Methodology

How Scoresight builds a verdict

A historical baseline, today's team news, and five analysts. The number you see is a win probability for the toss you chose.

The data

Scoresight indexes 6,300+ professional T20 matches: about 5,100 T20 internationals and 1,100 IPL matches. Each match keeps the teams, venue, toss, and ball-by-ball record.

The baseline model

Before any live news, the model looks at how these teams have played each other, how the ground treats batting first and chasing, and recent batting and bowling form. That produces a starting win probability.

Live team news and conditions

A separate pass reads public team news for the match: likely XIs, injuries, and conditions. It can move the baseline when the published squad is different from the one the history assumes. If that source times out, the report says so instead of inventing a lineup.

The five analysts

Four specialists run together. A chief analyst reads all four and writes the verdict.

Specialist 01 of 4

Head-to-head

How these two teams have played each other before.

Chief analystYour report

The chief analyst reads all four before writing the verdict.

How we measure accuracy

A call is correct when the team we favoured wins. Ties and abandoned matches are excluded. We publish the rate only from official predictions made before the match, not from reports people re-run afterwards. See the track record

Limitations

The report uses the toss you enter. It does not score the other toss. Dew, a late team change, and a small head-to-head sample can all move the result after the report is written. Thin samples are labelled as thin.

Responsible use

Scoresight is information for personal use. It is not betting advice and it does not accept wagers. Past results, once published, will not guarantee the next match.