Kalshi Politics — Mispricing Audit (Sentinel, managed)

Report date 2026-09-14 · Methodology v3.3 (conviction-weighted, EV-agnostic sizing; resolver-priced probabilities; no LOW-tier or sub-35¢ deployments) · Category-match mode: trading_events.category = 'Politics' · Horizon 45 days · Capital $1,000 · Picks count toward the public track record and are cleared for auto-copy.

Bottom line. An efficient week. Of 566 live Politics markets, 20 survived the mechanical screen and got full diligence, and only two cleared the v3 conviction and entry-band gates — both legs of one thesis: IMF PortWatch has printed 286–301 Suez Canal transit calls in every one of the last eleven weeks, and the Sep 14–20 ladder prices the 270 rung at 78¢ and the 260 rung at 86¢. Both deploy as MEDIUM conviction in the 60–90¢ favorites band, capped at 7% each and 13.9% for the cluster. $139 deployed, $861 held as cash. The headline-grabbing markets — the Xi state visit, Netanyahu at the UN, the Hormuz weekly ladder, the Bab el-Mandeb ladder after the Houthi seizure of the strait — were all either fully priced or genuinely unknowable and are logged as rejects.

1. How this was researched

Mode: category-match. lower(trading_events.category) = 'politics' returned 464 active markets across 158 events closing within 45 days (the DB mirror's Politics category includes Kalshi's geopolitics and shipping-chokepoint series, which is why the Hormuz/Suez/Panama ladders appear here). No theme-match fallback was needed.
Schema check first. GET /schema showed trading_markets, trading_events, resolution_precedents and the signals tables, but no market_snapshots table. All live prices, 24h volume, open interest, orderbooks and 14-day candlesticks therefore came from the Kalshi public API (/markets, /markets/{t}/orderbook, /series/{s}/markets/{t}/candlesticks), pulled for all 566 markets in the 158 events.
Stage 1 filter: dropped volume_24h < 1500, spread > 5¢, YES bid ≥ 95 or YES ask ≤ 5, close beyond 2026-10-29, "what will he post / say" noise buckets, and the six already-published tickers in $EXISTING_PICKS (e.g. KXCRYPTOSTRUCTURE-26JAN-OCT, KXHORMUZNORM-26MAR17-B261001). 20 candidates remained.
Resolver-first diligence. For the PortWatch ladders I pulled the resolver's own dataset — the IMF PortWatch Daily_Chokepoints_Data ArcGIS feature service — for all of 2026 and rebuilt the Monday–Sunday weekly totals for the Suez Canal, Panama Canal, Bab el-Mandeb and Hormuz. I then reconciled those totals against every settled Kalshi rung since Aug 2 (via /markets?status=settled) to confirm the mapping and to measure how much the Tuesday settlement snapshot differs from later revisions. The resolution_precedents table's Hormuz entries (settles on the Tuesday 09:00 ET PortWatch republish; AIS-only; 48-hour de-dup) were used as the resolver profile.
News diligence via web search and fetch of primary/first-tier sources: Lloyd's List Intelligence Red Sea Brief (3 Sep), Al Jazeera and NPR on the Houthi capture of Mocha and Mayyun (11–13 Sep), the Wikipedia/SCMP record of the Sep 24 Xi state visit, reporting on Netanyahu's Sep 23–24 UN trip, NBC/ABC on the press-secretary vacancy, Time on Tillis vs Hegseth, and CNN/CBS on the UAP-file release cadence.
Sizing: conviction tier × entry band only. EV figures are reported for calibration and did not set a single contract count.

2. Markets reviewed

Stage-2 candidates (post mechanical filter). Prices are YES bid/ask in cents at ~15:30 UTC on 2026-09-14. "Eff. entry" is the price paid for the side I would take.

