Kalshi Economics — Mispricing Audit
1. How this was researched
Mode: category-match. $CATEGORY = Economics matched 1,148 active markets (114 series) in the live Kalshi mirror closing within 45 days, so no theme-keyword fallback was needed. The mirror schema exposes trading_events / trading_markets only — there is no market_snapshots table in the current schema — so all prices, volume, open interest, 14-day history (candlesticks endpoint) and order books came from the Kalshi public API at pick time.
Stage 1 mechanical screen (volume24h ≥ 1,500 · spread ≤ 5¢ · not fully priced · no announcer-noise markets · nothing already in $EXISTING_PICKS) reduced 1,408 fetched open markets to 16 candidates across six thesis groups: the July FOMC complex, the AAA monthly gas-price ladder, July CPI YoY, the Bank of Russia July 24 decision, weekly jobless claims, and two fast-food price trackers. Each survivor got rules-text review, resolver profiling (Stage 2.5), 14-day price history, live order-book depth, and primary-source news diligence. Note: many July Economics staples (CPI MoM, payrolls, PCE, the Fed hold leg) are already live picks from prior runs and were excluded up front.
2. Markets reviewed
| Market | Question | YES bid/ask (¢) | Vol 24h | Disposition |
|---|---|---|---|---|
| KXAAAGASM-26JUL31-4.16 | AAA avg gas > $4.16 on Jul 31 | 82 / 83 | 2,213 | PICK — BUY YES (HIGH) |
| KXAAAGASM-26JUL31-4.18 | AAA avg gas > $4.18 on Jul 31 | 77 / 78 | 2,448 | PICK — BUY YES (MEDIUM) |
| KXFEDDECISION-26JUL-H25 | Fed hikes 25bp on Jul 29 | 23 / 24 | 658,558 | PICK — BUY NO (MEDIUM) |
| KXFED-26JUL-T3.75 | Fed funds upper bound > 3.75% after Jul 29 | 22 / 23 | 57,468 | REJECT — dominated by H25 NO |
| KXCPIYOY-26JUL-T3.3 | July CPI YoY > 3.3% | 69 / 72 | 3,196 | REJECT — fade is a banned tail |
| KXCPIYOY-26JUL-T3.4 | July CPI YoY > 3.4% | 27 / 28 | 12,188 | REJECT — fairly priced vs nowcast |
| KXAAAGASM-26JUL31-4.20 | AAA avg gas > $4.20 on Jul 31 | 62 / 65 | 5,207 | REJECT — thin edge, cluster cap |
| KXAAAGASM-26JUL31-4.22 | AAA avg gas > $4.22 on Jul 31 | 43 / 44 | 2,487 | REJECT — coin-flip band |
| KXAAAGASM-26JUL31-4.30 | AAA avg gas > $4.30 on Jul 31 | 5 / 6 | 3,158 | REJECT — tail band |
| KXCBDECISIONRUSSIA-26JUL24-C25 | Bank of Russia cuts 25bp Jul 24 | 58 / 59 | 1,587 | REJECT — edge already eaten |
| KXCBDECISIONRUSSIA-26JUL24-HOLD | Bank of Russia holds Jul 24 | 34 / 35 | 3,288 | REJECT — no edge either side |
| KXFEDCOMBO-26JUL-0-T0 | Hold AND ≥1 dissent | 59 / 60 | 2,524 | REJECT — fairly priced |
| KXFEDCOMBO-26JUL-0-0 | Hold AND 0 dissents | 17 / 18 | 2,002 | REJECT — tail band |
| KXJOBLESSCLAIMS-26JUL23-200000 | Initial claims ≥ 200k (w/e Jul 18) | 73 / 77 | 1,887 | REJECT — no edge, closes in hours |
| KXPOPCHICKSAND-26AUG02-T5.67 | Popeyes sandwich avg ≥ $5.67 (July) | 43 / 44 | 2,148 | REJECT — resolver risk (Spice Data) |
| KXCHIPBURRITO-26AUG02-T9.80 | Chipotle burrito avg ≥ $9.80 (July) | 5 / 6 | 1,672 | REJECT — tail + resolver risk |
3. Picks — detail and thesis
Macro backdrop (drives everything below): The U.S.–Iran conflict has largely blocked the Strait of Hormuz (~20% of global energy supply pre-war). Brent recovered to ≈$89 on Jul 21 amid a 10th consecutive day of U.S. strikes and Iranian missile attacks on Kuwait, with a mediated "10-day ceasefire" proposal floating. AAA national average gasoline: $3.83 (Jul 2) → $3.84 (Jul 9) → $3.94 (Jul 16) → $4.06 (Jul 22). June CPI printed soft (−0.4% MoM; YoY 3.5%), collapsing July Fed-hike odds from ~46% to the teens; new Chair Kevin Warsh holds the funds rate at 3.50–3.75% and has replaced forward guidance with data-dependence.
