Kalshi Business — Mispricing Audit

Report date: 2026-05-13  ·  Category: Business (Companies + Financials in Kalshi taxonomy)  ·  Horizon: contracts closing by 2026-06-27  ·  Capital: $10,000

Of 465 active business contracts closing in the next 45 days, the volume + spread + non-noise filters cut the field to a working set of 16 mid-priced markets. After diligence I keep six picks — two BUY NOs and four BUY YESs — totalling $9,000 deployed / $1,000 reserved. The biggest single edge is the BUY NO on KXCAVA-26MAYREST-457, where the market is pricing a Q1 restaurant count Cava has not yet reported with a degree of confidence that the historical opening cadence does not justify.

1. How this was researched

The audit ran in four passes:

  1. DB filter. Pulled all status='active' markets in trading_events.category IN ('Companies','Financials') from /tmp/bot.db with close_time in the next 45 days. 465 rows. Dropped FX/interest-rate range markets (KXUSDJPY, KXEURUSD, KXTNOTEW, the daily S&P/Nasdaq grid) — these are short-window noise, not business mispricings.
  2. Liquidity + price filter. Kept rows with volume24h ≥ 1500, top-of-book 5¢ ≤ yes_bid ≤ 95¢, and spread (yes_ask – yes_bid) ≤ 5¢. Cut to 14 names. Hand-added two adjacent strikes (KXCAVA-458, KXKLAR-118) to read the full distribution.
  3. Live verification. Hit /trade-api/v2/markets/{ticker} and /orderbook for each. Caught one stale row (the local DB still shows KXPM-26APRZYNSHIP-190 as active but the live API has status="finalized" with result "no" — Philip Morris reported 155M ZYN US-shipment cans on 22 April and the market closed the same day at 1¢).
  4. News diligence. For each survivor I pulled the rules text, mapped the resolution to a specific company KPI or event, and grounded a true-probability estimate in published earnings releases, Wikipedia launch logs, FAA/launch-schedule databases, and the issuing company's investor relations page. Primary sources are linked at the bottom.

2. Markets reviewed

Working set after Stage 1. Last column is my verdict; each BUY / SELL picks gets a dedicated section below. PASS = priced about right or thesis too fragile to bet.

Ticker Subject YES bid / ask Vol 24h OI Close Verdict
KXSPACEXSTARSHIP-12-26MAY19Starship IFT-12 by May 1963 / 694,3823,47405-20BUY NO
KXSPACEXSTARSHIP-12-26MAY31Starship IFT-12 by May 3193 / 9415,98584,20406-01PASS — fair
KXSPACEXCOUNT-26MAY-10>10 SpaceX launches in May93 / 942,36415,88606-01PASS — fair
KXSPACEXCOUNT-26MAY-12>12 SpaceX launches in May47 / 483,80820,73506-01BUY YES
KXSPACEXCOUNT-26MAY-13>13 SpaceX launches in May12 / 1310,85041,38206-01BUY YES
KXCAVA-26MAYREST-457Cava >457 restaurants Q175 / 7914,75712,29706-13BUY NO
KXCAVA-26MAYREST-458Cava >458 restaurants Q126 / 273,8193,62306-13WATCH
KXCAVA-26MAYREST-460Cava >460 restaurants Q18 / 940,559110,86806-13PASS — fair
KXCOST-26MAYCARDS-150MCostco >150M cardholders Q316 / 178,19215,39706-27BUY YES
KXKLAR-26MAYCONS-118MKlarna >118M consumers Q198 / 992,2742,24005-15PASS — capped
KXKLAR-26MAYCONS-120MKlarna >120M consumers Q179 / 825,5094,56405-15PASS — fair
KXKLAR-26MAYCONS-122MKlarna >122M consumers Q117 / 184,2304,37005-15BUY YES (small)
KXLOW-26MAYTRAN-195MLowe's >195M Q1 transactions69 / 701,5501,02605-21PASS — fair
KXLOW-26MAYTRAN-204MLowe's >204M Q1 transactions5 / 62,0641,20305-21PASS — fair
KXOMEGA-MAY-5534Bezel Omega Index >$5,53441 / 421,9111,37405-31PASS — opaque
KXPM-26APRZYNSHIP-190MPM >190M ZYN US-ship Q1REJECT — finalized NO 4/22

