calendar.
Liquidity/entry context: yes 8 bid / 9 ask; 3,968 NO contracts fillable at 92¢; vol24h 56,478; OI 854,333 — deepest book in the category. Price history: YES has sat 7–12¢ for two weeks (12→9¢ over 14 days) — no adverse 48h move.
Pick 2 — KXAPRPOTUS-26JUL31-40.2 · BUY NO @ 0.94 · HIGH CONVICTION
Will Trump's RCP approval average be below 40.2 at 11:00 AM ET on Jul 31, 2026?
Thesis. The RCP average is 40.9 today (Jul 27), from RCP's own machine-readable history. Resolving YES needs a −0.7-point move in 4 days — equal to the single largest 4-day move of the past 120 days (mean 0.23, median 0.20, 90th percentile 0.50), and it must happen against a mechanical headwind: the oldest poll in the current window is Pew's 34 (the low outlier), whose scheduled aging-out alone would lift the average ~+0.5. The average has not left the 40.3–41.2 band all month.
Resolver: RealClearPolitics curated approval average, snapshot at 11:00 ET Jul 31. Priced risks: −1.5 pts (RCP inclusion discretion — which polls enter/exit the window is editorial), −1 pt (a coordinated string of sub-39 gold-standard prints on Iran-war fatigue; CBS 39, Fox 39, Emerson 39, Echelon 38 are already the direction of travel), −0.5 pt (snapshot timing vs. morning poll drops). Evidence-based ~99.5% → 97%.
- Daily history (RCP JSON, Jul 27): Jul 20: 41.1 · Jul 23: 40.8 · Jul 25: 41.1 · Jul 27: 40.9. Disapprove rising (+1.9/wk) but approve stable.
- Expected prints before Jul 31: Rasmussen daily (42–45), Economist/YouGov Tue (~40 four weeks running), Morning Consult (44–45), possibly RMG Fri (44–46) — the scheduled releases skew Trump-friendly.
- Verified poll table sums exactly to the published 40.9 — the composition math is auditable.
Tail risk: two or three simultaneous sub-39 releases from high-weight pollsters Jul 28–30 while RCP retains Pew — the only realistic path to −0.7 in 4 days.
Liquidity/entry context: yes 6 bid / 7 ask; 3,416 NO contracts fillable at 94¢; vol24h 7,083; OI 16,411. Price history: bucket opened Jul 25 at 4¢ YES, drifted to 6–7¢ — no adverse move.
Pick 3 — KXCLARITYVOTE-26JUL-AUG08 · BUY YES @ 0.62 · MEDIUM CONVICTION
Will the Senate conduct any recorded vote on the CLARITY Act before Aug 8, 2026?
Thesis — this is a rules-vs-headline trade. The crowd is pricing something like "will the crypto bill pass or get a deal before recess" (Thune himself said passage will miss the window, and seven swing Democrats rejected the Lummis text on Jul 22 — hence the 77→39 crash). But the rules resolve YES on any recorded vote — a cloture vote on the motion to proceed counts even if it fails 53–47. Leadership has said, twice and on the record, that exactly such a vote is coming before the Aug 7 recess regardless of a deal: Thune Jul 23 ("I would like to at least get Clarity started"), and Punchbowl's Jul 27 pre-recess queue reporting a procedural crypto vote "regardless of whether Democrats back a proposed deal on ethics restraints." Forcing Democrats on record is politically useful to the majority even in failure.
Resolver: Senate roll-call record — any recorded vote on the CLARITY Act before Aug 8; fully objective and instantly checkable. Priced risks: −5 pts (no cloture motion actually filed as of the Mon Jul 27 schedule), −5 pts (floor congestion: Clayton DNI, Russia-sanctions vehicle, a two-week nominations package, approps prep — Thune may protect floor time and punt to a Sep 14 return, past the deadline), −3 pts (recess-cancellation fight scrambles the queue). Evidence-based ~89% → 76%.
- Bloomberg Jul 22: Thune expected a vote even without a Dem deal; motion to proceed possibly as early as Jul 27–28.
- CoinDesk Jul 23: Thune — bill will miss passage pre-recess, but "get Clarity started"; White House crypto adviser points to "the first week of August."
- Punchbowl Jul 27: procedural crypto vote in the pre-recess queue "regardless" of the ethics deal.
- Whipsaw anatomy: 77¢ (Bessent "1-yard line") → 39¢ (seven Democrats reject text; Thune says passage slips) → 56–61¢ (procedural-vote-anyway reporting). The market still hasn't fully separated vote from passage.