Ticker	Question (short)	YES bid/ask	Vol 24h	Side	Eff. entry	My prob	Disposition
KXSUEZWEEKLY-26SEP20-T270	Suez transits Sep 14–20 > 270	77/78	2,504	YES	78	86	PICK — MEDIUM
KXSUEZWEEKLY-26SEP20-T260	Suez transits Sep 14–20 > 260	85/86	1,776	YES	86	92	PICK — MEDIUM
KXHORMUZWEEKLY-26SEP13-T35	Hormuz transits Sep 7–13 > 35	43/46	5,512	NO	57	70	REJECT — coin-flip band
KXHORMUZWEEKLY-26SEP13-T30	Hormuz transits Sep 7–13 > 30	84/85	4,306	NO	16	45	REJECT — sub-35¢ tail
KXHORMUZWEEKLY-26SEP13-T40	Hormuz transits Sep 7–13 > 40	7/8	9,644	NO	93	95	REJECT — thin
KXHORMUZMAX-26SEP13-SEP11	Sep 11 is the peak Hormuz day	38/42	1,010	YES	42	—	REJECT — unpriceable
KXHORMUZWEEKLY-26SEP20-T30	Hormuz transits Sep 14–20 > 30	50/51	1,020	NO	50	55	REJECT — coin-flip, no edge
KXBABELMANDEBWEEKLY-26SEP13-T175	Bab el-Mandeb transits Sep 7–13 > 175	60/63	1,855	NO	40	60	REJECT — coin-flip band
KXBABELMANDEBWEEKLY-26SEP20-T175	Bab el-Mandeb transits Sep 14–20 > 175	27/28	1,029	YES	28	30	REJECT — tail
KXPANAMAWEEKLY-26SEP20-T200	Panama transits Sep 14–20 > 200	61/62	3,443	YES	62	63	REJECT — fairly priced
KXPANAMAWEEKLY-26SEP20-T225	Panama transits Sep 14–20 > 225	9/10	6,607	NO	91	88	REJECT — negative edge
KXXIUSA-26JUL07-OCT01	Xi in the U.S. before Oct 1	93/96	16,367	YES	96	95	REJECT — fully priced
KXNETANYAHUNYC-26JUL27-OCT01	Netanyahu in NYC before Oct 1	92/96	3,004	YES	96	93	REJECT — negative edge
KXHEGSETHOUT-26AUG-26OCT01	Hegseth out before Oct 1	4/6	2,722	NO	96	97	REJECT — thin
KXPRESSSECANNOUNCE-26AUG-OCT01	Next press secretary named before Oct 1	21/22	3,251	NO	79	82	REJECT — edge ≤3¢
KXUAPFILES-26AUG10-26OCT02	New UAP files released before Oct 2	63/68	1,590	YES	68	68	REJECT — no edge, 5¢ spread
KXTRUMPNOMNUM-26SEP13-T1	≥1 nomination sent to Senate this week	94/99	5,020	YES	99	—	Cut: 5¢ spread at 99, no room
KXAPRPOTUS-26SEP18-39.5	RCP approval 39.4–39.6 on Sep 18	25/27	3,398	—	—	—	Cut: 0.2-pt bucket, tail-band set
KXTRUTHSOCIAL-26SEP19-T240	>240 Truth Social posts this week	10/11	3,592	—	—	—	Cut: posting-count noise bucket
KXACTBLUETOP-26SEP18-SBRO	Sherrod Brown top ActBlue fundraiser Sep 18	6/7	2,291	—	—	—	Cut: tail, OI 614

Also seen but mechanically filtered (not logged as rejects): KXSUEZWEEKLY-26SEP20-T280 YES at 65/66 — the best-looking rung on the Suez ladder by edge, but 24h volume was zero and open interest 27, so it fails Stage 1 and sits on the watchlist (see execution notes). KXGOVTSHUTDOWN-26OCT01 (0/2), KXCLARITYVOTE-26JUL-OCT01 (95/99) and KXHOUSEADJOURN-26-26SEP26 (98/99) were fully priced.

3. Picks — detail and thesis

Pick 1 — KXSUEZWEEKLY-26SEP20-T270 · BUY YES @ 0.78 · MEDIUM CONVICTION

Will there be more than 270 transit calls through the Suez Canal from Sep 14 to Sep 20, 2026, as reported by IMF PortWatch?

Limit / current: 78¢ limit · 77/78 live (last 78)

My probability vs market: 86% vs 78% implied

Conviction · entry band: MEDIUM · 60–90¢ favorites

Cluster: suez-transits-week-sep14-20

Size: 89 contracts · cost $69.42 (6.9% of capital, MEDIUM cap 7%)

Max payout / EV: $89.00 · +8.0¢/contract · +10.3% · $7.12 edge (record only, did not set size)

Mispricing thesis. The market is pricing a generic left tail into a series that has not shown one. The resolver's own dataset has printed the following Suez weekly totals (Mon–Sun, revised series) since the Hormuz-closure regime settled in late June: 298, 295, 289, 291, 286, 291, 287, 286, 294, 301, 291. The minimum is 286, sixteen transits above the 270 rung; the standard deviation is about 5. The 78¢ ask says roughly one week in five breaks that floor. My read is that the crowd is charging a Red Sea risk premium after this weekend's Houthi seizure of Bab el-Mandeb without checking that Suez volumes were unmoved by the entire July–September Houthi attack regime, during which Bab el-Mandeb fell from 284 to 178 per week and Suez stayed flat at ~290.

