Do Kalshi alert prices match the order book?
In our first observed order book after an alert, the lowest displayed ask was above the logged alert price in 88 of 295 quoted MLB observations (29.83%) and 22 of 62 NFL observations (35.48%). The first requests were typically about one to two seconds after the alert timestamp.
This measures quote availability at observation time. It does not establish what anyone could have filled at the earlier alert timestamp, or that an alert caused a price move.
The first observed book
We collected 2,872 observations across 359 distinct MLB and NFL alert IDs, August 30–October 14. This sample has no BTC order-book observations. A displayed ask and displayed quantity can disappear before an order reaches the exchange.
| Sport | Eligible first books | Books with ask | Ask above logged price | At least 1 visible contract at logged price or better | Median time after alert |
|---|---|---|---|---|---|
| MLB | 296 | 295 | 88 / 295 (29.83%) | 206 / 296 (69.59%) | 2.19s |
| NFL | 63 | 62 | 22 / 62 (35.48%) | 40 / 63 (63.49%) | 1.42s |
One first book in each sport had no ask. Those books remain in the visible-depth denominator. The quoted-ask comparison uses only books with an ask. Fractional depth below one contract does not meet the one-contract threshold.
What was visible at later observations
These are separate scheduled observations, not continuous availability or order fills. The clock starts when the collector accepts an alert; the measured elapsed time from the alert is shown alongside it.
| Sport | Scheduled offset | Eligible active books | Median elapsed from alert | At least 1 contract at logged price or better |
|---|---|---|---|---|
| MLB | 30s | 296 | 31.59s | 177 (59.80%) |
| MLB | 60s | 296 | 61.65s | 153 (51.69%) |
| MLB | 120s | 296 | 121.65s | 143 (48.31%) |
| MLB | 300s | 291 | 301.62s | 122 (41.92%) |
| NFL | 30s | 63 | 30.98s | 36 (57.14%) |
| NFL | 60s | 63 | 60.94s | 34 (53.97%) |
| NFL | 120s | 63 | 120.92s | 33 (52.38%) |
| NFL | 300s | 60 | 300.98s | 32 (53.33%) |
Method and limits
After alert fan-out, a separate observer requests the public Kalshi order book at scheduled offsets of 0, 15, 30, 60, 90, 120, 180 and 300 seconds. We derive a YES ask from each resting NO bid: ask = $1 minus the NO bid. We add displayed quantities priced at or below the logged alert price. This follows Kalshi’s documented order-book representation.
- Exclude synthetic records and observations at or after October 15, 00:00 UTC. Sixteen synthetic snapshots are excluded; 2,872 real observations remain.
- Analysis rows require an active market, a successful market-status fetch, collector start within 10 seconds of the alert, and scheduled-time drift no greater than 3 seconds. The CSV retains every real observation and marks eligibility.
- Only ten returned price levels are measured. Visible depth is a lower bound on depth beyond that request, and is never a guaranteed fill.
- The order book and market status are fetched separately. Logged alert prices can come from a different quote source or side of the spread. This study cannot isolate latency from spread or quote-source differences.
- Repeated alerts in a game are correlated. This is a descriptive sample of our collected alerts, not an exchange-wide rate or an investment-return study.
There is no evidence here that our signal gives a tradable price advantage. Check the live order book and your own order confirmation before treating a logged price as executable.
Download and reproduce
All observations, timing and raw NO bid levels (CSV) · Summary tables (CSV) · Definitions, exclusions and input hashes (JSON) · Independent CSV check (Python).
Cite: DegenHedge Research, “Do Kalshi alert prices match the order book?”, October 15, 2026. Frozen data through October 14. The graded results log is a separate record of outcomes, not a record of fills.