Tail risk: cloture never gets filed — Thune opts not to burn 2–3 floor days on a vote he loses, and the bill waits for September. That's a real 1-in-4.
Liquidity/entry context: yes 60 bid / 61 ask; ~658 YES contracts fillable ≤62¢; vol24h 10,374; OI 48,813. Price history: 65→77→39→65→61 over 14 days; it moved 20¢+ twice in the last week, so the entry demands a hard limit — the thesis is priced at 61, not at 70.
Pick 4 — KXELECTIONEMERGENCY-26SEP01 · BUY NO @ 0.90 · MEDIUM CONVICTION
Will Trump issue any executive action declaring a national emergency regarding the 2026 midterm elections before Sep 1, 2026?
Thesis. The only known draft order is an outside-activist document — a 17-page draft by pro-Trump attorneys (Ticktin et al., dated Apr 2025), surfaced by the Washington Post on Feb 26, 2026 — and Trump personally disavowed it the next day ("never heard about it... not considering"). Five months later there is still no reporting of any administration-drafted order, and legal analyses are unanimous that no statutory authority exists. The market's 10–11¢ YES has been drifting down (21¢ two weeks ago) as nothing materializes.
Resolver: any executive action by Trump declaring a national emergency regarding the 2026 midterm election, before Sep 1. Priced risks: −3 pts (definitional looseness — an "election security"/foreign-interference emergency action, e.g. an EO 13848-style declaration aimed at midterm interference, might be read as qualifying), −4 pts (escalation: Trump's Jul 16 "election security" speech echoed the activist draft's framing per advocacy trackers, and he released "election documents alleging foreign threats" Jul 18 — the rhetorical groundwork is being laid). Evidence-based ~100% → 93%.
- WaPo Feb 26, 2026 + Democracy Docket full text: draft is activist-authored, not administration.
- Newsweek/The Hill Feb 27: Trump denies considering it.
- American Democracy Minute Jul 21: Jul 16 speech "appears to have elements of" the draft — rhetoric, not action; no emergency declared, no drafting reported.
Tail risk: an impulsive August declaration as midterm messaging escalates — the speech shows the theme is on his mind; 35 days is a long time. That's why this is MEDIUM, not HIGH, and sized at 7%.
Liquidity/entry context: yes 10 bid / 11 ask; 562 NO contracts fillable at 90¢; vol24h 1,927; OI 22,616. Price history: YES 21→15→17¢ over 14 days, 10–11¢ live quote — decaying, no adverse move.
4. Recommended $2,500 portfolio
Sizing discipline (v3): position size below is set only by conviction tier (HIGH ≤ 15% = $375; MEDIUM ≤ 7% = $175) and entry band, with a 15% per-cluster cap. The EV columns are recorded for calibration and did not drive contract counts — the resolved v1/v2 record showed stated EV was anti-predictive.
| Pick | Side @ limit | Tier / band | Contracts | Cost | Max payout | EV ¢/ct | EV % |
|---|---|---|---|---|---|---|---|
| KXHORMUZNORM…B260901 | NO @ 92¢ | HIGH / >90¢ | 407 | $374.44 | $407.00 | +3.0 | +3.3% |
| KXAPRPOTUS-26JUL31-40.2 | NO @ 94¢ | HIGH / >90¢ | 398 | $374.12 | $398.00 | +3.0 | +3.2% |
| KXCLARITYVOTE-26JUL-AUG08 | YES @ 62¢ | MEDIUM / 60–90¢ | 282 | $174.84 | $282.00 | +14.0 | +22.6% |
| KXELECTIONEMERGENCY-26SEP01 | NO @ 90¢ | MEDIUM / 60–90¢ | 194 | $174.60 | $194.00 | +3.0 | +3.3% |
| Total deployed | 1,281 | $1,098.00 | $1,281.00 | +6.3% blended |
Cash held: $1,402 (56.1%). Cash is a position — this category is fairly efficient this week and the two juiciest raw edges (Mamdani EO at 13¢, Hormuz T50 at 15¢) sit in the banned tail band. Dollar edge on deployed capital ≈ +$69; max payout $1,281 (+$183 if all four win, model-probability ≈ 65%).
Cluster exposure (cap: 15% = $375 per cluster)
| Cluster | Cost | % of capital | Cap |
|---|---|---|---|
| hormuz-reopening | $374.44 | 15.0% | 15% |
| rcp-approval-jul31 | $374.12 | 15.0% | 15% |
| clarity-floor-vote | $174.84 | 7.0% | 15% |
| election-emergency-eo | $174.60 | 7.0% | 15% |
All four theses are genuinely independent (shipping data, polling arithmetic, Senate floor mechanics, White House action). One cross-feed note: the live feed already carries Hormuz weekly-window picks from earlier runs; this pick rides reopening-by-September, not the Jul 20–26 data print, but followers holding both should know they share Iran-headline sensitivity.