Evidence.

IMF PortWatch Daily_Chokepoints_Data (queried 2026-09-14, latest published day Sep 6): Suez daily counts for Aug 31–Sep 6 were 30, 51, 41, 41, 56, 42, 30 = 291. Full-2026 weekly range 246–314; the only weeks ≤270 were in the Jan–Jun regime (246 holiday week, 266, 269, 269, 261, 265), before Saudi crude began rerouting north via Yanbu when Hormuz closed.
Lloyd's List Intelligence, Red Sea Brief, 3 September 2026: weekly Suez transits "290 ships between 24–30 August … on par with the 294 recorded the week prior … levels not seen since the start of 2024," driven by Saudi northbound rerouting ("partial loading at Yanbu and topping up at Sidi Kerir") plus bulkers, boxships, tankers and car carriers all up year-on-year. This is structural flow tied to the Hormuz closure and it does not cross Bab el-Mandeb.
Settlement reconciliation: every settled Suez rung since Aug 2 matches the PortWatch weekly total except one — KXSUEZWEEKLY-26AUG02-T290 settled NO although the revised series now shows 291 for Jul 27–Aug 2. So the Tuesday snapshot can sit a transit or two below the later-revised figure (and Panama's 26AUG23-T200 settled NO on a week now revised to 204). That is priced below.
The Sep 7–13 Suez ladder (closing tomorrow) has T280 at 92/93 and T290 at 61/62, i.e. the same crowd expects last week to print ~290 again. Nothing in the tracked data suggests a break in trend before the Houthi event.

Resolver profile (Stage 2.5). Source: IMF PortWatch Daily Chokepoint Transit Calls, Suez Canal (AIS-based, 48-hour per-ship de-duplication, UTC dates, republished Tuesdays 09:00 ET). Mechanism: Kalshi sums the seven UTC days Sep 14–20 from the Tuesday Sep 22 09:00 ET republish, which is also the market close (13:00 UTC). Definitional boundaries: "transit calls" is PortWatch's own count; no discretion. Priced risks: −4 pts (Houthi escalation spills into Suez volumes within the week, e.g. container lines re-suspend Red Sea strings), −2 pts (Tuesday snapshot undercount vs later revision), −1 pt (AIS/data outage or a missing day in the republish). Evidence-based 93% → model_prob 86%.

Tail risks, cleanest first. (1) Houthi control of Mayyun/Mocha turns into a sustained attack campaign this week and Maersk/MSC/CMA CGM pull the strings they only just restored — a 20+ ship drop would be needed, and the lag from decision to Suez count is typically several days, but the week is seven days long. (2) A wider Red Sea incident north of Bab el-Mandeb (a Yanbu-area strike like the Aug 24 VLCC hit) that halts Saudi northbound liftings. (3) A PortWatch publishing glitch that prints an incomplete week on Sep 22.

Liquidity / entry context. YES bid 77 (639 contracts), YES ask 78 (553 contracts fillable at the limit; next 100 at 81). 24h volume 2,504; open interest 2,332; market opened Sep 11 16:30 UTC. Price history (14d): first prints Sep 13 at 76¢; 2,272 contracts traded Sep 14 between 73¢ and 77¢ (mean 74.5¢); last 78¢. No ≥20¢ move; the price has drifted up 2–5¢ since open, so the edge has not been eaten but has narrowed.

Pick 2 — KXSUEZWEEKLY-26SEP20-T260 · BUY YES @ 0.86 · MEDIUM CONVICTION

Will there be more than 260 transit calls through the Suez Canal from Sep 14 to Sep 20, 2026, as reported by IMF PortWatch?

Limit / current: 86¢ limit · 85/86 live (last 86)

My probability vs market: 92% vs 86% implied

Conviction · entry band: MEDIUM · 60–90¢ favorites

Cluster: suez-transits-week-sep14-20 (same as Pick 1)

Size: 81 contracts · cost $69.66 (7.0% of capital, MEDIUM cap 7%)

Max payout / EV: $81.00 · +6.0¢/contract · +7.0% · $4.86 edge (record only, did not set size)

Mispricing thesis. Same series, one rung safer. 260 is 26 transits (about 9%) under the lowest week the current regime has produced and 31 under the mean. The 86¢ ask implies a one-in-seven collapse; the data support something closer to one-in-twelve even after charging for the Houthi shock. The two legs are deliberately taken together within one cluster: the 270 leg carries most of the edge, the 260 leg cushions the case where the week is merely weak rather than broken (a 261–270 print pays this leg and not the other).

