ICC rankings explained: three systems, not one table

A side loses a Test series two matches to one and comes out of it rated higher than it went in. Nothing has been miscounted. The table that produced that result is not counting series wins, and it never was.

Most attempts at getting the ICC rankings explained treat the subject as one object: a league table, published by the governing body, best team at the top. There is no one object. There are three separate systems sharing the name, built at different times to answer different questions, and almost every argument about cricket rankings is really an argument between two people looking at two of them.

The team rankings are a rating. The player rankings are a different kind of rating. The championship tables are a competition. Only the last of the three is a league table, and it is the one people least often mean.

ICC rankings explained: three tables, three questions

Start by separating them, because the rest of this is easier once they are apart.

The team rankings answer a question about the recent past: how well has this side been playing, per match, over the last three or four years, adjusted for the quality of the opposition it faced. There is one for each format and one for each of the men's and women's games. The output is a rating, typically a two or three digit number, and it is an average rather than an accumulation.

The player rankings answer a question about the present: if you were picking a world eleven this afternoon, who would be in it. The output is a figure on a fixed 0 to 1000 scale that moves after every innings a player plays, upwards when the innings beats what the system already expects of them and downwards when it falls short. It is a moving average of performances, not a summary of a career.

The championship tables answer a question about a defined competition: who has actually won what inside this cycle. The World Test Championship standings are the clearest example, and they behave like an ordinary tournament table with points, a percentage and a final at the end of it. They are governed by their own playing conditions, they reset when the cycle does, and they have nothing to do with the rating method at all. If you want that system in detail, the percentage table and the over-rate penalties that decide it are a separate subject with separate arithmetic.

The confusion is not the public's fault. All three appear under the same governing body's name, two of them cover Test cricket, and headlines refer to all of them as "the rankings". A side can be second in the Test rating and fifth in the championship table in the same week without either being wrong, because one is asking how good you have been and the other is asking what you have won.

• 1000Top of the player rating scale
• 1Points a Test win is worth on the team table
• 50Rating swing either side of an opponent's rating
• 40Rating gap that switches the team formula

These are the structural numbers in the ICC's published method, not measured values. The gap threshold and the 50-point swing are defined by the formula itself.

The team rating is an average, which is why it behaves so oddly

Here is the sentence that unlocks the team table, and it is the sentence nearly every summary leaves out.

A team's rating is its total ratings points divided by its total matches. Not its points. Its points per match.

The ICC's own analogy is a batting average, and it is a good one. A batting average is runs divided by dismissals, which is why a batsman who scores fifty in an innings can see his average fall. Fifty is a decent score and it is still below a career average of sixty, so the average drops. The team rating works identically. Every series a side plays produces a rate of points per match, and that rate either pulls the existing average up or drags it down depending on whether it beats the number the side is already carrying.

This single design choice explains almost every behaviour people find counterintuitive. A team does not accumulate rating points the way a football club accumulates league points. It cannot climb by playing more. It can only climb by playing better than it has been playing.

Two consequences follow immediately.

The first is that a strong side has more to lose than a weak side has to gain from the same fixture. If your rating is high, the bar your next series has to clear is high, and clearing it produces a tiny gain while missing it produces a real fall. If your rating is low, almost any competitive series raises it.

The second is that the denominator matters as much as the numerator. A side that has played thirty-five Tests inside the counting window has a heavy, stable average that a single series barely moves. A side that has played nine has a light, jumpy one. Two teams can produce identical performances in the same month and see their ratings move by wildly different amounts, purely because of how much cricket sits underneath the number. That is a scheduling artefact rather than a cricketing one, and it is the first of several places where the rating quietly reports on the fixture list rather than on the team.

Opponent strength is not a modifier, it is the whole mechanism

Most people assume there is a base value for a win with a bonus applied for beating a good side. That is not how it works. The opponent's rating is the value of the result. There is no base.