Conviction exposure
| Tier | Cost | % of capital | Per-pick cap |
|---|---|---|---|
| HIGH | $748.56 | 29.9% | $375 |
| MEDIUM | $349.44 | 14.0% | $175 |
| LOW | $0 | 0% | never deploys (v3.3) |
Risk profile
- Worst case −$1,098 (all four lose). It would take a full Hormuz reopening with queue overshoot, an unprecedented 4-day approval collapse, a punted crypto vote, and a midterm emergency declaration — four essentially uncorrelated failures.
- Best case +$183 (all four win; model P ≈ 0.65). Expected ≈ +$69 on model probabilities.
- Largest single loss: either HIGH pick at −$374 (model P of loss: 5% and 3%). The most likely loss is CLARITY at −$175 (24% chance) — which is why it carries the MEDIUM cap despite the largest stated edge.
- No cluster can flip the run's sign on its own: worst single cluster = 15.0% of capital.
Execution notes
- Limits, not market orders. All four entries are at the current ask; Kalshi taker fees (≈0.07×P×(1−P)) cost ~0.5¢ at 92–94¢ and ~1.6¢ at 62¢ — no pick flips sign net of fees, but paying 2¢ through the CLARITY ask would eat 14% of that edge.
- CLARITY (invalidation): if no cloture motion on the bill is filed by end of day Tue Aug 4, the floor-time window is functionally gone — exit at market. Conversely a filed cloture motion is a hold-to-resolution signal even if the vote is expected to fail.
- HORMUZNORM (invalidation): any credible US-Iran ceasefire/reopening headline — exit before the queue-flush repricing, don't wait for PortWatch prints (Tuesdays ~13:00 UTC).
- APRPOTUS (invalidation): two sub-39 prints from high-weight pollsters on Jul 28–29 put the −0.7 move in play — exit; otherwise hold the 4 days to resolution.
- ELECTIONEMERGENCY (invalidation): any report that the administration itself is drafting an election-emergency order (as opposed to activist pressure) — exit immediately; the Jul 16 speech thread is the thing to watch.
- Watchlist: KXHORMUZPEAK-26JUL26-T10 (YES fair value ~80 vs 70 ask, killed only by a 5¢ spread on a 6-lot book — take liquidity if a seller shows at ≤72); KXMAMDANIEO next-week reissues (if the series relists at tail prices, the shadow-track case for reinstating a capped tail slot builds here).
- Reserve: $1,402 (56%) held as cash for opportunistic adds — well above the 10% minimum.
5. What I rejected and why
Every Stage-2 survivor that didn't become a pick is machine-logged in picks.json and shadow-tracked to settlement — if the rejects outperform the picks, the screen itself is falsified, and that's a finding. Entry-band cuts are marked ⛔.
| Ticker | Would-have side @ price | My prob | Reject reason |
|---|---|---|---|
| KXMAMDANIEO-26AUG01-T0 | YES @ 13¢ | 42% | ⛔ Sub-35¢ tail — the strongest raw edge found this run (≈3×), banned by v3.2. 19 regular EOs in 29 weeks (Poisson P(≥1/wk) ≈ 48%), two signed in the five days before the window, and the market prices 13%. This is precisely the pick the shadow-track exists to adjudicate. |
| KXHORMUZWEEKLY-26JUL26-T30 | YES @ 89¢ | 91% | Edge (~2 pts) inside spread+fees, and PortWatch publishes the whole Jul 20–26 window at the exact minute of close (Tue 13:00 UTC) — with three tankers struck inside the window on Jul 20, the collapse scenario can't be priced from public data. |
| KXHORMUZWEEKLY-26JUL26-T50 | YES @ 15¢ | 35% | ⛔ Sub-35¢ tail; also the price collapsed 71→15 in 48h with zero published window data — informed or derisking flow either way, and the prior loss on this exact resolver (PortWatch printing 28 vs a tracker's 51) argues for humility. |
| KXHORMUZPEAK-26JUL26-T10 | YES @ 70¢ | 80% | 5 of the last 7 published days beat the strike, but a 4–5¢ spread over a 6-contract book fails the ≥3¢-slippage screen. Watchlist. |