Pick 1 — KXAAAGASM-26JUL31-4.16 · BUY YES @ 84¢ limit · HIGH CONVICTION
Current price: 82¢ bid / 83¢ ask (last 81¢)
My probability: 90% vs market implied: ~83% → +6¢ edge at the 84¢ limit
Entry band: favorites (60–90¢) · Cluster: hormuz-gas-jul31
Size: 150 contracts ≈ $126 cost (5.0% of capital; HIGH cap is 15%, deliberately undersized to book depth)
Thesis — the pass-through pipeline is mechanically loaded. AAA sits at $4.06 today with 8–9 days of drift left. The rung needs just +10¢ (~1.2¢/day) when the actual pace over the past six days was +2¢/day ($3.94 → $4.06). More important than momentum: crude has risen ≈$17/bbl since early July (≈+40¢/gal wholesale-equivalent) while retail is up only +23¢ — the classic rockets-and-feathers lag means stations keep raising prices for 1–2 weeks even if crude freezes today. For this rung to fail, crude must crash within days and retail must flatten almost immediately — retail gas essentially never falls quickly, even after ceasefires.
Evidence:
- AAA national average $4.06 on 7/22, up from $3.94 on 7/16 and $3.84 on 7/9 (AAA Fuel Prices newsroom, fetched today) — the resolver's own feed.
- Brent ≈$89 on Jul 21; Hormuz largely blocked; 10th consecutive day of U.S. strikes; Houthis threatening Red Sea routes (CNBC, Barchart, TradingEconomics).
- AAA's own 7/16 commentary attributes the 10¢/week climb to "instability along the Strait of Hormuz."
Resolver (Stage 2.5): AAA daily national average for regular, as published for Jul 31 — a snapshot of a slow-moving, thrice-decimal index the resolver itself publishes daily; "strictly greater than $4.16" is unambiguous at 3 decimals. Priced risks: −1 pt (AAA publication/timing quirk). Evidence-based 91 − 1 = 90.
Tail risk (cleanest way to lose): a credible ceasefire lands this week, Brent gives back $10–15 within days, and retail stalls under $4.16. Even then the pipeline math makes a sub-$4.16 close a stretch — hence 10 points, not 25.
Liquidity / entry context: book is thin on the offer — ~20 contracts at 83¢, ~28 within the 84¢ limit; 24h volume 2,213; OI 2,285. This is a resting-limit accumulation over 2–3 sessions, not a sweep; sized down to 150 contracts accordingly (see execution notes). Unfilled = cash, not loss.
Price history: the 4.20 sister rung swung 9¢→50¢→33¢ over two weeks as war headlines whipsawed; the ladder reprices daily off the AAA print, so the morning AAA number is the fill catalyst.
Pick 2 — KXFEDDECISION-26JUL-H25 · BUY NO @ 77¢ limit · MEDIUM CONVICTION
Current price: YES 23¢ bid / 24¢ ask → NO fills at 77¢
My probability (NO side): 90% vs market implied: 77% → +13¢ edge
Entry band: favorites (60–90¢ effective) · Cluster: fed-july-hold
Size: 227 contracts ≈ $175 cost (7.0% of capital = MEDIUM cap)
Thesis — retail flow is paying panic prices for a hike the committee can't pass. The hike leg collapsed to 4–7¢ after June CPI printed −0.4% MoM (core flat) and PPI −0.3%, then tripled today (7¢ → 24¢ on 1.46M contracts) as Brent hit $89 and Iran struck Kuwait. But CME fed-funds futures — the professional market — priced only 16.6% hike as of Jul 21, and the FOMC entered blackout Jul 18: no official can telegraph a hike this week, and the modern Fed does not deliver un-telegraphed hikes. The vote math is the anchor: a hike needs a Board majority, and the Board is stacked with holdovers and Trump-appointed doves (Waller and Bowman dissented for cuts in 2025; Cook, Jefferson, Barr are holds). The hawks priced >50% to dissent — Logan (~57¢) and Hammack (~51¢) — are regional presidents who can dissent but cannot carry a majority. Warsh himself answered "Not yet" on hikes after the June data. Fading the oil-headline spike at 77¢ NO is the trade.