For a match between two sides whose ratings are within 40 points of each other, the rule is short enough to state in one line. The winner scores the loser's rating plus 50. The loser scores the winner's rating minus 50. A tie or a draw scores each side the other's rating exactly.

Read that again and notice what is missing. Your own rating appears nowhere in what you earn. Only the opponent's does. Beat a side rated 130 and you bank 180 points for that match whether you are rated 40 or 120. What differs is whether 180 is above or below the average you already hold, and therefore whether banking it lifts you or not.

That is why the gain scales with the gap. It is not a bonus scheme. It is subtraction.

• Points scored for the win
• Above own rating of 100

Opponent rated 7012020

Opponent rated 10015050

Opponent rated 13018080

Constructed figures. The winning side is rated 100 in this example and every opponent is within 40 rating points of it, so the standard rule applies. Points scored is the opponent's rating plus 50. The second series is how far that sits above the winning side's own rating of 100, which is what actually moves the average.

Show the numbers

Worked example: one win, three different opponents

Item	Points scored for the win	Above own rating of 100
Opponent rated 70	120	20
Opponent rated 100	150	50
Opponent rated 130	180	80

The three pairs of bars are the entire competitive logic of the table. Beating a side well below you still helps, but only a little, and it costs you a fixture slot you could have spent on someone better. Beating a side well above you moves the number four times as far. A board that wanted to game the rating would not schedule easy series. It would schedule hard ones and try to win one Test in three.

There is a second regime for when the two sides are 40 or more rating points apart, and it exists to stop the first one producing absurdities. Once a gap that large opens up, the points switch from being derived from the opponent's rating to being derived from your own. The stronger side scores its own rating plus 10 for a win, its own rating minus 40 for a draw or tie, and its own rating minus 90 for a defeat. The weaker side scores its own rating plus 90 for a win, its own rating plus 40 for a draw, and its own rating minus 10 for a loss.

Look at what that does. A heavy favourite that wins gains almost nothing, because plus 10 barely clears its own average. A heavy favourite that loses takes a 90-point hit against its own number, which is enormous. And the underdog is protected: losing costs it 10 and beating the favourite is worth 90. The mismatch regime is not a fairness gesture. It is a stability mechanism, because without it a mismatch would generate huge point transfers in both directions and the top of the table would thrash about every time a strong side toured a weak one.

How a series becomes a change in a rating

Test cricket is scored by series rather than purely by match, which adds one more piece.

Each Test is worth one point to the winner and half a point to each side in a draw or tie. On top of that, the series carries a bonus point, awarded to the side that wins it, or split half each if the series is drawn. So a three-Test series has four points in it: three from the matches and one from the series. A five-Test series has six. A two-Test series has three.

Those points are then converted at the rate the gap rule sets, added to the side's running totals, and the rating is recalculated as the new total divided by the new number of matches, where the series bonus counts in the denominator as well as the numerator.

1. 1Before the seriesSide A is rated 100, carrying 3,600 points from 36 counted matches. Side B is rated 130. The two are 30 rating points apart.
2. 2The gap decides the formula30 is under the 40-point threshold, so the standard rule applies. The winner of each match scores the loser's rating plus 50, and the loser scores the winner's rating minus 50.
3. 3The series is scoredThree Tests plus a series bonus makes four points available. Side B wins the series 2-1, so it takes two match points and the bonus. Side A takes one match point.
4. 4Each point is convertedSide A's single point is worth Side B's rating of 130 plus 50, which is 180. The three points Side B took are worth 130 minus 50, or 80 each, to Side A.
5. 5Side A's series rate180 plus 240 is 420 ratings points from a series total of four, a rate of 105 per match.
6. 6The rate goes into the average4,020 points divided by 40 matches gives 100.5. Side A rounds to 101, having lost the series and climbed.
7. 7Meanwhile Side BSide B scores three points at 150 each and one at 50, which is 500 from four, a rate of 125. That is below its rating of 130, so the series winner's average falls.

