Spurs Banter Archive April 08 2015

08 Apr 2015 12:13:40
Disraeli is supposed to have said "there are three kinds of lies: lies, damned lies and statistics." Whenever the facts fail to corroborate our own arguments we like to remind ourselves of that quip.

Obviously we should be wary about over-interpreting football statistics.

For example, I always take pass completion rates with a pinch of salt. It is easy to look good on paper if you make a lot of short, safe, sideways passes, but one good, raking ball that opens up a defence, and results in a goal, will make up for a lot of similar passes that are just slightly off target and cede possession.

I also have doubts about 'assists'. This seems to mean that a player was the last person to touch the ball before the one who scored, but does this take account of whether or not the touch was decisive?

Last season, I saw a blog that compared Soldado with Torres. It concluded that Soldado was better at keeping possession. But that was because whenever Torres got the ball he was taking on defenders and trying to make things happen, while Soldado was content just to lay the ball off.

Statistics can never replace judgement, but they are sometimes a useful corrective. We all get bees in our bonnets about certain players, so it sometimes helps to take a reality check.

A few years ago, Peter Crouch was taking a lot of stick on this site. It was my impression that Spurs generally had a better shape, and played better, when he was on the pitch than when he wasn't, despite his obvious deficiencies as a player. I knew I was a lone voice at that time, so I did an analysis of our results when he started and when he didn't. I had no idea what it would show.

In fact it was quite revealing. At home, especially, we did noticeably better when he started games than when he didn't and he was far more likely to be brought on as a sub than be substituted himself.

I did something similar when it was Dawson that was in the firing line. Again, despite the obvious weaknesses in his game, we conceded fewer goals and picked more points when he played than when he didn't (home and away and in the league and cups).

These analyses didn't prove anything because the sample sizes were too small for that, but they did help to illustrate the gap that sometimes exists between our perception about what is happening on the pitch and what is really going on.

But that doesn't stop us preferring our prejudices.

PS: For what it is worth, I also think that Rose has grown steadily into a really good player and is the defender who has come out of this season with the most credit (except for Lloris, of course), but that is just my impression. I haven't checked it yet.

{Ed001's Note - pointless checking stats on that, as you are comparing him with Davies, just like you were only comparing Crouch with who replaced him or Dawson with his replacement. While I personally rated Crouch as a player, it doesn't necessarily mean he is good enough for the club just because he is better than the option currently there. The same with Dawson, being better than what else was available to the manager does not mean he is good enough for the club.

Stats need far more analysis than it is possible to do in one post. It takes months of analysis to be able to even get an idea of a player, let alone to then fully compare two players in the same position. That is why the likes of Comolli, with his over reliance on statistical analysis, is so hit and miss, his analysis is flawed and too shallow.}