Evidence:
- CME FedWatch: 83.4% hold as of Jul 21 (GrowBeanSprout FedWatch tracker); hike odds collapsed from ~46% to the teens after June CPI/PPI (TechTimes 7/15, Axios 7/15).
- June data: CPI −0.4% MoM, core flat, PPI −0.3% MoM (TechTimes 7/15). Warsh on hikes: "Not yet."
- Today's Kalshi spike is attributed by the financial press to oil-escalation flow, not any policy leak — no sourced report of hike deliberations exists (searches across Bloomberg/WSJ/CNBC coverage through today).
- Structure: NO on the 25bp-hike leg also wins on a cut or a >25bp hike (both priced ≤1¢) — it is strictly safer than the YES-on-hold leg already live in the feed.
Resolver (Stage 2.5): the FOMC statement / implementation note target range published Jul 29 — fully mechanical. Emergency intermeeting moves do not count (series rules); a cancelled/delayed meeting resolves as no-change, which favors NO. Priced risks: −1 pt procedural. Evidence-based 91 − 1 = 90.
Tail risk (cleanest way to lose): Warsh engineers a genuine surprise — a "credibility hike" into an oil shock, carried by a Board that publicly leaned dovish three weeks ago, announced with zero telegraphing. That is the 10%. Secondary flag, stated honestly: this pick fades a same-day 3× move on heavy volume, which v3 normally treats as "edge already eaten" — it deploys only because the catalyst is identifiable (oil headlines), the professional market disagrees with the panic, and blackout structurally blocks the confirmation path. That flag is why this is MEDIUM, not HIGH, despite the 13¢ modeled edge.
Feed-correlation note: KXFEDDECISION-26JUL-H0 YES (hold) is already a live pick from a prior run. This is a different contract with a strictly wider win-set, entered after today's dislocation, but subscribers holding both should recognize they share the no-hike thesis.
Liquidity / entry context: exceptional — 20,714 contracts fillable at 77¢ NO (resting YES bids at 23¢), 24h volume 658,558, OI 6.9M. Zero slippage at recommended size.
Price history: hike leg 17–21¢ (Jul 8–13) → 32¢ (Jul 14 peak, pre-CPI) → 4–7¢ (Jul 15–21, post-CPI collapse) → 13–24¢ today on 1.46M contracts.
Pick 3 — KXAAAGASM-26JUL31-4.18 · BUY YES @ 80¢ limit · MEDIUM CONVICTION
Current price: 77¢ bid / 78¢ ask (order book; screen quote 75/80)
My probability: 84% vs market implied: ~80% at the limit → +4¢ edge
Entry band: favorites (60–90¢) · Cluster: hormuz-gas-jul31
Size: 200 contracts ≈ $160 cost (6.4% of capital; MEDIUM cap 7%)
Thesis: same pass-through pipeline as Pick 1, one rung up: needs +12¢ in 8–9 days (~1.4¢/day) versus the running +2¢/day pace and the ~15–20¢ of un-passed wholesale increase. This rung is where the thesis is actually fillable at size — 135 contracts offered at 78¢ and 200 more at 80¢ right now — so it carries the deployable weight of the gas cluster while Pick 1 carries the best price. The extra 2¢ of strike raises the ceasefire-scenario failure odds, hence MEDIUM and a lower probability (84 vs 90).
Evidence: identical to Pick 1 (AAA $4.06 today, +12¢/6 days; Brent ≈$89; Hormuz blocked).
Resolver (Stage 2.5): same AAA snapshot mechanism; priced risks −1 pt → 84.
Tail risk: same ceasefire/crude-crash path as Pick 1, with ~2¢ less cushion — a deceleration to ~1.3¢/day average is enough to lose this rung while Pick 1 still wins.
Liquidity / entry context: 335 contracts fillable within the 80¢ limit; 24h volume 2,448; OI 2,247. Fillable today without slippage beyond the limit.