Evidence. As for Pick 1 — the eleven-week PortWatch run of 286–301, the Lloyd's List account of why Suez is elevated (Saudi Yanbu → Suez/Sidi Kerir crude flow plus returning liner strings), and the settlement reconciliation showing Kalshi settles off the Tuesday PortWatch republish with at most a few-transit undercount.

Resolver profile (Stage 2.5). Identical source and mechanism to Pick 1. Priced risks: −2 pts (Houthi spillover; a 31-transit drop is needed here, so the same scenario is less likely to reach this rung), −1 pt (snapshot undercount), −1 pt (data outage). Evidence-based 96% → model_prob 92%.

Tail risks. Exactly the failure mode of Pick 1, at larger magnitude: a liner re-suspension plus a halt in Saudi northbound liftings inside the same week. Because the two legs share it, the cluster cap is the risk control — see Section 4.

Liquidity / entry context. YES bid 85 (1,615 contracts), YES ask 86 (1,494 contracts fillable at the limit; next 100 at 90). 24h volume 1,776; open interest 1,707; opened Sep 11 16:30 UTC. Price history (14d): all 1,776 contracts traded Sep 14, between 84¢ and 86¢; last 86¢. No large moves.

4. Recommended $1,000 portfolio

#	Ticker	Side	Limit	Conviction	Band	Contracts	Cost	Max payout	Model p	EV ¢	EV %	Cluster
1	KXSUEZWEEKLY-26SEP20-T270	BUY YES	78	MEDIUM	60–90¢	89	$69.42	$89.00	86%	+8.0	+10.3%	suez-transits-week-sep14-20
2	KXSUEZWEEKLY-26SEP20-T260	BUY YES	86	MEDIUM	60–90¢	81	$69.66	$81.00	92%	+6.0	+7.0%	suez-transits-week-sep14-20
Totals							$139.08	$170.00		$11.98	+8.6%

Total deployed $139.08 (13.9%) · cash held $860.92 (86.1%) · max payout $170.00 · max profit $30.92 · blended EV +8.6% · dollar edge +$11.98. EV figures are for the calibration record; they did not drive contract counts — the sizes are the MEDIUM-tier cap ($70 each) truncated to whole contracts.

Cluster exposure (cap 15% = $150 per cluster)

Cluster	Legs	Cost	% of capital	Cap	Status
suez-transits-week-sep14-20	T270 YES, T260 YES	$139.08	13.9%	15.0%	within cap
cash	—	$860.92	86.1%	—	—

Conviction exposure

Tier	Picks	Cost	% of capital	Per-pick cap
HIGH	0	$0.00	0.0%	15% ($150)
MEDIUM	2	$139.08	13.9%	7% ($70)
LOW	0	$0.00	0.0%	never deploys (v3.3)

Risk profile

Worst case (−$139.08, −13.9%): Suez prints ≤260 for Sep 14–20. Requires a ≥31-transit (≥10%) collapse from the regime mean inside a single week; nothing in 36 weeks of 2026 data comes close except the Dec 29 holiday week (246) in a lower-baseline regime. Probability by my model ≈ 8%.
Split case (−$69.42 + $11.34 = −$58.08): Suez prints 261–270. Model probability ≈ 6%.
Most likely case (+$30.92, +3.1% on capital, +22% on deployed): Suez prints >270, which the series has done every week since June 22. Model probability ≈ 86%.
Concentration: 100% of deployed capital rides one resolver (PortWatch), one chokepoint and one week. That is deliberate and disclosed: it is the only thesis that cleared the screens, it is held under the 15% cluster cap, and no single outcome can cost more than 13.9% of the book. The alternative — padding with Hormuz coin-flips or 96¢ visit markets — is precisely what the v1/v2 record penalised.

