Kalshi Politics — Mispricing Audit (Sentinel, managed)

Report date: 2026-07-20 · Category: Politics (category-match mode) · Horizon: 45 days · Capital: $1,000 · Methodology: v3.3 (conviction-weighted, EV-agnostic sizing; resolver-priced probabilities; sub-35¢ tails hard-banned; LOW conviction never deploys). Picks count toward the public track record and are cleared for auto-copy.

1. How this was researched

Mode: category-match. Queried the live Kalshi mirror for all active markets where trading_events.category = 'Politics', closing between now and +45 days: 366 markets across 134 events. (Note: this mirror carries no market_snapshots table — the schema endpoint is authoritative — so all prices, depth and 14-day histories below come from the Kalshi public API: live /markets, /orderbook, and daily/hourly candlesticks.)
Stage 1 mechanical filters (24h volume ≥ 1,500 · spread ≤ 5¢ · not fully priced 5-95¢ · no announcer-noise markets · no ticker already live in the feed): 366 → 22 candidates. Cuts: 285 on volume, 21 noise, 13 already-published picks, 3 spread, 21 fully priced, 1 no quotes. The 35 already-published contracts in $EXISTING_PICKS were excluded from candidacy up front.
Stage 2 diligence: rules text pulled from the mirror for every shortlisted candidate; live orderbooks fetched; 14-day candlesticks checked for pre-emptied edges; four parallel primary-source research threads run (Federal Register / whitehouse.gov / USTR / govinfo for tariff and EO actions; IMF PortWatch ArcGIS API for actual Hormuz daily transit data; RealClearPolitics daily history JSON; ActBlue directory raw HTML; congress.gov + wire reporting for CLARITY/FISA/Hegseth/UFC).
Stage 2.5 resolver profiling: every probability below is a forecast of the named resolver's verdict at resolution time, with resolver risks priced as explicit point deductions shown per pick.
Stage 3/3.5/4: 18 candidates survived to full diligence; 15 were rejected and logged (shadow-tracked to settlement); 4 deployed. Sizing is by conviction tier + entry band only — EV numbers are recorded for calibration but did not set a single contract count.

2. Markets reviewed

Ticker	Question	YES bid/ask	Vol 24h	Verdict
KXNEWTARIFFS-26JUL-AUG01	Trump orders tariffs in Jul 2026?	74 / 75	17,237	PICK — BUY YES
KXCRYPTOSTRUCTURE-26JAN-SEP	Crypto market-structure law before Sep 1?	17 / 18	3,122	PICK — BUY NO
KXFISAEXTEND-26JUL-26SEP01	FISA 702 reauthorized before Sep 1?	8 / 9	6,012	PICK — BUY NO
KXAPRPOTUS-26JUL24-40.9	RCP approval 40.8-41.0 on Jul 24?	31 / 32	3,024	PICK — BUY NO
KXTRUMPBIBIMEET-26AUG01-AUG01	Trump-Netanyahu meet before Aug 1?	61 / 62	7,686	REJECT — fairly priced
KXHORMUZWEEKLY-26JUL19-T50/T45/T75	Hormuz transit calls Jul 13-19 (3 legs)	50-94	22,773	REJECT — no data in window
KXHORMUZPEAK-26JUL19-T15/T20	Peak Hormuz day Jul 13-19 (2 legs)	28-77	3,253	REJECT — no data in window
KXHORMUZWEEKLY-26JUL26-T50	Hormuz transit calls >50, Jul 20-26?	29 / 31	2,079	REJECT — edge too thin
KXHORMUZNORM-26MAR17-B260901	Hormuz 7-day MA >60 before Sep 1?	10 / 11	38,711	REJECT — edge too thin
KXEOWEEK-26JUL18-1	>1 Executive Order signed Jul 12-18?	7 / 8	4,029	REJECT — resolver conflation risk
KXACTBLUETOP-26JUL24-JOSS	Ossoff top ActBlue fundraiser Jul 24?	76 / 77	1,817	REJECT — leaderboard archetype
KXTRUMPUFC-26AUG-DJT	Trump attends UFC 330?	13 / 14	3,776	REJECT — banned tail band
KXTRUMPNUMSTATES-26AUG01-E8	Exactly 8 states visited in Jul?	23 / 24	3,633	REJECT — unauditable count
KXAPRPOTUS-26JUL24-41.2 / 40.6 / 41.5	RCP approval ladder (3 more rungs)	11-36	5,785	REJECT — band/ladder discipline
KXHEGSETHOUT-26APR-SEP01	Hegseth out before Sep 1?	7 / 8	1,623	REJECT — thin, duplicative
KXTRUMPPHOTO-26JUL26-4	Getty photo-days count = 4?	21 / 22	1,807	CUT — noise metric, not diligenced

