Kalshi Economics — Mispricing Audit

Report date 2026-06-24 · Horizon 45 days (through ~Aug 8) · Capital $1,000 · Methodology v3 (conviction-weighted, EV-agnostic) · Auto-copy track

Bottom line. Economics is an efficient category this week, and the one macro event that matters — the collapse of the mid-June oil/Hormuz war premium — is already reflected almost everywhere: gas, CPI-YoY, Treasury and payroll ladders are all priced around consensus. The single exception is the July FOMC "hold" contract, where the hike premium built up during the oil spike has not repriced after today's ~40% crude crash. That is the only edge I can defend. One pick, ~7% of capital deployed, ~93% held as cash. A near-empty book is the correct answer in an efficient week — padding it is what the historical record punished.

1 · How this was researched

Selection mode: category-match. $CATEGORY=Economics matched trading_events.category directly — 1,135 active markets close within 45 days, so no theme/keyword fallback was needed.

Schema first. Pulled /schema as source of truth (trading_markets, trading_events, market_snapshots).
Filter. Ranked the macro series by liquidity and grouped the 1,135 markets into ~80 series. Focused on the liquid, decision-relevant ladders closing in-horizon: July FOMC (rate decision + threshold), June/July U-3, June payrolls/ADP, June CPI/Core-CPI YoY & MoM, June PPI, Q2 GDP, month-end Treasury yields (Jun 30), AAA gas (Jun 30), Musk net worth (Jun 30). Cut volume24h<1500, spreads >5¢, fully-priced (≥95¢/≤5¢) and announcer-noise markets.
Diligence. For survivors: read the rules text, pulled 14-day intraday history from market_snapshots, hit the live Kalshi orderbook for depth, and grounded priors in primary sources (BLS, Federal Reserve, AAA/EIA, CNBC/Bloomberg market wires).
Screens (v3). Conviction tier set off defensibility (source quality + rule clarity + book depth), never off EV. Entry-band screen restricts coin-flips (35–60¢) to HIGH-conviction-only and defaults sub-35¢ tails to reject. Size set by tier + band; ev_pct/edge_cents are recorded for calibration only.

Macro backdrop established this run: current fed-funds upper bound 3.75% (3.50–3.75%, effective 3.63%); new Chair Kevin Warsh (sworn in May 22) ran a hawkish June 17 FOMC — easing language removed, 9 of 18 members projecting a hike this year, median dot 3.8% YE. May CPI +4.2% YoY (energy +23.5%), core +2.9%; May U-3 4.3%, payrolls +172k. Then on Jun 24 crude fell below ~$70–72 (−~40% from the wartime peak) as the Strait of Hormuz reopened and US–Iran talks advanced.

2 · Markets reviewed

Market	Ticker	Price (YES)	Read	Outcome
Fed holds 0bp · Jul 29	KXFEDDECISION-26JUL-H0	77¢	Hold premium stale vs oil crash	PICK · BUY YES
Fed +25bp · Jul 29	KXFEDDECISION-26JUL-H25	21¢	Inverse leg of the pick	REJECT (same cluster)
June U-3 > 4.1%	KXU3-26JUN-T4.1	92¢	Near-locked but thin book	REJECT (slippage)
June payrolls > +70k	KXPAYROLLS-26JUN-T70000	67¢	Coin-flip vs consensus	REJECT (no edge)
June CPI YoY > 3.6%	KXCPIYOY-26JUN-T3.6	92¢	Base-effect step-down priced	REJECT (efficient)
June core CPI YoY > 3.1%	KXCPICOREYOY-26JUN-T3.1	6¢	~20¢ spread, untradeable	REJECT (liquidity)
Gas > $3.85 · Jun 30	KXAAAGASM-26JUN30-3.85	44¢	Decline already priced	REJECT (coin-flip)
10Y yield > 4.45% · Jun 30	KXUST10M-26JUN30-T4.45	70¢	6-day yield noise	REJECT (coin-flip)
Q2 real GDP > 0.5%	KXGDP-26JUL30-T0.5	94¢	Deep favorite, illiquid	REJECT (no edge)

Also scanned and dismissed at the filter stage: Musk net-worth (Jun 30), June PPI YoY (tariff-distorted), core PCE (illiquid), ISM PMI, ADP, July U-3/threshold ladders (untraded), month-end 2Y/5Y/30Y, and the various non-US GDP/inflation series.

3 · The pick

KXFEDDECISION-26JUL-H0 · BUY YES @ 0.78 Medium Conviction

"Will the Federal Reserve hike rates by 0bps at their July 29, 2026 meeting?" — i.e. the Fed holds. Entry band: 60–90¢ favorite. Cluster: fed-july-rate-path

My probability 82%

→

Market-implied (at 77–78¢): ~77% · My estimate: 82% · Edge ≈ +5¢ / ~6.5% EV (recorded for calibration — did not set the size).

