T20, ODI, Test: What Is Actually Predictable in Each Format
The same question is easy in one format and near-random in another. A guide to where skill beats noise in each of cricket's three formats — and which markets to leave alone.
A forecaster's most valuable skill is knowing which questions are answerable. Cricket makes this unusually concrete, because the same question — will this batter score fifty? — has a completely different amount of signal in a Test match than in a T20.
Format changes the balance between skill and noise. Here is how, and what to do about it.
The core principle: more balls, less noise
The longer the format, the more the better team wins.
A T20 innings is around 120 deliveries per side. A single over can decide it. One dropped catch, one lucky edge, one bowler having an off day — each is a large fraction of the total sample. The better side wins comfortably more than half the time, but nowhere near as often as their quality gap suggests.
A Test match runs to a maximum of five days and often 2,000+ deliveries. Luck gets averaged out. Depth, discipline and conditions dominate. Upsets happen, but they are genuinely upsets rather than a routine Tuesday.
ODIs sit in between, closer to Tests than most people assume.
Practically: be more humble in T20 markets than your knowledge feels like it justifies. If you think a side is clearly better, that conviction should translate into a smaller probability edge in T20 than in a Test — because the format itself is injecting randomness that no amount of analysis removes. A team you rate as substantially superior might be a strong favourite over five days and only a modest one over twenty overs.
T20: predictable structures, unpredictable outcomes
What is genuinely predictable in T20:
- Phase structure. Powerplay, middle, death — each has stable, well-documented scoring and wicket patterns at a given venue. Totals markets are far more tractable than match-winner markets.
- Role certainty. Who opens, who bowls the death, who finishes. Roles in T20 are more rigid than in any other format, which makes player-specific markets more analysable.
- Venue effects. Because the sample is short, ground characteristics dominate. A small ground with a flat surface produces high totals almost regardless of who is playing.
- Match-ups. Specific bowler-type against specific batter-type is a real and reasonably stable effect over a season, and it is more visible in T20 than anywhere else because captains actively engineer it.
What is not predictable:
- Individual innings. A T20 batter's score is close to a lottery even for the best players. They face a small number of balls in a high-risk mode. Markets on "will X score 30+" are much closer to a coin flip than the batter's reputation suggests.
- Close-match outcomes. Anything decided in the last two overs is decided by variance.
The T20 rule of thumb: prefer structural markets to individual ones. Totals, phase-based questions and match-ups reward analysis. Individual milestone markets mostly reward luck, and they attract price distortion from fans backing their favourite regardless of the number.
ODIs: the middle format that rewards patience
Fifty overs is enough for the game to have a shape: a start, a consolidation, an acceleration. That structure makes several things tractable.
- Totals are meaningfully predictable from surface and conditions, more so than in T20, because the innings has time to regress toward what the pitch allows.
- Top-order batters get a real sample. A quality opener facing a full innings has a genuine chance to convert. Individual milestone markets carry more signal here than in T20.
- The chase has structure. Required rate, wickets in hand and overs remaining interact in ways that make in-progress reasoning much more reliable.
The trap in ODIs is stale mental models. Scoring norms in the format have moved substantially over the past decade, and a lot of intuition — what counts as a par total, when acceleration should begin, how many wickets a chasing side should keep back — is calibrated to an older version of the game. Check what recent matches at that venue actually produced rather than what a good total "should" be.
Tests: the format where research pays most
Test cricket is the best format for a serious forecaster, and the worst for a casual one, because it demands actual work and then rewards it.
What is predictable:
- Match outcome from conditions. A surface that will deteriorate, a side with two quality spinners, a home team with a strong record — these compound over five days into a genuinely large edge.
- The draw. The most under-analysed market in cricket. Weather, over rates, surface flatness and match situation combine into a probability that casual players routinely price near zero and that is often substantial. If you are willing to think about the draw seriously, you are competing with very few people.
- Session-level dynamics. New ball, second new ball, the first hour of day one, the last hour of day four. These are stable, well-understood pressure points.
- Individual output. With a full innings and a second one, quality batters have room to demonstrate quality. Milestone markets carry real signal.
What is not:
- Anything depending on declarations. These are captain's decisions driven by a specific match state and sometimes by pure temperament. Do not model them; note the uncertainty and price accordingly.
- Day-five weather-dependent finishes. You are forecasting the weather at that point, not the cricket.
Format-agnostic mistakes worth naming
Importing form across formats. A batter in extraordinary T20 form tells you very little about their next Test innings. Different skills, different risk profile, often different role. Treat cross-format form as weak evidence at best.
Ignoring the calendar. Fatigue after a long tour, a squad rotating heavily before a bigger series, a franchise resting a player with a niggle — schedule context routinely outweighs recent statistics, and it is public information the market is slow to price.
Treating home advantage as one number. Home advantage in cricket is not a constant. It is enormous where conditions are distinctive and preparation is targeted, and modest where they are not. It is also much larger in Tests than in T20, for exactly the sample-size reason above.
Assuming rules are stable. Playing conditions change between tournaments — fielding restrictions, substitutions, over-rate penalties, tie-breaking. A market can hinge on a rule you last checked two seasons ago.
How to choose which markets to play
Sort candidate markets by how much of the outcome is decided by things you can actually research:
- High signal: Test match results, Test draws, ODI totals, T20 venue-driven totals, series outcomes.
- Medium: ODI individual milestones, T20 phase markets, match-up-driven questions.
- Low: T20 individual scores, anything decided in a final over, anything depending on a captain's discretionary call.
You do not have to avoid the low-signal markets. They are fun, and fun is a legitimate reason to play. Just size them like the lotteries they are, and put your real conviction — and your real Stars — where the work you did actually changes the odds.
Keep reading
When Prediction Markets Get It Wrong: The Limits of Crowd Wisdom
Prediction markets are good aggregators — but they are not oracles. There are specific, identifiable conditions where market prices are systematically wrong, and knowing them is a genuine edge.
3 min readDew, Rain and Weather: The Variables Most Cricket Predictors Ignore
Dew changes T20 matches more than almost any other variable — and it is consistently underweighted by predictors. A practical guide to factoring weather into pre-match calls.
5 min readRecency Bias: Why Your Last Match Always Feels More Important Than It Is
The team that won last week feels unstoppable. The team that lost three in a row feels broken. Recency bias is the most common error in sports prediction — and the easiest to identify once you know what to look for.