What the live reading means

The app describes who leads and compares the observed shots, shots on target, possession, corners and cards. More possession or shots alone does not establish which team is stronger or what will happen next.

Experimental goal model

Where both teams’ xG and red-card counts are available after 20 minutes, we estimate each team's goal rate as xG ÷ elapsed minutes × 90. Independent Poisson projections use those constant rates over the approximated remaining 90-minute duration. This crude extrapolation is untrained and uncalibrated. It ignores tactical changes and added duration for stoppage time. No recommendation is emitted when the input is missing, too sparse, stale, outside the model range, at half-time, at 90+ minutes, or affected by dismissals.

Refresh is not guaranteed latency

Fixture requests use a 15-second server cache; statistics use a 60-second cache because the provider describes minute-level statistics updates. The browser refreshes every 30 seconds while visible. “Retrieved” timestamps record when our server fetched the data, not the time of the most recent on-field event. Background refresh pauses.

API allowance

One active browser watching one selected match uses approximately 180 provider requests per hour, plus initial requests and extra requests after score/phase changes or when switching matches. A local daily budget starts at 100 requests. The provider's free plan is useful for connection testing and has limited season coverage; it is not enough for continuous monitoring. Select a suitable allowance before increasing the local budget.