| KXHEGSETHOUT-26APR-SEP01 | NO @ 92¢ | 95% | Defensible on its own — Hegseth is firing others, not being fired — but the feed already carries Hegseth-out-Aug-1 NO live. Adding the Sep 1 leg is the correlated double-bet (cf. the Kash Patel blow-up) the v3 cluster rule exists to prevent. |
| KXPARDONSTRUMP-26JUL-17 | YES @ 92¢ | 89% | July stands at 11 individuals / 17 grants — in-bucket under either counting convention — but clemency drops in Friday batches, Jul 31 is a Friday, the Jul 3 batch alone was 17 grants, and R. Kelly's commutation is pending. Negative edge at the ask. |
| KXLAKECONF-26MAY12-SEP01 | NO @ 95¢ | 90% | The recess wall (out Aug 7 → Sep 14) supports NO, but Trump demanded the Senate stay in session "until the entire Executive Calendar is CLEAR" on the day of this audit, and Thune is "open." 5¢ of upside doesn't pay for that live tail. |
| KXTRUMPNUMSTATES-26AUG01-E10 | NO @ 28¢ (lean only) | — | Stage-2.5 resolver-opacity reject: public-schedule reconstruction finds 7 firm July states (SD, PA, NY, NJ, GA, DE, MI) + 2 ambiguous (VA golf, MD transits) against a market confidently pricing "exactly 10" at 72¢. The VISITAREA tally can't be independently verified; deductions exceed the 15-pt cap. |
| KXTRUMPNUMSTATES-26AUG01-E9 | YES @ 12¢ (lean only) | — | Same resolver opacity — no leg of the ladder is priceable without the resolver's own count; sub-35¢ besides. |
| KXTRUMPUFC-26AUG-DJT | YES @ 15¢ | 35% | ⛔ Sub-35¢ tail. UFC 330 is confirmed for Aug 15 in Philadelphia and Trump attends big cards regularly, but there is zero attendance signal either way. |
| KXAPRPOTUS-26JUL31-40.9 | YES @ 32¢ | 33% | ⛔ Sub-35¢ tail, and the status-quo bucket is structurally fragile: Pew's 34 aging out of the RCP window jumps the average ~+0.5 — out of this bracket — with no news at all. |
| KXAPRPOTUS-26JUL31-40.6 | YES @ 18¢ | 20% | Fairly priced: needs only a normal dip, but the composition drift fights it. No edge. |
Stage-1 judgment cuts (not shadow-logged, per policy): Truth Social weekly post-count buckets, Truth-Social posting-hour markets, and endorsement-count markets — behavior-noise series with no researchable prior.
6. Sources
- RCP approval average, daily history & poll table (primary): realclearpolitics.com/poll/race/8656/historical_data.json · polling_data.json · Wayback snapshot Jul 24
- Approval context: Forbes, Jul 27 · CBS News/YouGov, Jul 26 · Emerson July national
- IMF PortWatch Hormuz daily data (primary, chokepoint6): PortWatch ArcGIS feature service · PortWatch methodology · straits.live (closure day-149, publication lag) · 2026 Strait of Hormuz crisis timeline · CNBC on June reopening spike
- CLARITY Act: Senate floor schedule, Jul 27 · Punchbowl, Jul 27 · Bloomberg, Jul 22 · CoinDesk, Jul 23 · Seven-Democrat statement, Jul 22
- Kari Lake / recess: nomination transmittal · SFRC 12–10, Jul 22 · Trump recess demand, Jul 27 · 2026 Senate calendar
- Election emergency: WaPo, Feb 26 · Democracy Docket draft text · Trump denial, Feb 27 · Jul 16 speech analysis
- Mamdani EOs: EO 19, Jul 24 · nyc.gov emergency EOs · EO tracker
- Pardons: DOJ Pardon Attorney clemency grants · WaPo, Jul 4 · R. Kelly petition
- Hegseth: CNN Jul 21 (testimony) · Jul 26 (flag promotions)
- UFC 330 / Trump travel: UFC.com · Roll Call Factba.se calendar · daily schedules
- Market data: Kalshi read-only mirror (
trading_events/trading_markets) and Kalshi public trade API v2 (markets, orderbooks, candlesticks), all queried live 2026-07-27.
Data: Kalshi mirror DB + Kalshi public API (prices/orderbooks as of 2026-07-27, ~16:30 UTC) + primary web sources cited above. All probabilities are subjective estimates; binary contracts resolve to $1.00 or $0.00 and any position can go to zero. EV figures are recorded for calibration only and did not determine position sizes (v3 policy). Rejected candidates are machine-logged and shadow-tracked to settlement. This is not financial advice.