Kalshi Economics — Mispricing Audit

Category: Economics · Horizon: 45 days (markets closing by 18 Jul 2026) · Capital: $1,000 · Report date: 3 Jun 2026

Prices as of the 3 Jun 2026 ~20:46 UTC mirror snapshot + live Kalshi orderbook pulls at write time.

Bottom line. The Economics shelf this window is dominated by deep, efficiently-priced US macro-data ladders (May jobs 5 Jun, May CPI 10 Jun, FOMC 17 Jun) where I hold no informational edge — and where the few discrete rate-decision markets (Brazil Copom, Banxico, FOMC dissent count) screen as fairly priced or already repriced on public news. After a five-candidate diligence sweep with an adversarial second pass, exactly one thesis survived: a low conviction bet that the multi-week AAA gasoline decline does not fully stall by 8 Jun. I deploy $186 (19%) across two legs of that single thesis and hold $814 (81%) in cash. This is a thin-opportunity week and I am not going to pad it.

1 · How this was researched

  1. Universe. Queried the live Kalshi mirror for every active market whose event category = 'Economics' closing within 45 days: ~400 markets across 131 series.
  2. Liquidity screen. Joined the latest market_snapshots row per market and cut everything with volume24h < 1500, bid-ask spread > 5¢, or already pinned (YES bid ≥ 95 / ask ≤ 5). That left 39 liquid candidates, clustered in a handful of series.
  3. Structure check. Reconstructed the full price ladders for the mutually-exclusive / monotone sets (PAYROLLS, CPI, CPI-YoY, U-3, the central-bank decision trees, the gas ladders) and checked for monotonicity violations and leg-sum dislocations. None of the liquid legs were internally arbitrageable — these books are coherent.
  4. Diligence. For the researchable discrete events I pulled the resolution rules, a multi-day intraday price history, the live depth-of-book, and primary-source news (central-bank statements, Focus/economist surveys, AAA/EIA data, Reuters/Bloomberg). Five candidate theses ran in parallel.
  5. Adversarial pass. Every thesis that came back a BUY was handed to a second analyst whose only job was to refute it — check whether the cited consensus was current, whether the Kalshi price had already moved to fair value, and whether a definitional trap could flip resolution. One BUY (FOMC dissent) died on that pass; one (gas) survived with conviction cut.

2 · Markets reviewed

Representative cut of the 39-candidate liquid screen and the verdict each reached. Prices are YES unless noted.

Series / marketWhat it asksClosePxVol24hVerdict
KXAAAGASW-26JUN08AAA US avg gas vs ladder, value on 8 Jun8 Junladder~16k topBUY NO ×2 legs
KXCBDECISIONBRAZIL-26JUN17Copom: Cut-25 vs Hold17 JunC25 57 / H 413.6kPASS edge eaten
KXFOMCDISSENTCOUNT-26JUN-0Exactly 0 dissents at June FOMC17 Jun775.3kPASS refuted
KXCBDECISIONMEXICO-26JUN25-HOLDBanxico holds 25 Jun25 Jun93/970.8kPASS efficient
KXAAAGASD-26JUN04AAA US avg gas, value on 4 Jun4 Junladder~13k topPASS well-priced
KXPAYROLLS-26MAYMay nonfarm payrolls ladder5 Jun>100k 469.7k topPASS = consensus
KXU3-26MAYMay U-3 unemployment ladder5 Jun>4.3% 352.2k topPASS = consensus
KXCPI-26MAY / KXCPIYOY-26MAYMay CPI MoM & YoY ladders10 Jun>0.4% 739.4k topPASS = consensus
KXFEDDECISION-26JUN-H0Fed holds at June meeting17 Jun9691kPASS pinned
KXCHAICUTS-26JUN04-T1AI = #1 cited reason in Challenger May report4 Jun92/944.7kPASS ~priced + noisy

3 · Picks — detail and thesis

Ordered highest-conviction first. Both deployed picks are the same underlying thesis (US gasoline keeps declining), expressed at two thresholds with different risk/reward. The concentration is intentional and flagged in the portfolio section — it is the only edge that cleared the screen.

