Skip to content

Predictions

A prediction is a call that an upcoming event’s primary tickets will resell above their price: bought at the primary floor when the call is made (entry), it is graded on what actually sold on the resale market (marketplaces and Ticketmaster resale) between the call and the event, never on listings alone. The exit is the median price of those sales; the call is a hit when the exit clears the call’s requiredMarkup (fees and margin), a miss when it doesn’t, and void when there were fewer than 3 resale sales to grade it on.

Predictions need an API key with market:read; every plan includes them. They open to customers once they are published: the category’s current model has a track record of enough graded calls and status302 switched them on for your organization. Until then the calls and the track record answer 403 with the problem type https://status302.com/problems/predictions-unpublished.

Calls come in two categories, each with its own model, track record and publication. Pick one with category on the calls and the track record; tm is the default and an unknown value answers 400.

category Events Model looks at
tm (default) Ticketmaster Primary sales and inventory, TM resale and marketplace prices
dice DICE The lineup’s strongest act, the venue’s DICE history, DICE ticket-type demand and CrowdVolt resale

A dice call weighs:

  • The strongest act on the lineup: its Last.fm listeners, Spotify followers and popularity, how many of its past DICE shows sold out and how far above face they resold, and its Ticketmaster resale history when its name matches exactly one attraction.
  • The venue: how many of its past DICE shows sold out and how far above face they resold.
  • Demand: the share of the event’s DICE ticket types sold out, and whether a waiting list is open.
  • CrowdVolt resale: the markup over the primary floor, its momentum and the market’s liquidity.

A DICE event gets a call only once it is linked to a marketplace listing. DICE calls are graded exactly like TM calls: entry at the primary floor when the call is made, exit at the median price of the resale sales between the call and the event, a hit when the exit clears entry +20%.

GET /v1/predictions/availability tells whether a category is published yet:

{
"available": false, "reason": "unpublished", "currentModel": "v1", "graded": 41, "minGraded": 50,
"categories": [
{ "category": "tm", "available": false, "reason": "unpublished", "currentModel": "v1", "graded": 41, "minGraded": 50 },
{ "category": "dice", "available": false, "reason": "unpublished", "currentModel": "dice-v1", "graded": 3, "minGraded": 50 }
]
}

reason is unpublished and is absent once available is true (plan no longer occurs: every plan has predictions). The top-level fields describe the tm category; categories has one entry per category, tm first, since each is published separately.

Terminal window
curl -H "Authorization: Bearer $STATUS302_API_KEY" \
"https://api.status302.com/v1/predictions?category=dice&status=open&limit=50"
status Lists Order
open (default) Calls on events that haven’t started Lowest risk first
graded Hits and misses Most recently graded first
hit, miss, void One outcome Most recently graded first

Each call has its category, the event, the model version that made it, risk (the chance, 0–100, that the call is wrong; frozen when the event starts), riskAtCall, reasons (strongest factor first), entry and the latest resale floor in currency (minor units; see money) with their markup. Graded calls add gradedAt, exit, exitMarkup and outcome. Pages follow nextCursor (with the same status and category; see pagination).

GET /v1/predictions/track-record?category=dice says how one category’s calls turned out (tm by default), overall and per model version, all time and over the last 30 and 90 days (by grading time): hits, misses, voids, hitRate, meanExitMarkup, the brier score of the risk at the call (0 is perfect, 0.25 a coin flip) and bands, the calibration per 10-point band of the risk at the call (expectedMissRate against the observed missRate). It is recomputed at most every minute.

A must cop is an upcoming public on-sale (Ticketmaster or DICE, within 60 days) of an act whose latest known on-sales sold out instantly: at least 2 of the act’s last 3 (the last two years, both sellers) sold out within two hours of their public on-sale. A pick is a hit when its own on-sale sells out within two hours too. Picks are refreshed every 15 minutes.

Terminal window
curl -H "Authorization: Bearer $STATUS302_API_KEY" \
"https://api.status302.com/v1/must-cop?status=open"

GET /v1/must-cop needs market:read like the calls and takes the same status, cursor and limit: open picks soonest on-sale first, graded ones most recently graded first. Each pick has the event, the act, onSaleAt, reasons and the releases (the act’s on-sales it rests on, with their soldOutAt); record counts every graded pick. Must cops are published on their own: until then the list answers 403 with the problem type https://status302.com/problems/must-cop-unpublished. Like the calls, they are not financial advice.