Every figure here is invented to make the arithmetic legible. Side A is rated 100 with 3,600 accumulated points from 36 counted matches. Side B is rated 130. No real team, series or rating is being described.

Nothing in that sequence is a quirk. Side A performed, per match, at a level above the standard it had been holding, against a stronger opponent, and the average recorded it. Side B performed below its own standard against a weaker opponent, and the average recorded that too. The series scoreline is not what the system is measuring. It is measuring the rate.

If that feels like the table refusing to acknowledge results, consider the alternative. A system that only counted series wins would rank a side that won six home series against modest opposition above one that drew away in the two hardest places to tour. Every serious rating system in any sport eventually arrives at the same conclusion, which is that who you played is not context for the result, it is part of the result.

Why a side can lose a series and still rise

The worked example above is the clean case, but the general rule is easy enough to carry around.

A side's rating goes up when its rate for the series beats the rating it walked in with. Against a much stronger opponent, the rate for a respectable defeat can beat it comfortably. Against a much weaker opponent, even a clean sweep can fall short.

Take the second half of that. A side rated 120 tours one rated 85, a 35-point gap, so the standard rule still applies. It wins all three Tests and the bonus, four points from four, every one of them worth 85 plus 50, which is 135. Its rate for the series is 135, comfortably above 120, so it climbs. Now suppose it wins the series 2-1 instead. Its rate becomes three points at 135 and one at 35, which is 440 over four, or 110. It has won a series away from home and its rating has fallen ten points below where it was. Both figures here are invented, but the arithmetic is exactly what the published rule produces.

This is the honest bargain the rating makes. It will tell you which side has been the better team over three years, weighted for who they played. It will not tell you who won last week, and it was never designed to. If you want the second thing, the scoreboard is right there and it is free.

Old results are discounted, not deleted

The counting window runs roughly three to four years, and the way results leave it is the second thing that separates this system from a simple table.

Results do not sit at full value and then vanish. They step down. After the annual update, which the ICC applies around the start of May, the most recent block of results carries full weight, the block before it carries half weight, and anything older than that is dropped out of the calculation entirely. Both the points and the matches are halved together, so a series that has aged into the discounted band still counts in the average, just at half the influence it had a year earlier.

Most recent block of results100%

The block before that50%

Anything older0%

The weighting bands the ICC applies at its annual update. A discounted result has both its points and its matches halved, so it still counts towards the average at reduced influence rather than being partially removed from it.

Show the numbers

How a result's weight falls as it ages

Item	Value
Most recent block of results	100%
The block before that	50%
Anything older	0%

Halving both halves of the fraction matters more than it sounds. If only the points were halved, an ageing result would drag every side's rating towards zero. Because the matches are halved as well, the result keeps its character and loses only its say. A stale thrashing still says you were thrashed. It just says it more quietly.

The annual timing produces one genuinely strange effect that anyone following the table closely will have noticed. Ratings move in early May without a ball being bowled. A side whose oldest counted year was excellent loses a chunk of quality from its average and falls; a side whose oldest counted year was miserable loses a chunk of misery and rises. Nothing has happened on a cricket field. The window has simply slid, and the tables are recalculated on the new set of weights.

There is an argument for a gradual decay curve instead, with every result fading a little each month rather than dropping a band once a year. It would be smoother and it would track form better. It would also be impossible for an ordinary supporter to reproduce, and the annual step at least means that anyone can work out, in advance and to the day, what is about to leave the calculation.

Each format is a separate ledger, and the numbers do not translate

There is a rating table for Test cricket, one for one-day internationals and one for T20 internationals, in both the men's and women's games, and they share a scale without sharing a meaning.

They are computed from disjoint sets of matches. A Test result has no effect whatsoever on a T20I rating. They also have structurally different inputs: Tests carry a series bonus because Tests are played in series, while the shorter formats are largely scored match by match, and the pools of teams differ enormously in size. That last point matters more than the first. A Test table contains a handful of sides who play each other constantly. A T20I table contains dozens of teams, many of whom will never meet.

