Kalshi Economics — Mispricing Audit

Report date 2026-07-08 · Category Economics (category-match mode) · Horizon 45 days · Capital $1,000 · Methodology v3.1 (conviction-weighted, EV-agnostic sizing; resolver-profiled probabilities). Picks count toward the public track record and are cleared for auto-copy.

1. How this was researched

Mode: category-match. trading_events.category = 'Economics' matched 1,139 active markets closing within 45 days on the live Kalshi mirror, so no theme-mode keyword search was needed. The latest 5-minute snapshot cycle was joined against those markets and the Stage-1 mechanical screen applied (24h volume ≥ 1,500, spread ≤ 5¢, not already ≥95/≤5, no announcer-noise markets, nothing on the already-published list). That left 42 liquid contracts collapsing into ~16 distinct questions. Each survivor got the full Stage-2 treatment: rules text pulled from the mirror, 14-day intraday price history, live Kalshi orderbook, and primary-source news diligence (four parallel research sweeps: CPI data, gasoline/oil, the Maine Senate situation, and Musk/misc).

Context that shaped everything this week: the US–Iran / Strait of Hormuz war re-escalated on July 7–8 (tanker attacks, US strikes on 80+ targets, Trump declaring the June 17 MoU "over"). WTI rose ~5–7% and RBOB ~6% today, and every Kalshi gas market repriced violently intraday — which put the entire gas complex behind the Stage-3 "price already moved in the last 48h" screen. Separately, the June 17 MoU had crashed oil during June (AAA retail fell $4.24 → $3.83), setting up a rare negative June CPI headline print. The deployable edge this run is concentrated in the July 14 CPI release.

Stage-2.5 resolver checks earned their keep: the single "safest-looking" candidate (gasoline CPI index > 342, offered at 97¢) turned out to resolve on the seasonally adjusted FRED series CUSR0000SETB01 — not the NSA series the headline suggests. On NSA math it was a 14%-cushion lock; on the actual SA series my point estimate lands at 344.2, two points from the strike. The pick was killed, and the same analysis produced a better trade on a sibling strike (Pick 3). Note: two Maine Senate nominee markets surfaced under the Economics category (upstream miscategorization); they were diligenced as found and rejected on their merits.

Already-published picks were excluded up front. The only direct overlap was the July Fed decision cluster (KXFEDDECISION-26JUL-H0 YES is live), so its siblings (H25, KXFED-26JUL-T3.75) were rejected as duplicates of a published thesis.

2. Markets reviewed

Market	Question	Price (bid/ask)	Verdict
KXCPICOREYOY-26JUN-T2.7	Core CPI YoY > 2.7% (June)	84/85	PICK 1 — BUY YES
KXCPIYOY-26JUN-T3.7	Headline CPI YoY > 3.7% (June)	66/67	PICK 2 — BUY YES
KXUSGASCPI-26JUL14-T344	Gasoline CPI (SA) > 344 (June)	84/85	PICK 3 — BUY NO (lottery)
KXUSGASCPI-26JUL14-T342	Gasoline CPI (SA) > 342 (June)	94/97	REJECT — resolver reads SA; "lock" is negative-EV at ask
KXUSGASCPI-26JUL14-T345	Gasoline CPI (SA) > 345 (June)	41/43	REJECT — coin-flip band
KXCPI-26JUN-T-0.2	CPI MoM > −0.2% (June)	33/36	REJECT — band screen (edge logged)
KXCPI-26JUN-T-0.3	CPI MoM > −0.3% (June)	78/91	REJECT — no edge at ask, spread blew out
KXCPINDEX-26JUL14-T334.700	CPI NSA index > 334.700 (June)	54/56	REJECT — band screen (edge logged)
KXECONSTATCPIYOY-26JUN-T3.8	CPI YoY exactly 3.8%	48/52	REJECT — definitional, priced fair vs ladder
KXAAAGASM-26JUL31-3.80	AAA gas > $3.80 on Jul 31	65/68	REJECT — moved +35¢ today (48h rule)
KXAAAGASW-26JUL13-3.880	AAA gas > $3.88 on Jul 13	36/38	REJECT — moved +32¢ today (48h rule)
KXMEDNOMJUL-26AUG01-TJAC	Troy Jackson is ME-Sen Dem nominee on Jul 31	47/48	REJECT — coin-flip band, fluid politics
KXMEDNOMJUL-26AUG01-SBEL / -NSHA	Bellows / Shah is nominee on Jul 31	20/21 · 16/17	REJECT — tail band / edge within error
KXMUSKNW-26JUL31-T750	Musk > $750B (Bloomberg, Jul 31)	90/91	REJECT — fairly priced
KXU3-26JUL-T4.2	U-3 > 4.2% in July	55/56	REJECT — coin flip, no edge
KXTBCRUNCHWRAP-26AUG02-*	Crunchwrap avg price (July)	63/64	REJECT — unknowable resolver (Stage 2.5)
KXFED-26JUL-T3.75 (+H25)	Fed July decision cluster	23/24	EXCLUDED — duplicates published pick