3. Top picks

BUY NO KXSPACEXSTARSHIP-12-26MAY19 · Starship Flight 12 launches by May 19, 2026

Limit:
NO at $0.36 (= sell YES at $0.64)
Best live quote:
YES 0.63 bid / 0.69 ask  →  NO 0.31 bid / 0.37 ask
My true probability:
P(launch by May 19) ≈ 45%  →  P(NO) ≈ 55%
Edge:
+19¢ per NO contract  ·  +53% EV at fill

Mispricing thesis

SpaceX itself is publishing a no-earlier-than (NET) date of May 19, 2026, 5:30pm CT with a broader window of May 15–21. The Kalshi rules resolve YES only if the launch happens before May 20 — i.e. on the NET date itself with no slip. The market is pricing this at 63% YES, which assumes essentially no scrub risk on a debut V3 vehicle.

Evidence

Tail risks (what makes me wrong)

Liquidity & execution

Thin. Visible YES bid depth is ~50 contracts total across 0.55–0.62; deeper NO ask only fills materially at 0.37–0.41. I size this conservatively at $1,500 and place limit BUY NO at $0.34, willing to walk up to $0.38 average. If unfilled by close-of-business 5/15 I drop the limit — by then the rocket either rolls or the price gaps.

Price history (last 14 days)

Drifted from 70% YES (5/1) → 67% (5/6) → 66% today as the V3 reality set in. Hasn't broken lower because traders don't want to be short going into a confirmed launch window. Plenty of room to fall toward the actual NO probability if the rocket scrubs once.

BUY NO KXCAVA-26MAYREST-457 · Cava reports >457 restaurants in Q1 FY2026

Limit:
NO at $0.21 (= sell YES at $0.79)
Best live quote:
YES 0.75 bid / 0.79 ask  →  NO 0.21 bid / 0.25 ask
My true probability:
P(Q1 restaurant count > 457) ≈ 50%  →  P(NO) ≈ 50%
Edge:
+29¢ per NO contract  ·  +138% EV at fill

Mispricing thesis

The three adjacent strikes in this event imply the market is concentrated on Q1 2026 = 458 restaurants exactly — >457 YES at 78%, >458 YES at 26%, >460 YES at 8%. That gives an implied P(Q1 = 458) ≈ 52%, with only ~22% mass at ≤457. Cava has not yet reported — earnings are May 19 — and the historical Q1 opening cadence does not justify this much certainty.

Evidence

Tail risks

Liquidity & execution

OI 12k, decent depth on the NO side (~400 contracts at 0.14, 450 at 0.15, 840 at 0.17). Limit BUY NO at $0.20 should fill at least partially overnight; walk up to $0.23 average. Size $2,000.

Price history

Strike opened 5/12 at 31/69 wide quote, settled today at 75/79 on the back of 14.7k volume — but most of that volume looks like a single block accumulating YES exposure to read a 458 print. The adjacent >460 strike crashed 50% → 14% on 5/7 on 25k volume, which tells me the market took out the right tail aggressively — but did not necessarily nail the center.

BUY YES KXSPACEXCOUNT-26MAY-12 · SpaceX has >12 launches in May 2026

Limit:
YES at $0.48
Best live quote:
YES 0.47 bid / 0.48 ask
My true probability:
P(launch count ≥ 13) ≈ 60%
Edge:
+12¢ per YES contract  ·  +25% EV at fill

Mispricing thesis

Through May 12 SpaceX has flown 56 Falcon-family vehicles in 2026 (Wikipedia / Spaceflight Now). YTD through April was 52, so May-to-date is 4 launches in 12 days. The market has extrapolated the slow first half to a ~10–11 monthly total and priced >12 at 47% YES. The published launch schedule says otherwise.

Evidence

Tail risks

Liquidity & execution

OI 20.7k, tight 1¢ spread. Place limit BUY YES at $0.47, willing to lift to $0.49. Size $2,000.

BUY YES KXSPACEXCOUNT-26MAY-13 · SpaceX has >13 launches in May 2026

Limit:
YES at $0.13
Best live quote:
YES 0.12 bid / 0.13 ask
My true probability:
P(launch count ≥ 14) ≈ 28%
Edge:
+15¢ per YES contract  ·  +115% EV at fill

Mispricing thesis

Mirror of Pick 3. If my May-total distribution puts mass at 12–14 launches with a meaningful right tail, then >13 (i.e. ≥14) is a 25–35% event, not a 12% event. Same Spaceflight Now / Wikipedia schedule and FAA cadence data underlie both calls.