So the observation that a side is rated higher in T20 cricket than in Tests carries no information about which format suits it. The two numbers are averages over different populations, computed from different fixture densities, against different competition. Comparing them is like comparing a batting average to an economy rate because both are decimal numbers.

This is worth being blunt about because it is a mistake even good commentary makes. There is no combined ICC ranking, no aggregate of the three formats, and no defensible way to build one. Sides are not equally exposed to all three, and the structural differences between the formats mean a run or a wicket does not mean the same thing across them anyway.

The player rating is a moving figure, not a career summary

Switch tables now, because the player system works on a different principle and gets confused with the team one constantly.

Every internationally active batter and bowler carries a rating on a 0 to 1000 scale for each format. That figure is not a total, not an average of runs, and not a percentile. It is a weighted average of that player's individual performances, recalculated after each innings, with recent innings weighted most heavily and older ones fading in influence.

The scale is worth understanding because it is not intuitive. 1000 is a ceiling, not a target, and it has never been reached. The ICC's own guidance is that a rating above 900 is a rare and exceptional level, that the mid 700s and above is broadly where a format's top ten sit, and that around 500 is a solid, established international standard. So the difference between a rating of 850 and 800 is not a six per cent difference in quality. It is a large difference near the top of a compressed scale.

The mechanism is a comparison against expectation. Each innings produces a performance value, and that value is compared with what the system already expects of that player given their rating. Beat the expectation and the rating rises. Fall short and it falls. A batter rated 850 who scores forty on a difficult pitch may gain nothing at all, because forty is not what 850 predicts, while a batter rated 450 who scores the same forty in the same match may gain a good deal.

That is the piece that makes the table behave differently from anything derived from raw statistics. It is not measuring output. It is measuring surprise.

What actually feeds a batting or a bowling rating

The precise algorithm is not published in full, which is a real criticism and one this piece comes back to. What the ICC does state clearly is the set of factors the calculation uses, and knowing them is enough to explain most of the movements that look strange.

For a batter, the value of an innings depends on the runs scored, the quality of the bowling attack faced as measured by the bowlers' own ratings, the pitch and the scoring conditions read through what everybody else in the match managed, and the result of the match. A hundred is not a hundred. A hundred scored while the rest of both sides struggle past thirty is a far larger performance than a hundred in a match where four other players also made one, and the system sees that because it is looking at the innings in the context of the match around it.

For a bowler, the inputs mirror it. Wickets taken, runs conceded, the quality of the batters dismissed as measured by their ratings, the conditions read from the match as a whole, and the result. Removing three top-order batters who are themselves highly rated is worth substantially more than removing three tail-enders for the same figures, which is exactly the distinction a bowling average refuses to make.

Two edge cases in the rules answer questions people ask constantly. A batter who does not get to bat is not penalised; their rating simply does not change. And a bowler who goes wicketless is protected in low-scoring innings, where the total is small enough that there were never many wickets to be had. Both provisions exist to stop the system punishing players for match situations they had no part in creating.

The ICC does not rate wicketkeepers as keepers at all, on the reasonable ground that catches and stumpings depend so heavily on the bowling in front of them that the numbers do not carry a usable signal. Keepers appear in the batting table on their batting alone. All-rounders get an index rather than a rating of their own: batting rating multiplied by bowling rating, divided by 1000. That construction is deliberately punishing, because a player weak in one discipline cannot compensate with brilliance in the other. Multiply 800 by 100 and divide by 1000 and you get 80. The all-rounder index is a test of whether both halves are real.

Why a new player's rating is held down

A player who scores a double hundred on debut has, statistically, an unbeatable record. Every rating system has to solve this, and the ICC's answer is a damping factor applied at the start of a career.

For roughly the first twenty Tests, a player's published rating is deliberately held below the figure their performances alone would produce, with the damping easing as matches accumulate. The equivalent applies in the shorter formats over their own match counts. A newcomer who begins brilliantly climbs fast but starts from an artificially suppressed base, and only converges on their true rating once there is enough evidence behind it.