3. Picks — detail and thesis

Pick 1 — KXNEWTARIFFS-26JUL-AUG01 · BUY YES @ 0.77 limit · HIGH CONVICTION

"Will Trump order tariffs in Jul 2026?" — rules pay YES if Trump "has taken any executive action imposing new or increased tariffs on any country, provided the action specifies an effective date (even if such date occurs after expiration) in Jul 2026."

Current price 74 / 75 (last 75)

My probability 95%

Market implies ~75%

Entry band 60-90¢ favorite

Edge +18¢ vs limit (record only)

Cluster july-tariff-action

Mispricing thesis. The resolving event has already happened. On Jul 15, 2026 Trump signed a presidential memorandum directing a 25% Section 301 tariff on all goods of Brazil ("The Trade Representative shall impose a tariff of 25 percent on all goods of Brazil, with exemptions for certain goods"), with USTR's implementing notice setting the effective date at 12:01 AM ET, Jul 22, 2026. Both documents were published in the Federal Register today, Jul 20 (docs 2026-14654 and 2026-14542). A presidential memorandum is unambiguously "executive action"; it imposes a new tariff on a country; and it specifies an effective date.

Resolver (Stage 2.5): Kalshi settlement against the documentary record (Federal Register / whitehouse.gov / USTR). Mechanism: any qualifying action during July. The rules' one ambiguity — whether "in Jul 2026" modifies the action or the effective date — is moot here: the Brazil action was both signed (Jul 15) and takes effect (Jul 22) inside July. Redundant paths if Brazil were somehow disqualified: the global Section 301 "forced-labor" action (10%/12.5% on ~60 economies) is at final-action stage with completion trackers pointing to ~Jul 20-24, ahead of the Jul 24 Section 122 sunset; and the April pharma proclamation's first tranche takes effect Jul 31 (qualifies under the effective-date reading). Priced risks: −3 pts (resolver insists on a literal executive order, though the rules say "any executive action"), −2 pts (unknown-unknowns flagged by the intraday 94→75 fade). Evidence-based ~100 → model 95.

Evidence.

Presidential memorandum, signed Jul 15, 2026, FR doc 2026-14654 (govinfo mirror) — operative text quoted above. (govinfo)
USTR Notice of Action, FR doc 2026-14542: duties apply to Brazilian goods entered "on or after 12:01 eastern time on July 22, 2026." (govinfo; USTR PDF)
Contemporaneous coverage Jul 16: Washington Post, CNBC, NBC — "US launches 25% tariff on Brazil imports."
Second path queued: USTR global §301 final action expected before the Jul 24 Section 122 surcharge sunset (Fortune, Jul 16; USTR Jun 2 determination).

Tail risks (cleanest first). (1) Kalshi settlement reads "executive action" as EO-only and treats a §301 memorandum + USTR notice as agency action — this is the entire bear case and, I believe, why the market sits at 75 rather than 95+; the rules' own words ("any executive action") argue against it. (2) The Brazil action is suspended/withdrawn before settlement — it takes effect in 2 days; even then, the action was already "taken." (3) Settlement delay drags to Aug 1 close — carries no probability cost, only time.

Numbers. 190 contracts at ≤77¢ = $146.30 max cost (14.6% of capital, at the 15% HIGH/cluster cap). Max payout $190.00; EV +18¢/contract, +23.4% (recorded, not used for sizing).

Liquidity / entry context. Book at fetch: YES 74 bid / 75 ask; 253 contracts resting at ≤77 (1@75, 120@76, 132@77); 24h volume 17,237; OI 29,280. Slippage to fill 190 ≈ 1.5¢ — under the 3¢ screen.

Price history (14d). Traded 34-45¢ through Jul 15 (pre-announcement), gapped to 49 on Jul 16 (Brazil news), whipsawed 22-96 on Jul 17-19 as traders argued the definitional question, last 75-80 today on 17.9k volume. The move ate part of the edge (Stage-3 check acknowledged) — but 75 vs a mechanically-satisfied rule is still the widest defensible gap on the board.