3. Picks — detail and thesis

Pick 1 — KXCPICOREYOY-26JUN-T2.7 · BUY YES @ 0.86 · HIGH CONVICTION

Will core CPI inflation be above 2.7% for the year ending June 2026? Current 84 / 85 · my probability 94% vs market-implied ~85% · entry band: 60–90¢ favorite (86¢) · cluster cpi-june-report.

Thesis. Core CPI YoY printed 2.9% in May (BLS, Jun 10). For NO, the printed June value must fall two tenths to ≤2.7 — requiring a June core MoM around +0.05% against a +0.23% year-ago base. Core has run +0.2 to +0.4 every month of this tariff-and-war inflation cycle (Apr +0.4, May +0.2), and the Cleveland Fed's June core nowcast is +0.23% MoM / 2.85% YoY — comfortably above the 2.75% rounding boundary. The June price data already exists; only the measurement is pending, and every input points the same way. Crucially, the June gasoline collapse does not touch core.

May core CPI YoY 2.9%, core MoM +0.2% — BLS CPI news release, Jun 10, 2026.
Cleveland Fed Inflation Nowcast (Jul 8): June core MoM +0.23%, core YoY 2.85% — clevelandfed.org.
Year-ago base: June 2025 core MoM ≈ +0.23% SA, so a flat-to-2.8/2.9 print needs nothing unusual.

Resolver: BLS CPI release, Jul 14 2026 8:30 ET; printed one-decimal core YoY vs 2.7; delay clause in rules. Priced risks: −1 pt (rounding-boundary measurement noise, release mechanics). Cleanest way to lose: a never-in-this-cycle collapse of June core to ≈0.0% MoM (used vehicles + shelter + services all soft at once).

Size: 81 contracts @ 86¢ limit = $69.66 (7.0% of capital; HIGH tier cap is 15%, cluster cap binds first) · max payout $81.00 · edge +8¢ · EV +9.3% (recorded, not used for sizing) · Liquidity at entry: yes 84/85, asks 107 @ 85 + 1,008 @ 86 (fill assured), vol24h 2,140, OI 3,043 · Price history: ground from ~50¢ (mid-June) to 84–86¢ — thesis partially absorbed; residual edge is the last 8–10¢ of a 94% event.

Pick 2 — KXCPIYOY-26JUN-T3.7 · BUY YES @ 0.67 · MEDIUM CONVICTION

Will headline CPI inflation be above 3.7% for the year ending June 2026? Current 66 / 67 · my probability 76% vs market-implied ~67% · entry band: 60–90¢ favorite · cluster cpi-june-report.

Thesis. YES needs the printed YoY ≥ 3.8, i.e. a June NSA index ≥ 334.66 (May: 335.123; base June-2025: 322.561) — equivalently June SA MoM above roughly −0.22%. Both credible nowcasts sit clearly on the YES side: Cleveland Fed −0.06% MoM / 3.92% YoY and Goldman −0.13% MoM. The Kalshi MoM ladder is pricing a median near −0.19%, materially more deflationary than either. And the crowd's own books disagree with each other: the specialist gasoline-index ladder (Pick 3's market) implies June SA gasoline ≈ −9% MoM, which through a ~3.2% weight contributes ≈ −0.29pp; with Cleveland's +0.23% core that arithmetic gives headline ≈ −0.08%, not −0.19%. The MoM book appears to be over-extrapolating the June gas crash beyond what the hard EIA data supports; this market is the cleanest favorite-band expression of that.

May CPI: YoY 4.2%, NSA index 335.123 — BLS release Jun 10 + BLS API CUUR0000SA0.
Cleveland Fed nowcast Jul 8: June headline MoM −0.06%, YoY 3.92%. Goldman preliminary: −0.13% MoM (Yahoo Finance).
June gasoline: EIA weekly regular retail fell $4.305 → $3.831 across June (eia.gov), fully captured in the nowcasts.

Resolver: BLS printed one-decimal YoY, Jul 14 8:30 ET; unambiguous, delay clause present. Priced risks: −2 pts (rounding boundary sits inside forecast error; possibility the MoM crowd knows a soft non-energy component I can't see). Those are inside the 76% number. Cleanest way to lose: CPI's own gasoline sample catches the intra-June collapse harder than EIA weekly averages imply, dragging SA MoM below −0.22%.

