Football News: Stats And The Lack Of Context
Image from: elegantthemes.com
Stats and the Lack of Context
Anyone who has spent a lot of time on our banter sites has probably noticed that I have a bit of a bee in my bonnet about the misuse of stats. Not so much the stats themselves, but the use of them without context to 'prove' a point or to draw a conclusion from them, is what bugs me. Annoys me even, at times.
One that particularly got to me recently was something I saw while watching the build up to the Merseyside derby. The presenter made the claim that Everton's Idrissa Gueye was better than Chelsea's N'Golo Kante because he has made more tackles. There was no attempt at all to put any context around that stat before attempting to take a conclusion from it. No consideration of other parts of the game even, when there are so many factors to consider, even if you limit it to the defensive aspect of the game. Does Kante make less tackles because Chelsea have more possession in a game than Everton tend to do? How about his positioning, does he get into better positions to prevent a pass, so negating the need for a challenge? Has his growing reputation made players less likely to hold onto the ball around him than they do around Gueye? Then there are interceptions, blocks, clearances and fouls made, which would all affect who is the better player. Incidentally the pair have both made 2.4 interceptions per game, despite Everton's lower possession stats, which would suggest, though only suggest, that Kante is positioning himself better to achieve the same amount in less time per game. All stats are as subjective as watching the game, as it really depends on how highly you rate each aspect of the game and what value you place on one stat compared to another. Does the fact that Kante gets dribbled past less often per game mean he is a better one-on-one defender or does it just mean less opponents have the opportunity to take him on? Or it could even be that players choose not to attempt to dribble past him because of who he is.
There are no stats to determine so many of those factors outlined, which is why stats should not be taken on their own to determine how you rate a player or team.
While I am on the subject, there are a few more annoying examples from the media, such as a presenter's claim that Fabian Delph had made a 'massive impact' when Manchester City played against Chelsea. Now they might have been right about that, but it was the stat they chose to 'prove' their point that he made a massive impact that astonished me and left me shaking my head in disbelief. They chose to highlight the number of touches he had in the first half as proof, as he had the most touches of the ball in that half of all players on the pitch. So not goals, assists, crosses, tackles or any of the things which actually make an impact, but touches of the ball. Talk about ridiculous. You could have a thousand touches and do nothing at all. It was just another presenter making a nonsense of stats.
Another example was a program analysing a Tottenham Hotspur performance, which they won in a dominant fashion. The analyst pulled up Dele Alli's heat map to show that he was not really a midfielder but a forward or, as he termed it, a nine and a half, because most of his touches were in the final third. Except that the heat map only illustrates which areas he is receiving and touching the ball, not where his starting position is. There was no analysis of why he was receiving the ball there or look into where he was the rest of the game. Was he just making lots of forward runs because his team were dominating and only getting the ball on the end of runs? It was not like he needed to do much chasing back in the game in question, as it was a particularly one-sided affair. This is not some local journalist making this asinine analysis, it was an extremely well-paid former player, who has many years of experience as a football analyst. He was a great player, but clearly not great at analysing stats.
That does not mean stats are a bad thing, quite the opposite, they are a very useful tool, if used correctly. Some are just interesting bits of trivia, but not of any real use. For instance the distance travelled stat I saw earlier this season, which showed West Bromwich Albion's Craig Dawson as the player who had covered the most distance in the Premier League up to that point of the season. On its own, that stat is absolutely meaningless, you need to add more information to get any use from it, but it is still an interesting and surprising stat. I would, personally, love to have more context to it, but also I would be interested to find out if there are any quirks to the gameplan utilised by Tony Pulis which affected it. Perhaps dead ball situations could have an effect? Most teams do not throw defenders forward for long throws, for instance, but I have seen the Baggies do that. Could that affect the distance covered? Certainly his aerial ability at corners and free kicks means he is likely to add a few yards further to his distance travelled at each one, over most full backs. Usually the fullbacks will hang back, Dawson tends to make his way into the box to try and get on the end of it.
Sometimes there are ones that seem to give enough info to be useful, such as this one sent in to our Liverpool site recently:
TEAM ShotsOnTarget Goals Conceded %
Chelsea 84 24 29
Spurs 87 22 25
Liverpool 97 39 40
Man City 86 34 40
Arsenal 122 36 30
Man Utd 82 24 29
Everton 115 34 30
West Brom 135 40 30
Saints 90 37 41
As this is shots on target only, it is tempting to judge goalkeepers based on this stat. Some of it makes sense, Hugo Lloris (Tottenham Hotspur), David De Gea (Manchester United) and Thibault Courtois (Chelsea) making a higher save percentage, Claudio Bravo/Willy Caballero (Manchester City), Fraser Forster (Southampton) and Simon Mignolet/Loris Karius (Liverpool) having a poor save percentage, in comparison. Then you spot Everton, with Joel Robles/Maarten Stekelenburg having the same percentage as Petr Cech in the Arsenal goal. Anyone who has seen Robles play will know he is not a good keeper and that casts doubt on the usefulness of this stat on its own.
In fact it brings to mind a study I read from a professor of statistics, who had studied stats across a number of sports, and his conclusion when it came to football. He concluded that there were too few games across a season, with too few goals scored to effectively analyse trends, as there was not enough data from it to rule out quirks. While he was referring to score predictions, the need for more data still applies. Even when you have all the data in the world at your fingertips, it may still come down top personal opinion. After all, even now with all the data to influence people's opinion, it still comes down to personal preference when you ask who is the best player in the world.
Written by Tris Burke April 23 2017 04:41:10