Pick 2 — KXCRYPTOSTRUCTURE-26JAN-SEP · BUY NO @ 0.83 limit · HIGH CONVICTION

"Will a crypto market structure bill become law before Sep 1, 2026?"

Current price NO ask 83 (YES 17/18)

My probability (NO) 94%

Market implies ~83%

Entry band 60-90¢ favorite

Edge +11¢ (record only)

Cluster crypto-market-structure

Mispricing thesis. "Becomes law before Sep 1" requires the full chain — Senate passage, House passage of the same text, presidential signature — and the published congressional calendar has nearly closed it. The Senate has ~2.5 working weeks left (recess ~Aug 7, returns Sep 14); the House leaves at the end of July and returns Aug 31. As of Jul 17, the merged CLARITY draft has zero Democratic supporters and needs 60 votes; three Democrats (Murphy, Van Hollen, Merkley) formally oppose over the dropped ethics provision restricting officials' crypto business ties. Even a stunning Senate breakthrough leaves the House one day (Aug 31) to pass an untouched Senate text and get a signature before Sep 1.

Resolver (Stage 2.5): statutory fact — "becomes law before Sep 1, 2026." Clean binary; no snapshot risk. Definitional boundary: any "crypto market structure bill" counts, not just CLARITY — I checked for stealth vehicles (riders on must-pass bills) and found none moving in the window. Priced risks: −2 pts (surprise bipartisan deal + compressed House action Aug 31), −1 pt (unnoticed rider). Evidence-based ~97 → model 94.

Evidence.

Bill sits at Senate Calendar No. 423; no cloture filed, no floor vote scheduled (Yahoo Finance/Coinspeaker, early Jul; TechTimes Jul 4).
Jul 14-17: merged draft omits the ethics provision; Murphy/Van Hollen/Merkley formally opposed; Jul 17 text release lands with "no Democrats on board and 60 votes to find" (TechTimes Jul 15-16; Disruption Banking Jul 17).
House 2026 calendar: August recess with return Aug 31-Sep 3 (Roll Call Nov 18, 2025; Bloomberg Government). Senate: recess ~Aug 7, return Sep 14 (Senate.gov 2026 schedule).
CoinDesk (Jul 9) assesses near-term passage unlikely in a House "nearly paralyzed by Republican infighting."

Tail risks. (1) Cleanest: Thune makes it the pre-recess centerpiece, 7+ Democrats flip on a restored ethics title, the House delays recess — every link must click within ~10 working days. (2) Mark-to-market pain: a scheduled Senate floor vote could spike YES to 30-40¢ (it printed 35 intraday on Jul 19) without changing the Sep-1 math — hold through it. (3) Feed-correlation note: the feed already carries the Aug-1 leg (NO); this Sep-1 leg is implied by the Aug leg's failure mode and tightens the same thesis rather than diversifying it.

Numbers. 180 contracts at ≤83¢ = $149.40 max cost (14.9%, at the 15% HIGH/cluster cap). Max payout $180.00; EV +11¢, +13.3%.

Liquidity / entry context. NO side: 252 contracts at 83, 1,000+ at 85; 24h volume 3,122; OI 30,486. Fill at 83 with no slippage for 180.

Price history. YES ran 7→18 over Jul 14-16 on the new-text news, spiked to 35 intraday Jul 19, settled back to 15-18. The optimism spikes are exactly what the NO entry sells.

Pick 3 — KXFISAEXTEND-26JUL-26SEP01 · BUY NO @ 0.92 limit · HIGH CONVICTION

"Will legislation that reauthorizes FISA Section 702 authority become law before Sep 1, 2026?"

Current price NO ask 92 (YES 8/9)

My probability (NO) 95%

Market implies ~92%

Entry band >90¢ deep favorite

Edge +3¢ (record only)

Cluster fisa-702-reauth

Mispricing thesis. Section 702 authority lapsed Jun 12, 2026 after the House rejected a further short-term extension 198-218 (Jun 11) — Democrats protesting Bill Pulte as acting DNI, plus 7 GOP warrant-reform holdouts. No reauthorization bill is moving. Critically, the incentive to rush is gone: FISC certifications approved in March 2026 keep collection running until March 2027 despite the lapse. Trump has also linked any extension to the SAVE America Act, adding a second veto point. A law before Sep 1 requires a deal materializing and passing both chambers in the ~2 pre-recess weeks — no reporting points to one.

