How CricArcade Works About the games

CricArcade runs three games: The Chase, a one-day international game; the T20 Franchise Manager, built around the IPL and the Big Bash; and The Follow-On, which lets you captain any Test side in history. They share a set of principles, and nothing on this page goes further than these eight.

1. Built on real cricket data. Every player comes from his real record, and every result we compare against is a real match.
2. A separate simulation for each format. A Test, a one-day international and a T20 are different games, and each is modelled and measured on its own.
3. Players are profiles, not a single number. How a batter scores and how often he gets out are different things, and so are a bowler's wickets and his economy.
4. Match context matters. The score, the target, the wickets in hand and the stage of the innings change what every player is trying to do.
5. Conditions matter. Grounds, pitches and the wear of a long match change how the game is played.
6. Cricket is uncertain. A rating describes a player; it does not guarantee what he does today.
7. Strategy and captaincy matter. Batting orders, bowling changes, declarations and reviews are decisions, and they change results.
8. Eras are modelled as eras. A Test in 1930 and a Test in 2020 were played under different laws, on different pitches, at a different tempo.

The Follow-On's players come from the careers of everyone who has played Test cricket for the twelve full-member nations, from the first Test at Melbourne in 1877 onwards. The Chase draws on the men's one-day international record since 1971, and the T20 game on real IPL and Big Bash seasons. Where a player's record is thin, because he played a handful of matches or in an era that kept fewer statistics, the games treat that record with appropriate caution rather than taking a lucky fortnight at face value.

A player is described by several separate qualities rather than one overall score. For a batter, how heavily he scores and how quickly he scores are different questions; Chris Tavaré and Virender Sehwag both made Test runs, at very different speeds. For a bowler, taking wickets and conceding few runs are different skills. Some batters are better against pace than spin, pace bowlers usually take the new ball while spinners come into their own as a pitch wears, and a wicketkeeper and a fine fielder matter to a side in ways that do not show up in a batting average. Keeping these qualities apart is what lets two players with similar averages play very differently.

Each format has its own simulation because the formats reward different things. In a Test, a batter who survives is winning; in the last five overs of a one-day chase, survival without runs is a loss. The same player is not the same asset in both, and a single model stretched across all of them would get at least one badly wrong.

Every delivery is a contest between a particular batter and a particular bowler, and the match around them shapes it. A new batter is more vulnerable than one who has been in for an hour, which is why a wicket just before an interval matters so much in real cricket and why fielding captains attack a fresh batter. A side chasing a big total takes risks that a side batting for a draw does not. Field restrictions in the first overs of a one-day innings change what a good shot is. None of this is decoration: the same two players produce different cricket at 30 for 3 than at 300 for 3.

Conditions follow the same idea. Some grounds are known for pace and bounce and some for turn, a Test pitch on its fifth day is not the pitch of its first morning. The games carry these differences in a way you will recognise if you watch the real thing; they do not publish the numbers behind them.

The greatest batters in history still made ducks. Don Bradman was out for under ten in about one in five of his dismissals, and a Test team of all-time greats can still be bowled out cheaply on a bad day. The games keep that uncertainty. A higher rating makes good things more likely; it never makes them certain, and a weaker side that bowls well and holds its catches can win. Our guide on why your star still gets out looks at the real numbers behind this.

Captaincy is the part you control. When you choose the order, the bowlers and the moment to declare, the games respect the choice and play it out; when the computer captains the other side, it makes the same kinds of decisions a real captain would, in cricket terms: attacking a new batter, protecting a tired bowler, setting a target it thinks it can defend. We do not publish the rules those decisions follow, for the reasons at the end of this page.

Eras are treated as eras. Timeless Tests were played to a finish; the first World Cups were 60-over matches; fielding restrictions, powerplays, a second new ball and the review system each arrived at particular moments and changed the game when they did. In The Follow-On and The Chase, a match set in an era is played under that era's conditions, which is why the same XI can score differently in 1935 and 2015.

The honest test of a cricket simulation is whether its matches look like real ones, taken in the round. The charts below put the games' simulated results beside the real scorecards in our archives, at a deliberately coarse resolution: wide bands of totals and scores, rounded to whole percentages. They show the shape of the cricket, which is the claim we are making, without publishing any figure the simulation was tuned to.

Start with Test cricket in the modern era. Each pair of bars is the share of first innings that ended in a given band, from the 1,057 real Tests played between 2000 and March 2024, and from 800 simulated Tests.

The same comparison for the second half of the twentieth century, when draws were far more common and the game was slower, uses the 1,202 real Tests from 1946 to 1999:

Team totals can look right for the wrong reasons, if every batter scores a little too much and every bowler takes a few too many wickets. So the next chart looks at individual innings: how often a batter is out for under ten, how often he passes fifty, how often he makes a hundred. This is where a simulation that treats players as real people, rather than as a team average, has to earn its keep.

One-day cricket is measured against itself, not against Tests. These charts use completed ODIs between the eight leading men's sides since 2015, leaving out rain-shortened matches, and simulated matches between the same sides' current squads.

The charts are not identical, and we would rather say so than pretend. Simulated one-day totals are spread a little more widely than real ones: more very low scores and more very high ones, with fewer in the 250-to-349 range where 53% of real first innings land. Simulated Tests produce more totals between 400 and 499 than the real game did, clearly in the post-war chart and to a lesser degree in the modern one. Part of that is who is playing: the Test simulations here are all-time XIs, the best each nation has produced, while the real archive is every Test side that ever took the field.

We measure these differences every time the games change, against the same archives you can browse on this site, and we treat them as work to do rather than as a margin to hide. When a change to a game moves one of these charts, the chart on this page moves with it.

CricArcade's simulations were built independently, from the real records, over a long time. We publish the principles above and the results, because players deserve to know what kind of game they are playing and whether its cricket is believable. We keep the algorithms and their parameters private, for two reasons.

The first is fairness. The games have leaderboards and head-to-head matches, and a published formula is an invitation to play the formula instead of the cricket. The second is that the games keep improving: the differences in the section above get worked on, and a published recipe would turn every improvement into a broken promise. What we promise instead is what this page shows: real players, real conditions, real uncertainty, and results you can check against the history of the game.

If you want to see the principles at work, the fastest way is to play: The Follow-On, The Chase or the T20 Franchise Manager. The guides go deeper into the real cricket behind each game.

• CricArcade Test match archive: 2,537 real scorecards, 1877 to March 2024
• CricArcade ODI match archive: real scorecards since 1971
• Charts computed with scripts/build-how-it-works.ts over the archives and the games' own simulations
• MCC: The Laws of Cricket
• ICC Men's Playing Conditions