4. Recommended $2,500 portfolio
Sizing is set by conviction tier + entry band only (v3 rule). EV columns are recorded for calibration and did not drive contract counts.
| # | Market | Action | Limit | Conviction / band | Contracts | Cost | Max payout | Edge ¢ | EV % |
|---|---|---|---|---|---|---|---|---|---|
| 1 | KXAAAGASM-26JUL31-4.16 | BUY YES | 84¢ | HIGH / favorites | 150 | $126.00 | $150.00 | +6 | 7.1% |
| 2 | KXFEDDECISION-26JUL-H25 | BUY NO | 77¢ | MEDIUM / favorites | 227 | $174.79 | $227.00 | +13 | 16.9% |
| 3 | KXAAAGASM-26JUL31-4.18 | BUY YES | 80¢ | MEDIUM / favorites | 200 | $160.00 | $200.00 | +4 | 5.0% |
| Totals | 577 | $460.79 | $577.00 | 10.1% blended | |||||
Deployed: $460.79 (18.4%) · Cash held: $2,039.21 (81.6%) · Dollar EV ≈ +$46.50 · Profit if all win ≈ +$116.21. The heavy cash weight is deliberate: the Economics slate this week was largely efficient (CPI, claims, dissent-count and Russia legs all priced within a few cents of fair), and v3 treats cash as a position.
Cluster exposure (cap: 15% = $375 per cluster)
| Cluster | Picks | Cost | % of capital | Cap |
|---|---|---|---|---|
| hormuz-gas-jul31 | #1, #3 | $286.00 | 11.4% | 15% ✓ |
| fed-july-hold | #2 | $174.79 | 7.0% | 15% ✓ |
Conviction exposure
| Tier | Cost | % of capital | Per-pick cap |
|---|---|---|---|
| HIGH | $126.00 | 5.0% | 15% ($375) |
| MEDIUM | $334.79 | 13.4% | 7% ($175) |
| LOW | $0 | 0% | never deploys (v3.3) |
Risk profile
- Natural hedge: the two clusters are negatively correlated through the oil channel. A ceasefire/crude crash kills the gas legs (−$286) but makes a hike absurd (+$52 on Pick 2 → net ≈ −$234). An inflation-panic hike (−$175) almost certainly comes with surging gas (+$64 → net ≈ −$111). The all-lose scenario (crude crash and a hike into disinflation) is internally contradictory; the mechanical worst case −$460.79 (18.4%) is far tail.
- Best case: war grinds on, no hike — all three win, +$116.21 (+25% on deployed cost).
- Most likely: same as best case; the modal path has AAA in the $4.20s on Jul 31 and a hold (possibly with hawkish dissents, which don't affect Pick 2).
- Concentration: largest single thesis 11.4% of capital; no cluster can flip the run's sign.
Execution notes
- Pick 1 (4.16): book shows only ~28 contracts within 84¢ — rest a limit at 83–84¢ and accumulate over 2–3 sessions as the ladder reprices off each morning's AAA print. Do not chase above 84¢. Cancel resting orders on a signed ceasefire + Brent < $80.
- Pick 3 (4.18): 335 contracts fillable now within 80¢ — take the offer up to the limit. Same invalidation trigger as Pick 1.
- Pick 2 (H25 NO): 20k+ contracts at 77¢ — fill immediately. Invalidation: any sourced report (WSJ/Bloomberg) of the committee deliberating a hike, or CME FedWatch hike odds > 40% on data (not oil headlines) before Jul 29 — exit if either appears.
- Watchlist: KXCPIYOY-26JUL-T3.3 NO becomes actionable only if the effective NO entry rises out of the tail band (≥ 36¢, i.e., YES ≤ 64¢) while the Cleveland nowcast holds ≈3.37; the gas weekly series (KXAAAGASW) offers a faster-resolving echo of the gas thesis if depth improves.
5. What I rejected and why
All 13 rejects below are machine-logged in picks.json and shadow-tracked to settlement — if they outperform the picks, the screen is adding nothing and we want to know. Entry-band cuts are marked [band].
- KXFED-26JUL-T3.75 — NO @ 78¢, prob 90. Same no-hike thesis as Pick 2 but strictly worse (loses on a >25bp hike where H25 NO wins) and would breach the fed-cluster cap. Dominated.
- KXCPIYOY-26JUL-T3.3 — NO @ 31¢, prob 43. The genuinely interesting find of the CPI complex: the Cleveland Fed nowcast (updated today) has July YoY at 3.37%, only 2bp above the 3.35 rounding boundary for a printed 3.4 — nowcast-fair is ~55–60¢ YES, vs market 70¢. The fade has ~+12¢ of modeled edge — [band] and it is a sub-35¢ tail, hard-rejected with no exceptions (v3.2). Logged so the shadow track can argue.
- KXCPIYOY-26JUL-T3.4 — NO @ 73¢, prob 75. Fairly priced against the same nowcast (+2¢). No edge.