The thesis, in one line. The AAA US national average for regular gas was $4.261 on 3 Jun, down from $4.290 the day before and ~$4.459 a week earlier — a steady ~2.8¢/day slide for three straight weeks as the May crude slump works through the ~1–2 week pump lag. The 8-Jun ladder prices a near-stall (implied ~$4.205, ≈1.1¢/day). If the decline merely continues at a fraction of its recent pace, the higher gas thresholds resolve NO.

Pick 1 — KXAAAGASW-26JUN08-4.220 · BUY NO @ 0.61 · Low conviction

Market: “AAA US avg regular gas > $4.220 on 8 Jun?”Current NO ask: 61¢ (YES 39)Resolves 8 Jun 2026

Mispricing. NO wins if gas settles ≤ $4.22 on 8 Jun — i.e. it needs to fall just ~4¢ over five days versus an observed ~2.8¢/day pace. The market still assigns ~42% to gas being above $4.22 (a near-total stall). My forecast distribution (μ ≈ $4.19, σ ≈ $0.04) puts true NO ≈ 73%. The only path that loses is a fast crude-to-pump reversal — real, but less than 42% likely. This is the most robust expression: it pays unless the decline fully stalls.

Evidence (all public):

Tail risk (the cleanest way this loses): crude is rebounding (Brent ~$97–101, WTI ~$95–96; Iran/Hormuz strikes, 6th weekly inventory draw) and GasBuddy's De Haan warns relief “may be short-lived” with price-cycling states “running out of room” — a near-term stall/reversal that keeps gas above $4.22. Fortune, 3 Jun.

True prob (NO)
~73% · entry 61¢ · edge ≈ +12¢
Sizing
200 contracts @ limit 0.61 → cost $122, max payout $200, EV ≈ +$24 (≈ +20%)
Liquidity
NO fillable 61–62¢ for ~265 contracts before walking; spread 6¢ on the leg but depth adequate for this size.
Price history
YES leg drifted down 73 → 47 (1→3 Jun) as gas fell; no single ≥20¢ jolt — an orderly re-rate, not a spike.

Pick 2 — KXAAAGASW-26JUN08-4.200 · BUY NO @ 0.43 · Low conviction

Market: “AAA US avg regular gas > $4.200 on 8 Jun?”Current NO ask: 42¢ (YES 58)Resolves 8 Jun 2026

Mispricing. The higher-payout expression of the same view. NO wins if gas settles ≤ $4.20 (a ~6¢ fall, ≈ the market's own implied median). At μ ≈ $4.19 this is roughly a 58% NO proposition bought at 42¢ — a fatter edge but closer to a coin flip, because the strike sits right at the market's median. Included as a smaller, higher-beta sleeve of the gas thesis, not a second idea.

Evidence: identical to Pick 1 (same AAA trajectory, same crude lag). The distinction is purely where the strike sits relative to the forecast: $4.20 is the market's median, $4.22 is comfortably inside it.

Tail risk: same crude-rebound stall — and here it bites harder, since a settle of $4.20–$4.21 (a plausible deceleration outcome) resolves YES and this loses while Pick 1 still wins.

True prob (NO)
~58% · entry 42¢ · edge ≈ +15¢
Sizing
150 contracts @ limit 0.43 → cost $64.50, max payout $150, EV ≈ +$22.5 (≈ +35%)
Liquidity
NO fillable 42–45¢; ~140 contracts at 42, ~400 more at 45. Rest at 0.43 and accept partials to hold slippage < 3¢.
Price history
YES leg 76 → 68 → 56 (1→3 Jun); same orderly decline.

No third pick. The remaining four candidates were rejected — see §5. Forcing a Brazil, Banxico, Fed or US-data trade here would mean paying for prices that already reflect the public information.

4 · Recommended $1,000 portfolio

PickSideLimitQtyCostMax payoutTrue pEV $EV %Conv.
KXAAAGASW-26JUN08-4.220NO0.61200$122.00$20073%+$24.0+20%Low
KXAAAGASW-26JUN08-4.200NO0.43150$64.50$15058%+$22.5+35%Low
Deployed350$186.50$350+$46.5+25%
Cash reserve$813.5081%

Blended EV ≈ +25% on deployed capital; dollar edge ≈ +$46.50. Deployment is held to 19% by design.

Risk profile

Execution notes

5 · What I rejected, and why

The screen is only as good as what it throws away. Four researched candidates and several whole series were cut.