Resolver (Stage 2.5): statutory fact — reauthorization "become[s] law after Issuance and before Sep 1." Clean binary. Boundary: a short-term extension would count as "legislation that reauthorizes" — that is the real risk channel, not a full RISAA-style bill. Priced risks: −2 pts (Clayton's DNI confirmation, committee vote expected this week, removes the Democrats' Pulte objection and unlocks a fast extension), −1 pt (rider on a must-pass vehicle). Evidence-based ~98 → model 95.

Evidence.

House rejected extension 198-218, Jun 11 (Axios); authority expired Jun 12 (NPR; EFF).
Brennan Center 702 resource page (updated Jun 10): March-2026 FISC certifications valid to March 2027 — collection continues through the lapse.
Trump conditions extension on SAVE America Act linkage (ASA, Jun 17). Clayton DNI hearing Jul 15; committee vote week of Jul 20 (CNBC Jul 15, NPR Jul 16).
Congressional calendar: House out ~Aug 1-Aug 31, Senate out ~Aug 7-Sep 14.

Tail risks. (1) Cleanest: Clayton confirmed this week → Democrats' stated objection evaporates → leadership attaches a clean short-term extension to something moving before Jul 31. That path is real, which is why this is 95 and not 98. (2) Market microstructure: YES whipsawed 15→26→6 in the last 24h on ~6k volume — somebody is trading rumors; the fade back to 6-9 says nothing materialized, but watch it. (3) Thin-edge acknowledgment: +3¢ on a 92¢ entry is the floor of what v3 deploys; sized accordingly small.

Numbers. 100 contracts at ≤92¢ = $92.00 max cost (9.2%; deliberately under the 15% HIGH cap — deep-favorite band, don't over-allocate chasing cents). Max payout $100.00; EV +3¢, +3.3%.

Liquidity / entry context. NO side: 5,766 contracts resting at 92 — deepest book on the slate. 24h volume 6,012; OI 19,086. Zero slippage.

Price history. YES traded 42→20 on Jul 17 (12.4k vol), 15 through Jul 19, spiked to 26 overnight and collapsed to 6 by noon today. Net flow direction agrees with NO.

Pick 4 — KXAPRPOTUS-26JUL24-40.9 · BUY NO @ 0.70 limit · MEDIUM CONVICTION

"Will the President's approval rating be between 40.8 and 41.0% at 11:00 AM ET on July 24, 2026, according to RealClearPolitics?"

Current price NO ask 69 (YES 31/32)

My probability (NO) 79%

Market implies ~69%

Entry band 60-90¢ favorite

Edge +9¢ (record only)

Cluster rcp-approval-jul24

Mispricing thesis. The RCP average is 41.2 today (RCP's own data feed, build stamped Jul 20 09:28 ET) and has been at or above 41.0 every day since Jul 11, grinding up ~+0.05/day (40.3 on Jul 6 → 41.2 on Jul 20; disapprove fell 57.1 → 55.6). For YES to hit, the average must fall a net 0.2-0.4 in four days against the trend — and the mechanical poll-window rolloff points the other way: the two oldest, lowest polls in the 14-poll window (Quinnipiac 38, FT 36) are next to age out, which pushes the average up. Daily moves this month have never exceeded 0.2. The market charges 31¢ for a bucket the average is actively drifting away from.

Resolver (Stage 2.5): the displayed RCP average at 11:00 AM ET Jul 24 — a mechanical snapshot of a slow arithmetic series I can recompute from its 14 components. This is snapshot-risk territory, but unlike a leaderboard, the quantity moves in ≤0.2 daily steps and its inputs are public. Priced risks: −3 pts (a single shock poll can move the average ~0.2-0.4; RCP's intraday add-timing near the 11 AM read is not fully predictable). Evidence-based ~82 → model 79.

Evidence.

RCP daily history JSON: Jul 6-20 series 40.3, 40.5, 40.7, 40.8, 40.9, 41.0, 41.0, 40.9, 41.1, 41.1, 41.0, 41.1, 41.1, 41.1, 41.2 (realclearpolitics.com historical_data.json, fetched today).
Component polls verified: 14 polls averaging exactly 41.21/55.57 (Rasmussen 44, Morning Consult 45, RMG 44, 2Way/HarrisX 45 at the top; Quinnipiac 38 and FT 36 the stalest).
Wayback snapshot Jul 6 (40.4/57.6) corroborates the series' provenance.