Execution notes

Limits, not market orders. Rest 78¢ on T270 (553 contracts sit at the ask, so a full fill is available now) and 86¢ on T260 (1,494 at the ask). Do not chase above 80¢ / 88¢ — beyond that the MEDIUM-tier edge is gone.
Invalidators (exit or stop adding): (a) any report that Maersk, MSC or CMA CGM re-suspend Suez/Red Sea transits this week; (b) a Houthi strike on a vessel north of Bab el-Mandeb (Yanbu/Jeddah approaches) after Sep 14; (c) the Tuesday Sep 15 PortWatch republish showing the Sep 7–13 Suez week under 280 — a first crack in the regime would cut my probability by 10+ points and the correct response is to sell into the bid, not hold.
Watchlist add: KXSUEZWEEKLY-26SEP20-T280 at 65/66 has better edge-per-cent than either pick but failed the Stage-1 volume screen (zero 24h volume, OI 27). If it develops a two-sided book with ≥1,500 daily volume before Wednesday, it is a MEDIUM candidate inside the same cluster — but only by replacing part of the T260 leg, since the cluster cap has just $11 of headroom.
Hedges: none needed at this size. The natural hedge — NO on Bab el-Mandeb Sep 14–20 T150 (YES 70/73) — is exposure to the same Houthi scenario and would only be worth it if the Suez position were much larger.
Settlement watch: PortWatch republishes Tue Sep 22 at 09:00 ET; Kalshi closes at the same instant and typically settles within the hour. Query the ArcGIS feature service for dates 2026-09-14 … 2026-09-20 and sum n_total.

5. What I rejected and why

All fourteen rejects below are machine-logged in picks.json and shadow-tracked to settlement, so the selection screen itself is testable. Entry-band cuts (Stage 3.5) are marked.

Ticker	Side I'd take	Price	My p	Reason
KXHORMUZWEEKLY-26SEP13-T35	NO	57	70	Stage 3.5 — coin-flip band, conviction not HIGH. Five of the last six PortWatch Hormuz weeks were ≤35 (39, 31, 35, 30, 22) and live trackers put mid-September at 3–8 ships/day, so NO looks right on the data. But the week has no published dailies yet, the peak-day ladder (Sep 11 at 38/42) says traders believe there was a convoy spike I cannot see, and PortWatch is AIS-only so a US-escorted convoy would print. Edge ≈14¢, below the 15¢ bar for a coin-flip, and I cannot claim HIGH.
KXHORMUZWEEKLY-26SEP13-T30	NO	16	45	Stage 3.5 — sub-35¢ tail, hard reject. The most "mispriced-looking" contract in the category (market 84% for >30 vs a 22-transit week just printed), and exactly the kind of resolver-lag tail (cf. `KXMAXSHIPSHORMUZ-26JUL31-AL50`) that v3.2 removed. Logged so the shadow record can argue.
KXHORMUZWEEKLY-26SEP13-T40	NO	93	95	Deep favorite, 2–3¢ of edge, and it would stack Hormuz exposure onto a book that already holds a Hormuz NO (`KXHORMUZNORM-26MAR17-B261001`, an existing pick).
KXHORMUZMAX-26SEP13-SEP11	YES	42	—	Single-day peak with no data; unpriceable. 4¢ spread.
KXHORMUZWEEKLY-26SEP20-T30	NO	50	55	Stage 3.5 — coin-flip, no edge. Recent weeks straddle 30 and an Omani reopening plan (CNN, Sep 13) is a live upside shock.
KXBABELMANDEBWEEKLY-26SEP13-T175	NO	40	60	Stage 3.5 — coin-flip band, conviction not HIGH. The prior two weeks printed 184 and 178 (barely over the line) and the Houthis took Mocha and Mayyun over Sep 10–13, mid-week. Sep 7–10 were probably normal (~25/day); Sep 11–13 could be anything. Resolver deduction exceeds what I can price — a textbook "great forecast on an unknowable resolver."
KXBABELMANDEBWEEKLY-26SEP20-T175	YES	28	30	Stage 3.5 — tail, and the market already reflects the post-seizure collapse. No edge either way.
KXPANAMAWEEKLY-26SEP20-T200	YES	62	63	Fairly priced. Last five PortWatch weeks: 210, 208, 187, 204, 210; weekly SD ≈ 10. The market's 62% is my 63%.
KXPANAMAWEEKLY-26SEP20-T225	NO	91	88	Negative edge. Three of the thirteen weeks since June exceeded 225 (228, 230, and 266 — the last one on the back of an 83-transit PortWatch glitch day on Jul 13), so the 9–10¢ YES is fair to cheap.
KXXIUSA-26JUL07-OCT01	YES	96	95	Fully priced. The state visit is scheduled for Sep 24 (invited at Busan, Oct 30 2025; confirmed during Trump's May 2026 Beijing visit). A 96¢ ask leaves nothing for a postponement in an active Iran war.
KXNETANYAHUNYC-26JUL27-OCT01	YES	96	93	Negative edge. Netanyahu is scheduled to arrive late Sep 23, speak Sep 24 and fly home the same night; but a same-day trip during an ongoing Iran/Lebanon campaign, with ICC-warrant routing constraints, has a 5–8% cancellation risk that 96¢ does not pay for.
KXHEGSETHOUT-26AUG-26OCT01	NO	96	97	Deep favorite, ~1¢. Tillis's Sep 2 call to replace him and Driscoll's resignation letter have not moved Trump, but there is nothing to earn.
KXPRESSSECANNOUNCE-26AUG-OCT01	NO	79	82	Edge ≤3¢ after resolver pricing. NBC: the White House "might not name a new press secretary until after the midterms"; the Sep 15 leg at 1/2 confirms no acting/interim designation yet. But the rules pay YES on any Trump announcement naming an acting or interim secretary (−4 pts), and Scott Jennings is a reported front-runner. Not defensible as a MEDIUM.
KXUAPFILES-26AUG10-26OCT02	YES	68	68	No edge, 5¢ spread. Cadence May 8 → May 22 → Jun 12 → Jul 10 → Aug 7 (gaps 14/21/28/28 days) is already ten days overdue with nothing since Aug 7; the "any federal agency, any UAP document" breadth helps YES but also makes the resolver's call hard to predict.

