LIVE VS PREMATCH
Pinnacle live vs prematch odds drops.
Same Pinnacle odds API feed, two tempos. Prematch Pinnacle prices drift toward kickoff over hours; live prices react to the game in seconds. Detection rules that work for one rhythm misfire on the other, so it pays to know which one you are watching.
Two rhythms of the same Pinnacle odds feed
Prematch markets open days before kickoff and move on information: team news, confirmed lineups, weather, market weight. A 5% drift across an afternoon is unremarkable on a popular soccer market, and most of it arrives in slow steps rather than jumps.
Live markets move on events. A goal, a red card or a break point can deliver the same 5% in under a minute, and a tennis market in play can reprice on every point. The raw price series looks similar in a log; the cadence underneath is completely different, and your drop rule feels that difference first.
That difference is why one global rule underperforms. A threshold tuned to
catch prematch drift fires constantly in play, where 3% moves are routine.
A threshold tuned for live chaos almost never fires prematch, where moves
of that size are rare and slow. Splitting rules by phase is the standard
fix, and the feed already makes the split: /odds-drop carries
live drops and /odds-drop-prematch prematch ones, each on its
own connection.
Suspensions and bounce-backs
Live markets suspend when something material happens: a goal, a penalty call, a VAR review. While a market is suspended there is no tradable price, and the first price after it reopens can sit a long way from the last one before it.
Treat a fall measured across a suspension as a repricing, not a drift: the
gap may already be closed at your bookmaker by the time you act. If your
own rule keeps a window, reset it when a market resumes and compare from
the first price after the suspension. Prematch prices can also dip and
bounce back within seconds; the prematch stream's recheck=N
option holds each drop N seconds and sends it only if the price is still
down.
Volume differences
Live produces far more moves. A single tennis match in play can generate more price changes than its entire prematch lifetime. The feed's own guidance puts prematch at under one drop a second on a typical day and the live stream at tens a second in prime hours.
Size your consumer for the live case even if you mostly care about prematch: bounded queues, batched writes, and an explicit policy for falling behind, because the stream does not wait for a slow reader. Measure against your own sport mix before you commit to capacity.
If the consumer does fall behind, shed events rather than block the connection. A missed alert is recoverable on the next move; a stalled reader that forces a reconnect costs you the whole backlog. The SSE setup guide covers reconnects and idempotent processing in detail.
Rule examples for each mode
A prematch rule might be 5% within 30 minutes, top-flight leagues only, 1X2 and totals markets. It catches sustained drift on team news and ignores single-tick noise, and the alert volume stays low enough for a human to read.
A live rule might be 3% within 2 minutes, skipping moves measured across a suspension, deduping by event id, with a cap of one alert per selection per match so a chaotic minute does not page you twelve times. Shorter window, smaller threshold, stricter dedupe: the opposite shape from prematch, for the opposite rhythm.
Both are starting points, not finished rules. Read how thresholds and windows compose in the drop alerts guide, then connect the stream with the SSE setup guide.