Predictions
A win-probability model for every fixture on the desk, with methodology disclosed and last-five-form inputs.
01 What the desk publishes
For each fixture on the desk we publish a single probability — the team most likely to win, expressed as a percentage with a 90-percent confidence interval. We do not publish per-innings or per-player probabilities beyond that.
The model is a logistic regression over the last 30 completed innings in the same format, with three inputs: venue-adjusted runs, average powerplay score, and a captaincy-pressure surrogate (the team that lost the last match). The full method is at methodology.
02 Why a single number, not a card
A win-probability desk that fans out into per-batter and per-bowler forecasts drifts toward betting-product language. We do not want that. We publish one number per fixture because that is the single number the editorial desk has a defensible model for.
For per-player fantasy reads — captain, vice-captain, differential — the Fantasy tips and Captain & vice-captain desks do that work directly, with their own inputs.
03 The chart
On the homepage and on every match page, the prediction is shown as a small chart. The chart has two bars — one per team — and a confidence band on the larger bar. The chart is generated from the model output, not curated by an editor.
If you want to see how the chart moves across the season, the Players index carries a form-trend strip for the most-picked players in every match.
04 When the desk does not call it
When two teams are within two percentage points of each other, the desk publishes “TOO CLOSE TO CALL” with the underlying model output hidden. The reasoning: a 51–49 probability is, for any practical purpose, a coin flip and we do not want the editorial weight of a number that is not meaningfully different from 50–50.
A “TOO CLOSE” tag will resolve at the toss or shortly after if the conditions force a clearer model output.
15 Frequently asked
The questions readers most often send to the desk, with answers rooted in the public editorial standards and the platform’s published terms.
What model does the desk use?
A logistic regression over the last 30 completed innings in the same format. Inputs are venue-adjusted runs, average powerplay score and a captaincy-pressure surrogate. Full disclosure at methodology.
How is the model kept honest?
Each match-day prediction is logged before the toss with the timestamp and the published probability. After the result, we publish a calibration chart that compares all in-season calls to the actual outcomes — so readers can see for themselves how the model tracks.
Do you ever modify a published number?
No. Once a probability is published for a fixture, it stands. Mid-match updates are separately published as “live” and clearly labelled.
A note from the editor
Dream11 Zone is an independent editorial desk. We are not the operator of Dream11 and we do not place bets. Real-money fantasy contests are restricted in Andhra Pradesh, Assam, Odisha, Telangana, Sikkim, Nagaland and Tamil Nadu. If you are under 18, do not register. If you are in a restricted state, do not deposit. If you are in a permitted state, deposit only what you can afford to lose and use the platform's deposit limits and self-exclusion tools. Editorial standards at methodology; state rules at is Dream11 legal?; contact at contact.