The table ranking system in football is flawed

Whenever a football match is played, teams can expect one of three possible outcomes: a win, a loss or a draw (except in knockout stages). These outcomes award teams three points, nothing or one point respectively. One problem with this outcome is immediately obvious: luck is rewarded. But the real trouble is that effort is not even considered.

A team may play incredibly well, on the whole, for 90 min, being let down in-between by slips or individual mistakes. This happens week in and week out. A team might also play pathetically, constantly losing possession or unable to build an attack at all. This too happens week in and week out. And it may so happen that the latter team exploits an individual mistake on the part of the former team and slot the ball in their net. Unfortunately, this too happens more often than we would like.

The trouble with the current system

The end result of this—an outcome that colours the table of standings for the league or cup not just the game week but for the whole year—is that the latter team would be rewarded with points on a technical win, while the former team would go home empty-handed.

The importance of a more complex points system, and the elephant in the room that the current reductivist system does not address, is not only that the efforts of the losing team were not rewarded but—more important—that the unfair bit of luck that benefitted the winning team was not called out. The 3–1–0 points system is adamantly discrete in a world where everything is continuous.

One may then ask, “But is there not some luck in every match?” There is, and a more complex points system would address that. The current three-tier points system is of an ancient era when both time and points were kept track of manually. Today, balls flying an inch past the goal line or offsides by the width of a hair are flagged using high-speed cameras and beefy computers. The “complex” points system we have been talking about would in fact be pretty simple for these computers to handle.

How to build a fairer points system

A weight map is the first aspect that needs to be put in place for the new points system to be meaningful. There is no doubt that goals should continue to carry the most weight because that is the aim of the game after all. But the result itself should clearly be a closer win than a three-point gap would suggest. Moreover, doing this would take goal differences into consideration at the points stage itself, greatly reducing tables with equal points. Should they happen anyway, we have head-to-head results to compare as usual.

The next hefty factor should be team communication in attack and defence and in transition. Following that, with a negative impact, should be fouls and dives—this has regrettably become a core aspect of football and teams should be punished for indulging in it. Next, consistent individual performances should be rewarded. So should a team’s overall performance in excess of the expected goals (xG) metric, a relatively recent introduction present across all leagues and cups in football. Next up, ball recovery time and keeper efficiency.

This is just the start but a good start anyway. A lot of these metrics are either independently measured already or measured across a season rather than a match (see AWS for the Bundesliga, for instance). They, along with the few additional measures, need to be incorporated to form a more fluid points system.

Building an equation

Metric 1: Goals

The goals could carry an exponential weight just so we can be sure that more goals scored would absolutely guarantee a win. For example, a 3–1 result could mean an exponential 1.5 v 0.5 (or a 3 v 1 but that would be too much because of the exponential). That way, even if both teams played exactly as well and scored exactly the same on every metric, a 3–1 result would still declare a proper winner with, say, 5.2 and 1.7 points.

Several sub-metrics may be defined for goals: in addition to a standard base score per goal, points may be added for individual build-up play, one-touch and volley goals, acrobatic goals etc. A key sub-metric for goals scored should be deflections. Goals scored straight into the net should award the kicker more points on top of the base score than goals scored through deflections, especially deflections off opponent defenders.

Metrics 2–4: Team communication

Now what about team communication? For each attack this is best quantified as the weighted sum of the net passes, net dribbles, number of passes during the attack attempt, and shot distance, in that order of weight. These weights could simply be in arithmetic descension. Similarly, for defence, this can be the weighted sum of net recoveries, net clearances with posession, net interceptions, net standing or sliding tackles, net clearences to the sidelines, and net clearances to the back line, in that order of weight. The weights could once again be in arithmetic descension.

Metric 5: Individual performances

Individual performances are easier to monitor for being more universal: despite their set positions, everyone in football can potentially tackle, dribble, shoot, throw, pass, lob, cross, head, volley, lose or gain possession, intercept, and of course commit fouls. These ought to be weighed up or down as sensible and summed for each player. This can then be summed for the entire team to arrive at our fifth parameter.

Metric 6: Overall performance v. xG

The expected goals metric is an esoteric number in regular use not only in football matches but also in betting set-ups. Simply put, a larger xG value means you are more likely to score a goal. Going further, a rounded xG value can also be interpreted as the literal number of goals your team is expected to score in a match. But xG is itself an incredibly complex metric and like all statistics is best taken with a pinch of salt. However, it can be reasonably thought of as an assessment of the performance of a team if performances strictly arise out of goals scored. Until we have a better metric for this parameter, let us go with exceeding xG as an indicator of good performance, but, since it is just one among many other parameters in this scoring method, it will no longer imply that goals alone determine good performance.