- KXAAAGASM-26JUL31-4.20 — YES @ 66¢, prob 69. Passes every screen but the marginal edge (+3¢) is inside model noise and the gas cluster budget is better spent on the 4.16/4.18 rungs. Cluster-cap allocation cut.
- KXAAAGASM-26JUL31-4.22 — YES @ 44¢, prob 48. [band] Coin-flip band without HIGH conviction — this rung is a pure war-continuation bet.
- KXAAAGASM-26JUL31-4.30 — YES @ 6¢, prob 8. [band] Sub-35¢ tail.
- KXCBDECISIONRUSSIA-26JUL24-C25 — YES @ 59¢, prob 62. The cut leg repriced 29¢ → 60¢ in the last 48h (Kalshi now ahead of Polymarket's 42% cut); whatever signal drove it is already in the price, and a CBR board decision two days out is not a HIGH-conviction object. [band]-adjacent coin-flip; edge eaten.
- KXCBDECISIONRUSSIA-26JUL24-HOLD — NO @ 66¢, prob 65. Complement of the cut leg; no edge after the same repricing.
- KXFEDCOMBO-26JUL-0-T0 — YES @ 60¢, prob 58. Market-implied P(dissent | hold) = 77% matches what the individual dissent legs imply (Logan ~57¢, Hammack ~51¢, correlated). Fairly priced, and it would stack the fed cluster.
- KXFEDCOMBO-26JUL-0-0 — YES @ 18¢, prob 17. [band] Tail, and no edge anyway.
- KXJOBLESSCLAIMS-26JUL23-200000 — YES @ 77¢, prob 76. Claims ran 216k → 208k the last two weeks; the ladder implies a ~204k mean. Fair, 4¢ spread, closes in under 24 hours. No edge, no time.
- KXPOPCHICKSAND-26AUG02-T5.67 — YES @ 44¢, no probability formed. Stage 2.5 resolver reject: resolves off "Spice Data," an opaque scraper whose July running average is ~70% determined but unobservable to us — exactly the resolver-lag profile that produced the program's worst tail loss (Hormuz transit counts). Unknowable resolver ⇒ no forecast ⇒ no trade.
- KXCHIPBURRITO-26AUG02-T9.80 — YES @ 6¢. [band] Tail + same opaque resolver family.
6. Sources
- AAA Fuel Prices — newsroom — national average $4.06 (7/22), $3.94 (7/16), $3.84 (7/9), $3.83 (7/2); Hormuz commentary. (Resolver source for Picks 1 & 3.)
- Cleveland Fed — Inflation Nowcasting — July 2026 CPI YoY 3.37%, MoM 0.04% (updated 7/22).
- CME FedWatch via GrowBeanSprout — 83.4% hold probability for Jul 29 as of Jul 21.
- TechTimes (7/15) — June CPI −0.4% MoM, core flat, PPI −0.3%; hike odds 46% → teens; CPI YoY 3.5%, PPI YoY 5.5%.
- Axios (7/15) — Warsh comments + June data make a July hike unlikely.
- Chase (July 2026 Fed preview) — funds rate 3.50–3.75% under Chair Warsh; "prices are too high."
- CNBC (7/13) and CNBC (7/9) — pre-CPI hike-odds runup; split committee context.
- Bloomberg (7/14) — hike bets before the June CPI print.
- CNBC oil coverage (7/8–7/15) and Barchart — Hormuz blockade, Brent ≈$89 (7/21), Kuwait strikes, ceasefire mediation.
- Bank of Russia press release and Moscow Times (6/19) — key rate cut 25bp to 14.25% on Jun 19.
- Polymarket — Bank of Russia July — 42% cut / 36.5% hold cross-reference.
- IndexBox and VerifiedInvesting — initial claims 208k (w/e 7/11), 215–216k (w/e 7/4).
- Kalshi public trade API (api.elections.kalshi.com) — live quotes, order books, candlesticks; Kalshi mirror DB (trading_events/trading_markets) — rules text and market universe.
Data: Kalshi mirror DB (rules, universe) + Kalshi public API (quotes, depth, history) as of 2026-07-22; news and nowcasts as cited. Methodology v3.3: conviction-weighted, EV-agnostic sizing; 60–90¢ entry band; sub-35¢ tails hard-rejected; 15% cluster cap; LOW conviction never deploys; rejects logged and shadow-tracked. All probabilities are subjective estimates; every contract can resolve to zero; nothing here is investment advice.