Tail risks. (1) Cleanest: one bad poll (a 37-38 print from an Echelon/YouGov-type house) lands Jul 22-23 and parks the average at exactly 40.8-41.0 — this is a ~20% scenario, priced. (2) Today's intraday spike to 56¢ on the YES side says someone disagrees; no new poll visible behind it as of the 09:28 build. (3) Ladder note: this shorts one rung of a mutually-exclusive ladder — adjacent-rung NOs cannot both lose, and I logged (not deployed) the correlated 40.6-rung NO.

Numbers. 100 contracts at ≤70¢ = $70.00 max cost (7.0%, at the 7% MEDIUM cap). Max payout $100.00; EV +9¢, +12.9%. Resolves in 4 days.

Liquidity / entry context. NO side: 52 at 69, 509 at 70 — 561 fillable at ≤70. 24h volume 3,024; OI 7,575.

Price history. Only 3 days of trade (new weekly ladder): YES 22-29 through Jul 19, spiked to 56 intraday today before closing 31. The spike, absent a visible poll, reads as ladder noise rather than information.

4. Recommended $1,000 portfolio

#	Ticker	Action	Limit	Contracts	Max cost	% cap	Conviction / band (set the size)	Max payout	EV ¢	EV %
1	KXNEWTARIFFS-26JUL-AUG01	BUY YES	77	190	$146.30	14.6%	HIGH · 60-90¢ favorite	$190.00	+18	+23.4%
2	KXCRYPTOSTRUCTURE-26JAN-SEP	BUY NO	83	180	$149.40	14.9%	HIGH · 60-90¢ favorite	$180.00	+11	+13.3%
3	KXFISAEXTEND-26JUL-26SEP01	BUY NO	92	100	$92.00	9.2%	HIGH · >90¢ deep favorite (kept small)	$100.00	+3	+3.3%
4	KXAPRPOTUS-26JUL24-40.9	BUY NO	70	100	$70.00	7.0%	MEDIUM · 60-90¢ favorite	$100.00	+9	+12.9%
Totals				570	$457.70	45.8%	Cash held: $542.30 (54.2%)	$570.00	—	+14.4% blended

EV disclosure: blended EV is +$66.00 (+14.4% on deployed cost). Per v3, these EV figures are recorded for calibration only — no contract count above was derived from EV; sizes come from the conviction-tier caps (HIGH ≤15%, MEDIUM ≤7%) and entry bands.

Cluster exposure (cap: 15% of capital per cluster)

Cluster	Cost	% of capital	Cap	Note
july-tariff-action	$146.30	14.6%	15%	Single leg
crypto-market-structure	$149.40	14.9%	15%	Single leg in this run; note the feed's existing Aug-1 NO rides the same thesis
fisa-702-reauth	$92.00	9.2%	15%	Single leg
rcp-approval-jul24	$70.00	7.0%	15%	One rung only; correlated 40.6-rung NO logged as reject

Crypto-NO and FISA-NO both lean on the August-recess calendar, but the bills, vote math, and failure modes are independent — passing one implies nothing about the other — so they are scored as separate clusters. Their combined 24.1% is noted here for transparency.

Conviction exposure

Tier	Cost deployed	% of capital	Picks
HIGH	$387.70	38.8%	3
MEDIUM	$70.00	7.0%	1
LOW	$0.00	0%	0 (never deploys, v3.3)

Risk profile

Worst case (all four lose): −$457.70. No single cluster can flip the book by more than $149.40 (14.9%).
Best case (all four win): +$112.30 net (+$570.00 payout on $457.70 cost, +24.5% on deployed, +11.2% on total capital).
Most likely (probability-weighted): picks 1-3 win, pick 4 is the coin-most leg at 79% — modal outcome all-win; expected value ≈ +$66. The single most damaging plausible event is a Kalshi EO-only reading of "executive action" (pick 1, −$146.30); I assess it ≤5%.
Concentration: three of four picks are resolver-mechanical (a published federal document, two statutory deadlines against a published congressional calendar); one is a slow-moving polling average. No pick depends on a third-party tracker, a leaderboard snapshot, or unpublished data — the three archetypes that produced the historical losses.

