Phoenix, AZHow this worksTerms
SeatResale

How SeatIQ works

Two numbers do the work: a Fair Price Score on every listing, and a buy-now-or-wait forecast on every event. This page explains both, and publishes how often the forecast is right.

The Fair Price Score

Every listing gets a score from 0 to 100. It answers one question: is this a good price for these particular seats? — not “is this cheap”, which you could work out by sorting.

We normalise each listing’s all-in price by the quality of its section (view and proximity), then rank it against every comparable listing for the same event. A front-row seat costing more than a nosebleed is not a worse deal, and the score reflects that.

The score always comes with its reasoning in plain language. If we can’t explain why a listing scored what it did, we don’t show a number.

Current model: seatiq-heuristic-v1.0. Version 1 is a transparent heuristic, not machine learning. It ships first because it is explainable and good enough to be useful; the learned model replaces the scoring function without changing anything you see.

The buy-now-or-wait forecast

Ticket prices move most violently in the final 72 hours before an event. That is where a forecast is worth the most, so that is where we concentrate. We project the median all-in price at doors and show it as a range, never a single number.

When we tell you to wait, we offer to watch the event and alert you at your target price. A recommendation you can’t act on isn’t advice, so the two always ship together.

How often are we right?

This is the part most price-prediction tools don’t publish. We do, because a forecast you can’t audit is worth nothing.

CohortForecastsDirection correctActual inside range
Inside 24h
24–72h
3–14 days
Over 14 days
The bar we set ourselves, before measuring. The forecast does not ship below 78% direction-correct, per time horizon rather than blended, and it must beat a naive “always say wait” strategy by at least 10 points — otherwise it is a constant wearing a lab coat. The “wait” call is held to that bar on its own, because a wrong “wait” costs you money or the seat. If a horizon misses the bar, that horizon ships with no forecast at all.

We are publishing this standard before we have results, so we cannot quietly move it afterwards.
Not yet measured. These cells are deliberately empty. The forecast trains on a longitudinal record of how listing prices moved as each event approached, and that record starts accumulating when ingest goes live. Publishing invented accuracy figures would defeat the entire purpose of this page, so the table stays blank until there is a real backtest behind it.

What we don’t do

Image credits

Event artwork comes from the ticketing feed itself. Ticketmaster’s Discovery API returns images supplied by the promoter, venue or rights-holder and cleared for display by API partners — so the picture you see on an event really is that performer, supplied by the people entitled to supply it.

Where an event has no feed artwork, we fall back to our own atmospheric artwork keyed to the venue type — a stage, a ballpark, an auditorium. Those images are never presented as depicting the act, and we do not generate likenesses of real performers, teams, mascots or logos.

Nothing here is scraped, and nothing is licensed editorial stock. Below is the provenance of every fallback image we ship, including the model that produced it.

Used forSourceOriginLicence
arena-stageGenerated: a large indoor arena during a live music concert, seen from high in the standscogview-4-250304 (z.ai)AI-GENERATED
arena-courtGenerated: an empty indoor basketball arena before doors opencogview-4-250304 (z.ai)AI-GENERATED
ballpark-duskGenerated: an outdoor baseball stadium at dusk under floodlightscogview-4-250304 (z.ai)AI-GENERATED
amphitheatre-nightGenerated: an outdoor amphitheatre at night during a concertcogview-4-250304 (z.ai)AI-GENERATED
theatre-warmGenerated: an ornate empty theatre auditorium lit warmly before a performancecogview-4-250304 (z.ai)AI-GENERATED

Generated images are produced in two stages: an art-direction model writes the scene, then an image model renders it. Both stages are instructed to exclude faces, real individuals, team marks, logos and text. Every generated image is reviewed before it ships.

Questions or corrections

If something here is wrong — a price, an event detail, an image we should not be showing — tell us at support@seatresale.com. Content-removal requests are actioned within 24 hours.