Current YES 77¢ bid / 78¢ ask Limit 78¢ Size 89 contracts Cost $69.42 (6.9% of cap) Max payout $89.00

Mispricing thesis. The July "hold" contract carried ~91¢ through June 16. It then fell to ~76¢ over June 17–19 on two shocks: (1) the hawkish June 17 dot plot under new Chair Warsh, and (2) the mid-June Hormuz/Israel–Iran war that spiked crude and revived July-hike fears. Shock (2) has now fully reversed — crude is back below ~$70 (−40% from peak), Hormuz is reopening, and US–Iran talks are advancing — removing the energy-inflation impulse that justified the July-hike premium. Yet HOLD traded ~150k contracts on Jun 24 and closed unchanged at ~76¢: the price has not absorbed the disinflationary news. With the Fed having just held, a back-to-back hike now requires the data to force it — and the data catalyst just moved the Fed's way.

Evidence (primary sources).

Jun 17 FOMC: rates held at 3.50–3.75%; statement turned hawkish (easing language removed; 9/18 dots show a hike this year; median 3.8% YE). federalreserve.gov 20260617
Jun 24 oil crash: crude below ~$70–72, lowest since late February, −~40% from the wartime peak; Hormuz tanker traffic resuming; a 60-day US waiver on Iranian crude; US–Iran talks advancing. CNBC Jun 24
Price history (this DB): HOLD bid 91.5 (Jun 16) → 86.4 (Jun 17) → 78.5 (Jun 18) → 74.5 (Jun 19) → 75.6 (Jun 23) → 75.6 (Jun 24). The Jun 24 crash did not move it.
Rules confirm exact-hold semantics: "If the Federal Reserve does a Hike of 0bps on July 29, 2026, then the market resolves to Yes." A cut also resolves NO — but a July cut is priced ~2¢ (C25 at 1–2¢), so this is functionally a hold-vs-hike bet.

Cleanest way this loses: the June CPI print (Jul 14) comes in hot on sticky shelter/services or tariff pass-through, and a credibility-seeking Warsh Fed hikes 25bp on July 29 regardless of cheaper oil. That is a real ~18% path — which is exactly why this is sized as MEDIUM, not HIGH. Secondary risk: the Iran ceasefire breaks and crude round-trips before resolution.

Why MEDIUM, not HIGH. Per v3, HIGH requires the outcome to be mechanically locked. This is a 5-week-out forecast with a live binary catalyst (Jul 14 CPI) before it, and the price fell for genuine fundamental reasons, not a pure book error. It is a defensible directional lean with a real way to be wrong — the textbook MEDIUM tier.

Max payout/contract $1.00 EV/contract +5¢ EV% ~6.5% model_prob 82

Liquidity / entry context (live book, Jun 24). YES 77¢ bid / 78¢ ask; resting NO depth of tens of thousands of dollars within 2–3¢ (≥6,000 contracts fillable at ≤78¢); 24h volume ~127k, open interest ~949k. An 89-contract clip fills with no slippage — this is one of the most liquid books on the venue.

Price history. 91¢ plateau (Jun 10–16) → step-down to ~75¢ (Jun 17–19 on the dots + war) → flat 75–78¢ since, including through today's oil crash. No move ≥20¢ in the last 48h on this thesis — the disinflation news is unabsorbed, so the edge is not yet eaten.

4 · Recommended $1,000 portfolio

Pick	Side	Limit	Contracts	Cost	Max payout	EV%	Conv.	Cluster
KXFEDDECISION-26JUL-H0	BUY YES	78¢	89	$69.42	$89.00	6.5%	Med	fed-july-rate-path
Deployed				$69.42	$89.00	6.5%
Cash reserve				$930.58	—	—	93.1% of capital

EV% is shown for the record only. Under v3 it did not drive the contract count — size came from the MEDIUM tier cap ($70) at the 78¢ entry.

Cash is the position. 93% of capital is held as cash by design. Economics is efficient this week and only one contract cleared every screen. The reserve is dry powder for opportunistic adds — chiefly if HOLD still hasn't repriced into the Jul 14 CPI, or if a clean disinflation mispricing opens in the Treasury/gas ladders.

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

Cluster	Cost	% of capital	Cap	Status
fed-july-rate-path	$69.42	6.9%	15%	OK

Only one underlying thesis carries risk; no single event can flip the run's sign. The inverse hike leg (H25) and any rate-sensitive Treasury bet were deliberately excluded so this cluster stays a single, capped position.