Execution notes

Limits, not market orders. Pick 1: work 190 at 76-77 (book holds 253 ≤77; expect ~1.5¢ average slippage; do not chase above 80 — the thesis does not improve with price). Pick 2: 180 at 83 fills against 252 resting. Pick 3: 100 at 92 fills instantly (5.7k deep). Pick 4: 100 at 69-70; resolves Friday.
Invalidation triggers. Pick 1: any Kalshi clarification implying EO-only reading → exit; conversely the global §301 final action landing (expected by ~Jul 24) is a free confirmation. Pick 2: a scheduled Senate cloture vote on CLARITY with ≥5 Democratic yes-signals → reassess; a mere floor-vote headline spike is noise to hold through. Pick 3: Clayton confirmed AND a clean 702 extension attached to a moving vehicle → exit immediately (this is the one rumor-sensitive leg; the overnight 26¢ spike shows the market probes it). Pick 4: any new national approval poll ≤38 released before Thursday → exit at market.
Watchlist (not deployed). Hormuz Jul 20-26 weekly NO at ≥75 becomes interesting if PortWatch's Tuesday drop confirms sub-8/day traffic; UFC 330 YES stays banned but is the shadow-track case to watch; ActBlue re-examine only if Kalshi ever moves the series off the leaderboard snapshot.
Cash reserve: $542.30 (54.2%) held — well above the 10% minimum. An efficient week: most of the board was fairly priced, and the discipline is the point.

5. What I rejected and why

All 15 Stage-2 survivors below are machine-logged in picks.json and shadow-tracked to settlement — if the rejects outperform the picks, the screen adds nothing, and we want to know that.

Ticker	Side I'd take	Price	My prob	Reject reason
KXTRUMPBIBIMEET-26AUG01-AUG01	YES	62	65	Fairly priced. No meeting scheduled; Axios (Jul 16): WH says "no meeting has been scheduled," Netanyahu "trying to will a meeting into existence." Both expected at the Graham funeral Jul 28 — but whether a funeral-sideline encounter "counts" as meeting in person is a definitional dispute I can't price (Stage 2.5 screen).
KXHORMUZWEEKLY-26JUL19-T50	NO	50	—	Entry-band screen (coin-flip). PortWatch has published ZERO days of the Jul 13-19 resolution window; the data drops Tue Jul 21 ~9 AM ET — the exact minute the market closes. Betting it is guessing. This cluster produced the program's worst resolved loss (a resolver-lag tail, −75%).
KXHORMUZWEEKLY-26JUL19-T45	YES	91	90	No published data in the window; no edge at 91¢ against an IRGC total-traffic ban declared Jul 19 and last-print 10 calls/day.
KXHORMUZWEEKLY-26JUL19-T75	NO	94	88	Escorted-convoy spike days (28/day on Jul 6-7) make >75 live; likely negative edge at 94¢.
KXHORMUZPEAK-26JUL19-T15	YES	77	—	Unpriceable: not one day of the resolution window is published.
KXHORMUZPEAK-26JUL19-T20	YES	29	—	Entry-band screen: sub-35¢ tail (hard reject, v3.2) — and no data besides.
KXHORMUZWEEKLY-26JUL26-T50	NO	71	74	Directionally right (IRGC ban, war-risk premiums at 3-10% of hull value, 9 ships attacked since Jul 6) but ~3¢ of edge is too thin for MEDIUM conviction, and a mid-week ceasefire rebound is live.
KXHORMUZNORM-26MAR17-B260901	NO	90	92	7-day MA is 16.4 and falling; even June's post-MOU recovery peaked near MA 29 vs the 60 threshold. But ~2¢ edge in the deep-favorite band isn't worth more Hormuz exposure.
KXEOWEEK-26JUL18-1	NO	93	95	Verified 0 EOs signed Jul 12-18 (Federal Register API, whitehouse.gov, American Presidency Project all agree — most recent EO is 14414, Jun 25). But the White House's own video titled the Jul 13 monument proclamations "President Trump Signs an Executive Order" — that press-vs-statute conflation is exactly the Stage-2.5 risk, and it eats the ~2¢ of edge at 93¢.
KXACTBLUETOP-26JUL24-JOSS	YES	77	70	Resolver risk — the leaderboard archetype. Resolves off the leftmost card of ActBlue's directory "top fundraisers this week" module at one instant. Ossoff is #1 today and won last week, but the module demonstrably churns (El-Sayed traded 70¢ on Jul 16; Talarico 51¢ on Jul 17) and El-Sayed's Aug 4 primary is a textbook pre-primary-surge catalyst. Deductions exceed the ~15-pt cap → auto-reject. This is the same shape as the program's costliest historical loss.
KXTRUMPUFC-26AUG-DJT	YES	14	45	Entry-band screen: sub-35¢ tail — hard reject, no exceptions (v3.2). Logged loudly: Trump attended UFC 327, hosted UFC Freedom 250, and was -500 to attend UFC 329; UFC 330 is a title card in swing-state Philadelphia in Bedminster season. My 45 vs 14 is the largest modeled edge on the board — and it stays on the shadow track, where it can argue for the band's reinstatement with data instead of capital.
KXTRUMPNUMSTATES-26AUG01-E8	NO	76	—	Bespoke VISITAREA counting rules, 11 days of unknowable travel left, no reliable public tracker → resolver deductions unpriceable (Stage 2.5 screen).
KXAPRPOTUS-26JUL24-41.2	YES	36	42	Entry-band screen: 35-60¢ coin-flip without HIGH conviction. The average sits in this bucket today, but four days of churn on a 0.3-wide bucket is a coin toss.
KXAPRPOTUS-26JUL24-40.6	NO	89	95	Real ~6¢ edge (needs a 0.5+ drop in 4 days against an uptrend — hasn't happened all month) but it's the same ladder as the deployed 40.9 rung; cluster discipline kept the single larger-edge rung.
KXHEGSETHOUT-26APR-SEP01	NO	93	94	No departure reporting in a month (last concrete pressure: the Jun 17 Senate travel-budget fight); wartime incumbency. But ~1¢ edge, and the feed already carries the Aug-1 NO on the same actor.