Position rationale

I am buying both >12 and >13 because they hedge nicely: if total = 13 exactly, the >13 ticket loses but >12 wins; if total = 14+, both win. The >13 ticket is the right-tail call at 13¢, where I'm risking $0.13 to make $0.87 if SpaceX matches the kind of week-three sprint we saw in March (5 launches in 8 days).

Liquidity & execution

OI 41.4k, tight 1¢. Limit YES at $0.12. Size $1,000.

Tail risks

If a Starship anomaly or vehicle stand-down lands in the next 18 days, Falcon throughput drops and 14+ becomes very hard. The recent price drift from 60% → 12% on this strike does suggest the market is fading the right tail aggressively.

BUY YES KXCOST-26MAYCARDS-150000000.0 · Costco reports >150M cardholders in Q3 FY2026

Limit:
YES at $0.17
Best live quote:
YES 0.16 bid / 0.17 ask
My true probability:
P(Q3 cardholder count > 150M) ≈ 30%
Edge:
+13¢ per YES contract  ·  +76% EV at fill

Mispricing thesis

Costco's cardholder count grows ~1.7–2.1M per quarter. The trajectory through FY2026 is on a knife-edge with 150M:

PeriodCardholdersΔ vs prior Q
Q1 FY26 (end Nov 2025)145.9M
Q2 FY26 (end Feb 2026)147.8M+1.9M
Q3 FY26 (end May 2026, reported late May)?consensus ~+1.7M → ~149.5M

At trend that's 149.5M, just under threshold. But April retail-month sales (4 weeks ending 5/3) printed +13.0% net / +11.6% comp — among the strongest comp months in two years, with Q2's +13.6% membership-fee growth still trending. A +2.1–2.4M cardholder quarter would clear 150M cleanly.

Evidence

Tail risks (real, not theoretical)

Liquidity & execution

OI 15.4k, tight 1¢ spread, ~7k contracts resting at 0.16. Limit BUY YES at $0.16, walk to $0.17. Size $1,750.

BUY YES KXKLAR-26MAYCONS-122000000 · Klarna reports >122M active consumers in Q1 2026

Limit:
YES at $0.18
Best live quote:
YES 0.17 bid / 0.18 ask
My true probability:
P(Q1 consumers > 122M) ≈ 25%
Edge:
+7¢ per YES contract  ·  +39% EV at fill

Mispricing thesis

Klarna reports Q1 2026 tomorrow, 14 May, before market open — the market closes 5/15. Quarterly consumer adds have been accelerating with US growth: 111M (Q2 2025) → 114M (Q3) → 118M (Q4). Extrapolating the +3–4M/quarter ramp puts Q1 2026 at 121–122M, with a real chance of overshoot given the US card / banking-services momentum the company called out in the Q4 release.

Why this is sized small

The >122M strike crashed from 39% to 17% during today's session on 2k volume. That is consistent with someone reading a leaked or whispered number, or simply with a re-pricing as the report gets closer. Either way I do not have the same information edge as I do on Cava or Starship. I take a small position to express the view that even ~25% probability is meaningfully above 18% market price.

Liquidity & execution

OI 4.4k, tight 1¢ spread. Place limit at $0.17. Size $750.

Hedge

The adjacent >120M strike at 79–82% YES is roughly fair on my model (I have P(>120M) ≈ 75%). If I wanted to fade the same information, that's where I'd short — but the edge is <5¢ so it isn't worth a separate ticket.

4. Recommended $10,000 portfolio

# Ticker Side Limit Contracts Cost Max payout P(win) EV ¢ EV %
1KXSPACEXSTARSHIP-12-26MAY19NO0.364,170$1,500$4,17055%+19+53%
2KXCAVA-26MAYREST-457NO0.219,520$2,000$9,52050%+29+138%
3KXSPACEXCOUNT-26MAY-12YES0.484,170$2,000$4,17060%+12+25%
4KXSPACEXCOUNT-26MAY-13YES0.137,690$1,000$7,69028%+15+115%
5KXCOST-26MAYCARDS-150MYES0.1710,300$1,750$10,30030%+13+76%
6KXKLAR-26MAYCONS-122MYES0.184,170$750$4,17025%+7+39%
Deployed$9,000$40,020
Reserve (opportunistic adds)$1,000

Expected return

Risk profile & concentration

Execution notes

5. What I rejected and why

KXPM-26APRZYNSHIP-190M — REJECT, already finalized

The local DB still flags this market active. The live API says status:"finalized", result:"no" as of 22 April 2026. Philip Morris reported ZYN US shipment volume of 155M cans for Q1 2026 (vs ~175M offtake) on 4/22; the market closed at 1¢. Surfacing this is the audit's main reminder to never trust the local DB's status alone.