Conviction exposure (caps: HIGH 15% · MEDIUM 7% · LOW 3% per pick)

Tier	Deployed	% of capital	Notes
High	$0.00	0%	No mechanically-locked edge cleared this week
Medium	$69.42	6.9%	1 pick at the MEDIUM cap
Low	$0.00	0%	—

Risk profile

Worst case: the Fed hikes (or, marginally, cuts) on Jul 29 → lose the full $69.42 (6.9% of capital). No correlated second leg amplifies it.
Best case: Fed holds → collect $89.00, a +$19.58 profit (+28% on cost).
Most likely (my 82%): Fed holds; expected P&L ≈ +$3.56 on $69 at risk. Small in dollars by design — the discipline is the point.
Concentration: one thesis, 6.9% of capital, 93% cash. This book cannot blow up on a single correlated cluster — the specific failure mode the v3 caps target.

Execution notes

Limit, don't chase. Post YES at 77–78¢. The book is deep enough that 89 contracts fill at the touch; do not pay above 79¢ (it erodes the ~5¢ edge).
Watchlist add: if HOLD is still ≤78¢ the morning after the Jul 14 CPI and CPI is in-line/soft, add up to the MEDIUM cap — the edge will have widened.
Invalidation triggers: (a) HOLD reprices to ≥84¢ — edge gone, stand down; (b) a hot Jul 14 CPI (headline ≥ +0.5% m/m surprise) — the hike path reopens, exit; (c) Iran ceasefire collapses and crude round-trips above ~$85 — thesis broken.
Hedge: none needed at this size. The natural hedge (selling the H25 hike leg) is the same cluster and is intentionally left on the table.

5 · What I rejected and why

Every Stage-2 survivor below is machine-logged to picks.json and shadow-tracked to settlement. If the rejects resolve as well as the pick, the selection screen added nothing — and that is a finding worth proving either way. Entry-band cuts (coin-flips and tails) are called out so the discipline is visible.

Ticker	Would-be side	Price	My prob	Band / tier	Why rejected
KXU3-26JUN-T4.1	YES	92¢	96%	>90¢ deep fav	~4¢ edge, but thin YES book → fills pay ≥3¢ slippage up to 95¢. Edge eaten by execution.
KXPAYROLLS-26JUN-T70000	YES	67¢	70%	60–90¢ / not HIGH	Data-release coin-flip with no informational edge vs the ~100k consensus — the exact medium-conviction bucket that bled money in v1/v2.
KXCPIYOY-26JUN-T3.6	YES	92¢	92%	>90¢ deep fav	Efficient. The YoY base-effect step-down (May 4.2% → mode ~3.7–3.8%) is fully modeled; zero edge.
KXCPICOREYOY-26JUN-T3.1	YES	6¢	~30%	<35¢ tail	Near-zero bid with a ~20¢ spread (5/24). Untradeable; fails liquidity + spread screens. No lottery slot used.
KXAAAGASM-26JUN30-3.85	NO	58¢*	58%	35–60¢ coin-flip	NO effective entry ~58¢ and not HIGH conviction. The oil-crash gas decline is already priced; also an existing-pick cluster.
KXUST10M-26JUN30-T4.45	YES	70¢	68%	60–90¢ / not HIGH	6-day yield random walk (10Y ~4.51%, 6bp cushion). Not mechanically locked; no defensible directional edge.
KXGDP-26JUL30-T0.5	YES	94¢	94%	>90¢ deep fav	5+ weeks out, no recent volume, no edge vs GDPNow-anchored consensus. Thin liquidity.
KXFEDDECISION-26JUL-H25	NO	21¢	84%	same cluster	Inverse of the HOLD pick. Logged for transparency, not deployed — it would double the fed-july-rate-path cluster.

*NO effective entry = 100 − YES bid. The gas YES leg trades 42/47.

6 · Sources

Federal Reserve — June 17, 2026 FOMC statement & June projections (dot plot)
Federal Reserve — FOMC meeting calendar (July 28–29, 2026)
CNBC — Fed holds rates June 2026 (Warsh's first meeting; hawkish turn)
CNBC — Oil & Hormuz coverage (June 2026); crude back below ~$70 / Hormuz reopening / US–Iran talks
BLS — Employment Situation, May 2026 (U-3 4.3%, +172k payrolls)
BLS — Consumer Price Index, May 2026 (+4.2% headline YoY, +2.9% core, energy +23.5%)
AAA — National fuel price average (~$3.93/gal, Jun 24)
U.S. Treasury — Daily par yield curve (10Y 4.51%, 2Y 4.24%, 30Y 4.95%, Jun 22)
Live market data: Kalshi public API (orderbooks) & the read-only Kalshi mirror (market_snapshots price/volume history).

Data sources: read-only Kalshi DB mirror (trading_markets, trading_events, market_snapshots); Kalshi public trade API for live orderbooks; web search/fetch for primary-source grounding.

Disclaimer: All probabilities are subjective estimates, not guarantees. Prediction-market contracts can resolve to zero and you can lose 100% of capital at risk. This is research/education, not financial advice. Figures captured around 2026-06-24 16:00 UTC; live prices move. v3 picks count toward the public performance record and are cleared for auto-copy — size accordingly.