Entry-band screen scorecard: the Stage-3.5 screen alone cut two sub-35¢ tails (UFC 330 YES, Hormuz PEAK-T20) and one coin-flip (APRPOTUS-41.2 YES). The UFC reject is the one to watch on the shadow track — it is the honest cost of the tail ban.

6. Sources

Brazil §301 memorandum (FR 2026-14654): govinfo.gov · USTR Notice of Action (FR 2026-14542): govinfo.gov · USTR final action PDF · USTR press release
Tariff context: CNBC Jul 16 · Washington Post Jul 16 · USTR global §301 determination Jun 2 · Fortune Jul 16 (§122 sunset Jul 24) · Pharma §232 proclamation (Apr; first tranche Jul 31)
CLARITY Act: Senate Banking May 14 · TechTimes Jul 15 · Disruption Banking Jul 17 · CoinDesk Jul 9 · House 2026 calendar (Roll Call) · Senate 2026 schedule
FISA 702: Axios Jun 11 (198-218 vote) · NPR Jun 12 (lapse) · Brennan Center (certifications to Mar 2027) · CNBC Jul 15 (Clayton)
RCP approval: RealClearPolling page · RCP data feed (trump_1st_ja JSON, build Jul 20 09:28 ET) · RCP historical_data.json daily series · Wayback Jul 6 snapshot
EO count: Federal Register EO list · whitehouse.gov presidential actions · American Presidency Project
Hormuz: IMF PortWatch chokepoints (daily data queried directly from the ArcGIS FeatureServer; latest published day 2026-07-12; weekly updates Tuesdays) · USNI Jul 13 (blockade) · CNBC Jul 17 (attacks) · The National Jul 17 (war-risk premiums)
Trump-Netanyahu: Axios Jul 16 · Times of Israel Jul 16 · Roll Call Jul 17 (funeral Jul 28)
ActBlue: secure.actblue.com/directory (raw HTML, trendingCampaigns props) · Kalshi series metadata
UFC 330: UFC.com (Aug 15, Philadelphia) · UFC Freedom 250 (White House card, Jun 14)
Hegseth: Washington Post Jun 17 · Military Times Jul 13
Market data: Kalshi mirror query API (schema-verified: trading_events, trading_markets; no snapshots table present) · Kalshi public API v2 (/markets, /orderbook, candlesticks), all fetched 2026-07-20.

Footnote. Data: Kalshi mirror DB (read-only query API) + Kalshi public trade API v2 (live quotes, orderbooks, candlesticks) + primary web sources linked above, all as of 2026-07-20. All probabilities are subjective estimates by the audit model; every contract can and regularly does resolve to zero — the worst case for this book is the loss of the full $457.70 deployed. EV figures are recorded for calibration and did not drive position sizing (v3 is EV-agnostic by design). Rejected candidates are logged and shadow-tracked to settlement. This report is not financial advice.