This is unpopular with supporters of whichever brilliant young player is currently being held down, and it is obviously correct. Without it the top of the table would be reshuffled by every debutant hundred and every five-wicket haul in a first Test, and the ranking would stop functioning as a statement about the best players in the world. The cost is real and worth naming: a genuinely great player is under-rated for two years or so, and the table lags their arrival. The system is trading responsiveness for credibility, which is the same trade the damping in any rating system makes.

The other end of a career is handled by qualification periods rather than damping. A player must have appeared in a match within a defined recent period to stay on the published table, longer for Tests than for the shorter formats because Tests are played less often. Retire, or simply stop being picked for long enough, and you drop off the list. The rating itself is retained rather than reset, so a player who returns after a long absence re-enters where they left rather than starting from nothing, and then finds out quickly whether the figure still fits.

When the rankings and the averages disagree violently

They disagree constantly, and the disagreement is informative rather than embarrassing.

A batting average is cumulative, permanent and context-free. Every run counts identically whether it was scored last week or eleven years ago, against a great attack or a poor one, on a green seamer or a featherbed, in a dead rubber or a decider. That is a feature, not a flaw: an average is a record of what a player has done across a career, and it is not trying to be anything else. The same applies to the way a bowler's average, economy and strike rate divide up the same three numbers.

A rating is the opposite on every axis. Recent, weighted, adjusted for opposition, adjusted for conditions, sensitive to the match result. It is trying to describe a player now.

So the two diverge most sharply in three situations, all of them predictable. A player whose great years were six seasons ago holds a strong average and a middling rating, because the average remembers and the rating has forgotten. A player two years into a breakthrough holds a modest average, dragged down by an early struggle, and a high rating, because the rating has already discounted the struggle. And a player who has feasted on weak attacks in benign conditions holds a handsome average and a rating that quietly declines to be impressed.

Which is more useful depends entirely on the question. For selection, form and who is the best in the world right now, the rating is better, because it is built to answer exactly that and the average is not. For career assessment, historical comparison and any argument about greatness, the average and the raw record are better, because a rating is a snapshot of a moving figure and has no memory to speak of. Neither is a substitute for reading the record properly, and both are more useful once you understand what an average is actually dividing by, and what a strike rate is not telling you.

The one thing you should not do is treat a rating as a precise ordering. The difference between the third and fourth ranked batter in a format is frequently smaller than the noise in a single innings. The tables are read as leaderboards because they are presented as leaderboards, but the honest reading is in bands: this group is clearly the best, this group is behind them, and the ordering inside each group changes with the weather.

ICC rankings explained: what the numbers are actually used for

If this were only a talking point, none of the design decisions would matter much. It is not.

Qualification. Global tournaments have repeatedly used a ranking cut-off as a direct qualification route, with a date published years in advance and the teams above a line going straight through while everyone below plays a qualifying event. That converts a rating into a fixture list, a broadcast deal and a budget. Teams have been known to schedule series with an eye on the date, which is a rational response to a rule that pays for position. The mechanics of a cut-off qualification are clearest in the way the shortest global tournament assembles its field.

Seeding. Rankings are used to seed groups at world events, which decides who a side meets in a group stage and therefore how hard the path is. A seeding is worth more to a team's chances than most single-match advantages, and it is awarded by a number calculated months earlier.

Prize money and silverware. The ICC has attached a trophy and cash to the top of a ranking table on a cut-off date. A rating that moves half a point in April can therefore be worth a very large sum, and it can move because of a weighting update rather than because of any cricket.

Selection and contracts. Player rankings feed into award shortlists, into national contract discussions, and increasingly into franchise valuations, where a rating is a convenient single number for a recruitment department to argue with. The economics of a franchise auction run on a lot of things, and a defensible objective ranking is one of them.