Metric 7: Ball recovery time

Although we have talked specifically about attacking and defensive football in metrics 2–4, those pictures alone would be incomplete if we do not account for a general picture of these means of playing. In other words, if the attacking and defensive parameters discussed above specify parts of the game, ball recovery time ties them all up with a nice ribbon.

Ball recovery time is essentially a measure of how quickly a team shifts from defensive to attacking football. This naturally gives rise to certain trends: first, shorter ball recovery time means greater recurrences of attacking football. I say “greater recurrences” because it does not mean greater periods spent playing attacking football: we do not know that the attacking football is not so bad that they keep losing the ball and therefore having to recover it. So we do not reward occurrences of ball recovery, just the swiftness of it.

But how does ball recovery time tie up metrics 2–4? If you play more proactively in defensive, only opening up to seize counterattacking opportunities, you get rewarded because this is a legitimate style of play e.g. gegenpressing. If you play a constant but monotonous, uncreative attacking football reliant on (sometimes meaningless) possession, you still get rewarded because you know how to pass the ball around and keep the opponents at bay e.g. tiki-taka. So neither attacking nor defensive styles alone should award points or penalise teams. How quickly they can get back to their own mode of play—how effectively they can counteract or how quickly they can regain the ball to continue possessing it—must also be accounted for if we want to tell a more complete picture.

Metric 8: Keeper efficiency

This last one is our only goalie-centric parameter. It does not imply the ineffectiveness or unimportance of a keeper, but it does speak to the difficulty in justifiably parametrising a keeper’s intentional contribution. However, in my view, there is another metric that is really also a measure of goalkeeping: the number of goals scored by the opponent, which we placed as our all-important first metric.

Keepers play less on their feet than their outfield teammates, but when they do their contribution is exponentially more critical. A keeper’s split-second response could mean more to turn a game on its head than a winger’s shuffling around with the ball on the sideline. Unlike the other metrics, where some additional work may be needed statistically to realise this new points system, we already have all we need for the keeper: efficiency.

AWS for Bundesliga defines keeper efficiency as a complex variable based on the algebraic sum of xSaves (the goal-keeping equivalent of xGoals). Their precise description is as follows:

Using derived characteristics of a shot, the model generates the probability that the shot will be successfully saved by the goalkeeper. Some of the factors considered by the model are: distance to goal, distance to goalkeeper, shot angle, number of players between the shot location and the goal, goalkeeper positioning, and predicted shot trajectory…

Adding up all xSaves values of saved and conceded shots by a goalkeeper yields the expected number of saves a goalkeeper should have during a match or season. Comparing that against the actual number of saves yields the Keeper Efficiency.

In other words, AWS uses a bunch of on-field positioning and projectile motion data to determine the likelihood of a keeper stopping a shot for every shot taken by the opposite team. They end up with numbers for the likelihood of shots stopped versus let in and the algebraic sum of these is the efficiency of a keeper. Like all efficiency, this value should ideally be as large as possible. It can be suitably weighed in relation with the other metrics to contribute to the overall performance score.

The trouble with using this metric alone is rather serious: unlike our other metrics which are based on a player’s actual performance, this one is purely statistical. Any influence a keeper can have on this is long-term, by performing consistently well across a number of matches.

One way to get around this is by introducing a few weighted pointers that are directly influenced by and relevent to just the one game. For instance, we could have the keeper’s positional awareness and reaction time accounted for i.e. how close in the direction of the ball did the keeper dive (recall those near misses where the ball grazes a keeper’s hand), or how much of a touch did the keeper get (grabbing a ball successfully versus merely punching it over the backline, the former being clearly preferable).

These actions, and other like them, can well make a meaningful reward system centred around the keeper and should be considered in addition to long-term keeper efficiency metrics to provide a suitably weighted match-wise keeper efficiency rating.

These metrics can seem complex on the face of it and they probably are. But we have machines that can easily handle these computations now. Implementing this sort of a points system not only rewards victory but also good behaviour en route to it. It retains the footballing trichotomy of points derived from win–lose–draw scenarios based on the singular metric of goal scoring; but it does this without underrating the skills, spirit and drive players needed to have demonstrated to have achieved such a result.

With a bit less black-and-white and a lot more grey, we may start to see, beautifully quantified, that winning and losing teams are not separated by as much as we thought they were. The repercussions can be deep and wide—even uncomfortable at first—but if sportsmanship can be rewarded and diving can be punished, we already have a welcome solution on our hands.