Discipline summary. Stage 3.5 cut four coin-flips and two tails, including the two contracts with the largest raw "edge" in the category (Hormuz T30 NO and T35 NO). That is the screen working as designed: the resolved record says those are where the book bled. If the shadow track shows these rejects settling in my favour at a rate that beats the picks, that is a finding for v4 — and it is now measurable.

6. Sources

IMF PortWatch — Daily Chokepoints Data (resolver dataset): dataset page · ArcGIS feature-service query endpoint (queried 2026-09-14 for 2025-12-29 … 2026-09-06)
Kalshi public API — settled KXSUEZWEEKLY markets, orderbooks and candlesticks for KXSUEZWEEKLY-26SEP20-T270 / -T260
Lloyd's List Intelligence — Red Sea Brief, 3 September 2026 (Suez 290/week; Bab el-Mandeb decline; Yanbu/Sidi Kerir rerouting; last Houthi attack Aug 24 off Yanbu)
Al Jazeera — Red Sea nations watch as Houthis seize Bab al-Mandeb strait (13 Sep 2026)
NPR — Houthi rebels seize a strategic port and island (11 Sep 2026) · Houthis capture a Red Sea island in threat to shipping (12 Sep 2026)
CNBC — Houthis deploy missiles and drones to attack ships in southern Red Sea (22 Jul 2026) · Al Jazeera — Six killed in Houthi attack on Bab al-Mandeb ship (12 Aug 2026)
CNN — Iran meeting with Gulf countries to discuss Strait of Hormuz deal (13 Sep 2026) · straits.live Hormuz tracker (cites PortWatch Sep 6 = 6 transits)
Wikipedia — 2026 state visit by Xi Jinping to the United States (scheduled from Sep 24) · SCMP — Xi's 2026 trip to US
5 Towns Central — Netanyahu to make brief New York trip for UN speech (14 Sep 2026)
NBC News — White House might not name new press secretary until after midterms · NPR — Karoline Leavitt is leaving (12 Aug 2026) · ABC — Who will replace Karoline Leavitt?
Time — Republican Senator calls on Trump to replace Hegseth (2 Sep 2026) · The Hill — Driscoll resignation
CNN — Pentagon releases fifth batch of UFO files (7 Aug 2026) · The Hill — fourth batch (10 Jul 2026) · CNN — first batch (8 May 2026)
Internal: resolution_precedents rows for KXHORMUZWEEKLY-26JUL19-T40 and KXHORMUZPEAK-26JUL19-T15 (resolver mechanism: Tuesday 09:00 ET PortWatch republish; AIS-only; 48-hour de-dup).

Data sources: Kalshi mirror DB (read-only query API; schema fetched 2026-09-14, no market_snapshots table present), Kalshi public trade API v2 (prices, orderbooks, candlesticks, settled results), IMF PortWatch ArcGIS feature service, and the news sources linked above. Disclaimer: all probabilities are subjective estimates by the author; prediction-market contracts resolve to $0 or $1 and the entire cost of a losing position is lost. EV and edge figures are recorded for calibration and did not determine position sizes. Nothing here is investment advice. Methodology v3.3; picks are cleared for auto-copy and count toward the public track record; rejects are shadow-tracked.