None of this makes the rankings more accurate. It makes them consequential, which is why the criticisms below are worth taking seriously rather than treating as pedantry.

The three honest criticisms

The formula is only half published. The team rating method is fully specified and reproducible. Anyone with the published ratings, a series scoreline and a calculator can derive the new numbers, which is why the worked examples in this piece are possible at all. The player rating is not. The ICC describes the inputs and states that the process is fully automated with no human judgement in it, but the algorithm that turns an innings into a performance value is not in the public domain. That means no independent party can audit a player rating, reproduce a movement, or check a disputed calculation. No human intervention is a claim about process, not about correctness, and a system with real financial consequences that cannot be checked from outside is carrying a weakness it does not need to carry.

The fixture list distorts the team table more than the cricket does. This is the deepest problem and it is arithmetic rather than opinion. Because the rating is an average, its stability depends on how many matches sit underneath it, and cricket's schedules are wildly unequal. Worse, the series bonus is a fixed one point regardless of the length of the series, so it is a much larger share of a short series than a long one. Win a two-Test series one-nil with a draw and you take 2.5 of the three available points, or 83 per cent. Win a five-Test series one-nil with four draws and you take four of six, or 67 per cent. Identical superiority, materially different reward, decided entirely by how many Tests a board could sell. Sides that get five-match series against the strongest opponents and sides that get two-match series against whoever will tour are not being measured on the same instrument, and the rating cannot correct for something it cannot see.

The associate nations problem has no solution inside the system. Teams without regular Test cricket cannot hold a Test rating at all, because the eligibility rule requires a minimum volume of matches inside the window and there is no route to that volume without invitations from boards who have no obligation to issue them. In the T20I table, where associates do appear, the opponent-strength mechanism creates a closed loop. Beating another associate scores you their rating plus 50, and their rating is low because they mostly play each other, so the whole group is anchored to a low common baseline that no amount of winning within it can lift very far. Real movement requires fixtures against full members. Those fixtures are allocated partly on the basis of ranking. A ranking gates the games that would improve the ranking, which is a structural trap rather than a competitive one, and it is a large part of why the gap between the full members and everybody else closes so slowly.

There is a fourth criticism made often that is less persuasive. People complain that the rankings do not match the World Test Championship table, or the recent results, or their sense of who is playing well. That is not a fault. It is the three systems doing their three separate jobs, and the fix is to read the right table for the question rather than to change the method. The same instinct in tennis produces endless arguments about a list that is a rolling twelve-month receipt rather than an opinion about quality, and it is the same category error each time.

How to read a ranking table without being fooled by it

Four checks, all of them public, and between them they explain nearly every movement you will see.

Count the matches in the denominator. Before believing a rating, find out how much cricket it is averaging. A rating built on nine matches and one built on thirty-four are different kinds of statement, and the first will swing three times as far on the same series. This is the single most useful thing to know about any side outside the top three, and it is the least often reported.

Check the gap before the series, not after it. Whether the two sides are inside or outside 40 rating points determines which formula runs, and therefore whether the underdog is protected or exposed. Under 40, the favourite is genuinely at risk and a defeat is expensive. At 40 or over, the favourite gains almost nothing by winning and takes a heavy hit for losing, which is why the mismatch tours that look pointless are, for the higher rated side, all downside.

Know what is about to drop out in May. The annual reweighting is arithmetic anyone can do in advance. Look at which year is about to move from full weight to half, and at which is about to disappear, and you can predict the direction of a side's rating before a ball is bowled. A team about to lose a terrible year will rise. A team about to lose a great one will fall.

Ask which of the three tables you are looking at. If the number is out of 1000, it is a player rating and it is a statement about now. If it is a two or three digit team number, it is a three-year weighted average and it is a statement about a period. If it is a percentage with a final at the end of it, it is a championship and it is a statement about a competition. Nothing else follows from any of them, and most of the arguments about where a side really stands in the international game turn out to be two people answering different questions correctly.