Size: 45 contracts @ 67¢ limit = $30.15 (3.0% of capital; MEDIUM tier cap 7%, cluster cap binds) · max payout $45.00 · edge +9¢ · EV +13.4% (recorded, not sizing) · Liquidity: yes 66/67, 134 @ 67 + 1,501 @ 68, vol24h 6,562, OI 156,347 · History: stable 65–73 for two weeks — no 48h-move contamination.

Pick 3 — KXUSGASCPI-26JUL14-T344 · BUY NO @ 0.16 · LOW CONVICTION · explicit asymmetric lottery

Will the June 2026 gasoline CPI index (SA, CUSR0000SETB01) print above 344? Current yes 84 / 85 (NO costs ~16¢) · my probability the NO wins: 45% vs market-implied ~16% · entry band: <35¢ tail — deployed under the single lottery slot, capped at 1.6% of capital · cluster cpi-june-report.

Thesis — a variance disagreement with an overconfident book. Hard inputs: May SA gasoline index 378.660 (FRED); June seasonal factor 1.0484 (stable to ±0.04% across 2023–25); EIA weekly retail gives a daily-weighted June/May decline of ≈ −9.1%. That maps to a June SA print of ≈ 344.2. The strike ladder prices a cliff — T344 at 84, T345 at 42 — i.e. the book claims to know the print to within ±0.6 index points. But the CPI's own-outlet gasoline sample historically deviates from EIA-weekly-implied projections by ±1.5 points; and a plausible Monday-only EIA construction (May avg $4.479 vs June $4.050 = −9.59%) lands at 342.3, six points below where the book is centered. At 16¢ the NO needs only ~16% to break even; my σ ≈ 1.6 around 344.2 puts it near a coin flip. This is priced as a lottery, not a conviction bet, precisely because the edge is a second-moment claim.

FRED CUSR0000SETB01: May 2026 SA = 378.660; CUUR0000SETB01 NSA = 396.961 (fredgraph.csv, fetched today).
June seasonal factor: NSA/SA = 1.0484 (2025), 1.0487 (2024), 1.0492 (2023); May 2026's factor matches May 2025 exactly → 2026 factors unchanged.
EIA weekly regular retail: May Mondays 4.452/4.500/4.490/4.475; June 4.305/4.146/4.052/3.914/3.831 (eia.gov LeafHandler, release Jul 7).

Resolver: FRED/BLS SA series named explicitly in rules_secondary — verified directly; this same check is what killed the T342 YES "lock". Priced risks: −3 pts equivalent (all-grades vs regular-grade mix, May daily-average construction). Cleanest way to lose: the specialists are right — CPI's June sample tracks the mid-month average (~345) rather than my daily-weighted path, and the print lands 344.5–345.5. Honesty note: this leg's YES was bid up from ~59 to 84 within the last hour — someone is actively buying the ladder today; and the market sits below the run's 1,500 vol24h screen (it entered via diligence on its liquid sibling T342). Both facts argue for the tiny size it got.

Size: 100 contracts @ 16¢ limit = $16.00 (1.6% of capital; ≤2% lottery cap, only tail this run) · max payout $100.00 · edge +29¢ · EV +181% (recorded — and pointedly NOT what sized it; it is the smallest position in the book) · Liquidity: yes 84/85, exactly ~100 contracts of NO available at 16¢ — size matched to the visible book, zero slippage tolerance · vol24h 557, OI 599 · History: traded 57–61 this morning, 84/85 now.

4. Recommended $1,000 portfolio

#	Market	Side	Limit	Contracts	Cost	% cap	Conviction / band	Max payout	Edge	EV%
1	KXCPICOREYOY-26JUN-T2.7	YES	86¢	81	$69.66	7.0%	HIGH · 60–90¢	$81.00	+8¢	+9.3%
2	KXCPIYOY-26JUN-T3.7	YES	67¢	45	$30.15	3.0%	MEDIUM · 60–90¢	$45.00	+9¢	+13.4%
3	KXUSGASCPI-26JUL14-T344	NO	16¢	100	$16.00	1.6%	LOW · tail (lottery slot)	$100.00	+29¢	+181%
Total deployed				226	$115.81	11.6%	Cash held: $884.19 (88.4%)	$226.00	+$39.53	+34.1%*

* Blended EV is dominated by the lottery leg's recorded +181%; per v3 policy EV numbers are calibration data only — every position size above came from conviction tier + entry band + cluster cap, and the highest-EV pick is deliberately the smallest.

