Research example · October 27, 2026 snapshot
TSA passenger screenings
A daily estimate of nationwide airport checkpoint screenings, tested on later TSA results and mapped to Kalshi's passenger-count thresholds.
All research examplesHistorical test
What the data shows
TSA observations through October 25, 2026. Errors are passengers per day; lower is better.
This is a forecast accuracy result, not a trading return. The available Kalshi series began in October 2026. A prospective record of predictions, executable prices, fees, and settled outcomes is needed before claiming a market edge.
Outcome and contract
The outcome is the nationwide checkpoint travel number for a specified date. Kalshi's listed thresholds resolve against the first published TSA count, without rounding. Later TSA revisions are excluded by the contract rule.
In the saved snapshot, 255 finalized threshold results were checked against the recorded TSA counts and all 255 matched. Several thresholds share the same daily outcome, so these are not 255 independent trading tests.
Forecast and uncertainty
The model combines the same weekday 52 weeks earlier, adjusted by the recent year-over-year traffic ratio, with the mean of the prior four matching weekdays. Its 90% annual-analog weight was selected using 2024–25 only.
Each 2026 test forecast uses data available at least two calendar days before the target. Error bands and threshold probabilities come from 2024–25 residuals. The 80% band covered 84.2% of held-out 2026 days.
Recent held-out dates
| Date | TSA count | Model estimate | Absolute error |
|---|---|---|---|
| Oct 25, 2026 | 2,837,625 | 2,808,805 | 28,820 |
| Oct 24, 2026 | 2,117,643 | 2,089,741 | 27,902 |
| Oct 23, 2026 | 2,729,432 | 2,715,314 | 14,118 |
| Oct 22, 2026 | 2,781,103 | 2,720,804 | 60,299 |
Selected recent days illustrate the saved test record; the 298-day summary above uses all held-out dates.
Method
From source count to tracked decision
Collect
Normalize TSA daily counts and record the exact contract settlement rule.
Estimate
Use earlier annual and recent weekday patterns to forecast each target date.
Validate
Tune on 2024–25 and compare the 2026 holdout with a previous-week baseline.
Track
Save predictions and pre-publication quotes, then compare settled results after costs.
Client project scope
Controls and handoff
A client implementation can be tailored to its data and review process.
Adjustable controls
Possible controls include the annual-pattern weight, recent-history window, probability band, threshold, and minimum price difference required before further review. Each change should be tested on dates that were not used to select it.
Operating process
The deliverable can include a dated source refresh, forecast log, quote capture, outcome check, and written instructions for the client's team. Holiday shifts, publication delays, revised counts, liquidity, and fees remain part of the review.
Source record
Published data and contract
This is a dated research example, not a current forecast or a recommendation to trade. The private workspace holds forecast and quote snapshots; no realized return is stated.
Discuss a model project