KXSPACEXSTARSHIP-12-26MAY31 — PASS, price right

YES 93/94, P(launch by 5/31) ≈ 90–93%. Even with V3 vehicle / Pad 2 risk, a 12-day buffer past NET captures essentially the entire scrub-and-retry distribution. Edge ≤2¢, not worth crossing 1¢ spread for.

KXSPACEXCOUNT-26MAY-10 — PASS, price right

YES 93/94. With ≥9 launches already locked in by 5/22, hitting 11 by 5/31 is a near-certainty (~95–97%). Edge ~3¢ at fill, not enough for the capital cost.

KXCAVA-26MAYREST-460 — PASS, market correctly skeptical of right tail

YES 8/9. My model puts P(Q1 ≥ 461) ≈ 5–8%. Price is right; no edge.

KXCAVA-26MAYREST-458 — WATCH, modest YES edge but bigger sister trade in pick #2

YES 26/27. My P(Q1 ≥ 459) ≈ 31%, so a small (+4¢) YES edge exists. The reason I don't take it: the BUY NO on the >457 strike (Pick 2) expresses the same view on the joint distribution with a much larger edge, so doubling up on adjacent strikes adds correlation without adding meaningful new alpha. If Pick 2 fills early at a great price, I might add a small >458 YES with the reserve.

KXKLAR-26MAYCONS-120M — PASS, market roughly fair

YES 79/82. My P(Q1 > 120M) ≈ 75%. With 3¢ spread, edge is sub-2¢ net of slippage. Not enough.

KXKLAR-26MAYCONS-118M — PASS, fully priced

YES 98/99 with Klarna already publicly claiming "over 118M" users in marketing materials. Essentially resolved already; 1¢ left isn't worth the capital.

KXLOW-26MAYTRAN-204M / KXLOW-26MAYTRAN-195M — PASS, info-limited

Lowe's Q1 reports 5/20. Management telegraphed Q1 comps "below midpoint" of full-year guide due to February winter storms. Range of likely Q1 transactions is ~190–202M; the strikes at 195 and 204 are placed exactly to bracket the consensus. 70% YES on >195 and 5% YES on >204 both look about right; my edges are within the spread. Skip.

KXOMEGA-MAY-5534 — PASS, opaque

Bezel Omega Index resolves on a single 3rd-party-published price as of 5/31. The index methodology, current level, and constituent watches are not easily verifiable from primary sources within the audit window. With OI 1.4k and ~50% pricing, there is no way to claim a defensible edge.

~110 strikes never crossed Stage 1 (FX, intraday S&P/Nasdaq ranges)

The Companies + Financials category in the DB is dominated by KXUSDJPY, KXEURUSD, KXNASDAQ100, KXTNOTEW, and KXINX grids — hourly or daily strike-range markets. These aren't business-mispricing opportunities; they're short-window microstructure plays where any edge would be eaten by the spread. Filtered out at Stage 1.

6. Sources

All numbers in this report trace to one of these primary sources or to the local Kalshi snapshot DB.

Kalshi (market data)

SpaceX / Starship Flight 12

SpaceX May 2026 launch cadence

Cava Group

Costco

Klarna

Philip Morris (for the rejected ZYN market)

Data sources: Kalshi public REST API (api.elections.kalshi.com/trade-api/v2) for live quotes and orderbooks; local Kalshi snapshot DB (/tmp/bot.db, last refreshed during this audit) for price history and 24h volume; company investor-relations pages and major news outlets for KPI grounding. Probability estimates in this report are subjective and reflect my best interpretation of public information at 13 May 2026. Kalshi contracts that resolve against the holder pay $0.00; the maximum loss per pick equals the capital cost shown. This audit is not investment advice; sizes are illustrative and ignore taxes, fees, and broker withdrawal frictions.