Cluster exposure

Cluster	Picks	Cost	% of capital	Cap (15%)
cpi-june-report	1, 2, 3	$115.81	11.6%	$150 — OK (77% used)

Yes — the whole book rides one release (BLS CPI, Jul 14, 8:30 ET). That is a deliberate, capped concentration: it was the only place in this category where the screens found defensible edge this week. Two mitigants: total exposure is under the 15% single-thesis cap, and the legs are not all the same direction — Pick 3 (gasoline-soft) partially hedges Pick 2 (headline-firm): if gasoline surprises soft enough to kill the headline YES, the lottery pays 5.3:1 against it.

Conviction exposure

Tier	Cost deployed	% of capital	Per-pick cap
HIGH	$69.66	7.0%	15% ($150)
MEDIUM	$30.15	3.0%	7% ($70)
LOW	$16.00	1.6%	3% ($30) / 2% lottery

Risk profile

Worst case (−$115.81, −11.6%): soft core (≤2.7) AND soft headline (≤3.7) AND firm gasoline (>344) — an internally awkward combination (firm gas usually props the headline) I put at ~1–2%.
Realistic bad case (−$34.81): headline lands exactly 3.7 with gasoline ~344.5 — Picks 2 and 3 lose, Pick 1 wins.
Most likely outcomes: gasoline prints 344–346 → Picks 1+2 win, lottery loses: +$10.19 (~35%); gasoline prints ≤344 → all three win: +$110.19 (~36%).
Concentration: one macro release; capped at 11.6% of capital with an internal hedge leg. 88.4% cash is a position — this category was mostly war-driven chaos and fair prices this week.

Execution notes

All limit orders, GTC through Jul 13 (all three markets close at the Jul 14 release). Do not chase: core above 87¢ or headline above 68¢ erases the sized edge; the T344 NO is strictly 16¢ — the visible book is exactly ~100 contracts and the leg was bid up this hour, so partial fills are acceptable and slippage is not.
Invalidation triggers: (a) any BLS release-delay notice — exit nothing, but expect close-time extension per rules; (b) for Picks 2–3: EIA's monthly gasoline average or a revised weekly print implying a June/May decline shallower than −8.5% kills the T344 NO (and strengthens Pick 2) — rebalance toward headline YES; deeper than −10% strengthens the NO and endangers Pick 2.
Watchlist (no capital): KXCPI-26JUN-T-0.2 YES below 30¢ and KXCPINDEX-26JUL14-T334.700 YES below 50¢ — both carry modeled edge but are band-blocked this run; the Jul-31 AAA gas ladder once the Hormuz headline volatility settles (re-evaluate after ~Jul 11 under the 48h rule).

5. What I rejected and why

All 15 rejects below are machine-logged in picks.json and shadow-tracked to settlement — if the rejects outperform the picks, the screen itself is the finding. Stage-3.5 entry-band cuts are marked [BAND].