PASS Brazil Copom — Cut-25 KXCBDECISIONBRAZIL-26JUN17-C25 @ 57¢

The cleanest discrete edge on paper — and it already moved. Cut-25 repriced from ~31¢ (2 Jun) to a 79¢ intraday spike, settling ~57¢ on 3 Jun, when Brent fell ~20% off its 2026 peak (US–Iran ceasefire optimism), clearing the ~$114 “pause” trigger that the May Copom minutes had flagged. My calibrated cut probability is ~63%; Polymarket independently sits ~65%. At 57¢ that is a ~6¢ edge — below the 7¢ bar — and the HOLD tail is genuinely live: 2026 IPCA expectations have climbed 11 straight weeks to 5.04%, above the 4.5% ceiling, and the May IPCA print lands 12 Jun, pre-meeting. Edge eaten, two-sided risk. MercoPress 7 May · Boletim Focus via Rio Times · Polymarket

REFUTED FOMC dissent count = 0 KXFOMCDISSENTCOUNT-26JUN-0 @ 77¢

This one came back a confident BUY NO (“the most divided Fed since 1992 won't vote unanimously”) — and the adversarial pass killed it. The load-bearing claim was a cross-check that “Polymarket prices 0-dissents at only ~46%, so Kalshi's 77¢ is an overshoot.” On inspection that 46% figure was a stale page dated 21 May; the live Polymarket reads ~70% and has converged with Kalshi — no outlier. Worse, the April hawk dissents (Hammack, Kashkari, Logan) were over easing-bias language, not the rate vote, and that bias is likely dropped in June given the oil shock — removing their reason to dissent; and serial dissenter Miran resigned in May. Calibrated P(0 dissents) is ~60–65%, so NO at 24¢ carries only a fragile, model-dependent edge. Textbook example of a thesis built on an out-of-date receipt. Rejected. Fed statement 29 Apr · the stale 21-May page

PASS Banxico holds KXCBDECISIONMEXICO-26JUN25-HOLD @ 93/97

Tempting to fade a 96% favorite — but the primary sources back it. On 7 May Banxico cut to 6.50% and issued its strongest hold guidance of the cycle (“the Board considers it appropriate to maintain the reference rate at its current level”), with BBVA and BofA framing it as the final cut and inflation still sticky (Apr headline 4.45%). Fair hold ≈ 93%; buying NO-on-HOLD at ~8¢ implies ~92% — about 1¢ of edge the wrong way. No trade. BBVA Research 8 May

PASS The big US macro ladders — payrolls / U-3 / CPI

The deepest, most-traded Economics markets, and they sit on top of published consensus: May NFP implied median ≈ +85k vs the Reuters survey +85k; U-3 modal 4.3% vs consensus 4.3%; CPI MoM ~0.45% and YoY ~4.2% both consistent with the energy-driven step-up. The one apparent gap — a 0.27% Cleveland Fed CPI nowcast — is a stale (5–7 May) artifact that predates the May gasoline surge; the gasoline math (~+0.3pp to headline) actually validates the market's 0.45%. These books digest the 8:30 ET prints within minutes; any genuine edge is captured at release, not held by me three days out with a January knowledge cutoff. PASS on all four.

PASS 4-Jun daily gas, and the noise markets

The KXAAAGASD-26JUN04 daily ladder is well-calibrated — its implied median (~$4.236) sits exactly one day's decline below today, so no defensible edge. Cut without deep diligence: announcer/“mention”-style and idiosyncratic single-name markets (Musk net worth, Fed tweets, Stripe/OpenAI IPO-by-date, tech-layoff counts) and the Challenger “AI = #1 reason” market (already ~93¢ and definition-sensitive) — these fail the “skip mention/noise” rule or offer no quantifiable edge.

6 · Sources

Data sources: live Kalshi mirror (trading_markets, trading_events, market_snapshots) for the universe, ladders and history; the Kalshi public API for depth-of-book at write time; web search/fetch for primary-source news. Prices captured 3 Jun 2026.

Disclaimer. Not investment advice. Every probability here is subjective — my own estimate, not a market truth — and 5-day gasoline forecasting in particular has real variance. Prediction-market contracts resolve to zero when wrong; you can lose the entire stake on any position. Do your own diligence and size to what you can afford to lose.