KXUSGASCPI-26JUL14-T342 — the audit's cautionary tale. At first read (NSA index 396.96 in May, strike 342): a 14%-cushion lock at 97¢. The rules resolve on the seasonally adjusted series: May SA is 378.66 and my June estimate is 344.2 ± 1.6 — the boundary is two points away, and YES at 97¢ is negative EV (model ~90%). This is the v1 leaderboard lesson replayed: right about the world, nearly wrong about the resolver. The residual NO-tail (model ~10% vs 6¢) was passed over — inside model error, 135 contracts of depth, and the lottery slot went to the better-shaped T344.
KXUSGASCPI-26JUL14-T345 NO @ 59¢ — [BAND] same variance thesis as Pick 3, but expressed in the 35–60¢ coin-flip band without lock-grade conviction. Model 69%.
KXCPI-26JUN-T-0.2 YES @ 36¢ — [BAND] my model says 64% (+28¢): the crowd's −0.19% MoM median looks too bearish against Cleveland (−0.06), Goldman (−0.13) and its own gasoline ladder. But a nowcast-vs-crowd dispute can never be HIGH conviction, and the 35–60 band requires HIGH. The shadow track will grade this screen.
KXCPINDEX-26JUL14-T334.700 YES @ 56¢ — [BAND] nearly the same boundary as Pick 2 (334.700 vs 334.657) trading 11¢ cheaper — a real cross-market inconsistency, but it sits in the coin-flip band, in a day-old market with a 232-contract book. Model 73%.
KXCPI-26JUN-T-0.3 YES @ 91¢ — model 91%, zero edge at the ask, and the live bid side collapsed to 78 (13¢ spread) while I watched.
KXECONSTATCPIYOY-26JUN-T3.8 — looked like a 25¢ mispricing vs the threshold ladder until the rules revealed it's an "exactly 3.8%" bucket, not a threshold. Priced consistently (implied 42–47 vs 48/52). Definitional, not exploitable.
KXAAAGASM-26JUL31-3.80 YES @ 68¢ and KXAAAGASW-26JUL13-3.880 YES @ 38¢ — the Hormuz re-escalation repriced these +32–35¢ today; whatever edge existed was eaten before this audit began (Stage-3 48h rule). What remains is a war-path coin toss (model 63% and 45% respectively) in thin books.
KXMEDNOMJUL-26AUG01-TJAC NO @ 53¢ — [BAND] after the Jul 6 assault allegation and Sanders' Jul 7 call to step aside, Jackson-at-47¢ requires both Platner's withdrawal by Jul 13 and a ~100-member state committee choosing the Platner-camp favorite the party just publicly rebuffed. Model 68% for the NO — but it's the coin-flip band, committee politics is not lock-grade, and it correlates with an already-published Platner pick. SBEL YES @ 21¢ (model 19%, fair) and NSHA NO @ 84¢ (model 88%, edge inside handicapping error) also logged.
KXMUSKNW-26JUL31-T750 YES @ 91¢ — Bloomberg printed $1.01T on Jun 30, but SPCX (3–5% float) produced a −29%/week drawdown in June; ~90% is fair. Model 90%, no edge.
KXU3-26JUL-T4.2 — genuine coin flip (June printed 4.2 on falling participation; payrolls decelerating; claims improving). No consensus exists yet, no defensible side.
KXTBCRUNCHWRAP-26AUG02-T6.67 — Stage-2.5 mandatory reject: resolves to a private vendor feed (Spice Data), no public history, first-ever settlement month of the series. Unknowable resolver > any forecast.
KXFED-26JUL-T3.75 / KXFEDDECISION-26JUL-H25 — same thesis as the already-live published pick (Fed holds in July); re-emitting would duplicate feed exposure. The H0/H25 pair sums to ~100¢ — no structural arb either.

6. Sources

BLS CPI news release (May 2026 report, Jun 10) — bls.gov/news.release/cpi.nr0.htm; release schedule — bls.gov/schedule; BLS API series CUUR0000SA0, CUUR0000SETB01.
FRED series data (fetched 2026-07-08) — CUSR0000SETB01 (gasoline CPI, SA), CUUR0000SETB01 (NSA).
Cleveland Fed Inflation Nowcasting (Jul 8) — clevelandfed.org; Goldman June CPI preview via Yahoo Finance.
EIA weekly retail gasoline — EMM_EPMR_PTE_NUS_DPG (release Jul 7, 2026); EIA July STEO — eia.gov/outlooks/steo.
AAA gas prices — gasprices.aaa.com; AAA newsroom (Jun 18, Jul 2 releases).
Hormuz re-escalation: Al Jazeera liveblog (Jul 8); CNN (Jul 8); Al Jazeera (Jun 17 MoU); Wikipedia: 2026 Strait of Hormuz crisis; TradingEconomics crude/RBOB quotes (Jul 8).
Maine Senate: BDN primary results; CNN allegation (Jul 6); PBS: Sanders (Jul 7); BDN: Jackson exploratory (Jul 7); Maine Public: replacement process; WaPo: deadlines (21-A M.R.S. §374-A).
Jobs: CNBC June jobs report; FRED UNRATE/PAYEMS/ICSA; DOL weekly claims.
Musk: Kalshi settled markets KXMUSKNW-26MAY31/-26JUN30 (Bloomberg $735B → $1.01T); Wikipedia: Wealth of Elon Musk; CNBC SpaceX IPO; Fortune (Jun 23).
Market data: Kalshi mirror query API (schema-verified; 1,139 Economics markets, latest 5-min snapshots) and Kalshi public API api.elections.kalshi.com/trade-api/v2 (rules_secondary, live orderbooks, Jul 8 ~16:00–16:30 UTC).

Data: Kalshi mirror DB (5-minute snapshots through 2026-07-08 16:23 UTC), Kalshi public trade API, BLS/FRED/EIA/AAA primary data, cited press. Method: v3.1 — conviction-weighted and EV-agnostic sizing, entry-band screen, 15% cluster cap, resolver-profiled probabilities with priced deductions; rejects machine-logged and shadow-tracked. Disclaimer: all probabilities are subjective estimates by an automated research process; prediction-market contracts can and do resolve to zero; nothing here is investment advice. Picks in this report enter the public track record and the auto-copy feed.