To the moaning statto minnies

http://www.bbc.co.uk/blogs/timvickery/2009/01/analysis_over_numbers_goals_be.html

Analysis over numbers, goals before statistics

• Tim Vickery
  
• 12 Jan 09, 11:24 AM
  

Great blog on the BBC website this morning, drawing attention to the limitations and misleading nature of football statistics. Here's an extract:

"There have been many times in the Maracana stadium when I have been sitting next to the team collecting match statistics. "Accurate pass by the number 5," the team leader would call out, though the ball had been blasted calf-height on the recipient's wrong foot, keeping the play so tight that loss of possession was inevitable, or "inaccurate pass by the number 8," after he had played an inspired ball inside the opposing full -back which might have set up a chance if his team-mate had been bright enough to read it.

Witnessing the match stats being compiled has made me acutely aware of their limitations. Football is too fluid for the rigidity of the statistical mind. Has the ball been used well? This depends, surely, on the situation of the game, the zone of the pitch - on considerations that cannot be reduced to a statistic...

Football is never just about what you do. It's also about how you do it - and that is a difficult concept to stick a number on."

Worth bearing in mind when the match stats screen shows you have 20 shots on target, they have 2 and you lose 0-1 etc etc.

Exactly. The world does not go according to plan, there will always be freak results etc.

Er, that's missing the point again. I love the irony of your username.

I always remember Harry Redknapp saying that the only time he ever used pro-zone was to see if his son was really any good at passing the ball. He judges how well the players look when in training and during the match rather than looking at the numbers.

I understand the point of the post and whilst it's very true in the real world, the FM world runs on stats and percentages and the numbers are far more related to general performance than real life. The match engine isn't in depth enough to cover the variety of possibilities such as the examples in the OP.

Regarding shots on target, it's true that in real life a team could have 20 shots on target and be playing badly and none of the 20 shots could be from good opportunities. But in FM, even watching a match on key highlights will show the chances as being a lot more clear cut and having 20 shots on goal will reflect you dominating a game.

On a slight side note, I actually think SI have shot themselves in the foot by adding in the "Clear Cut Chances" because now people are focusing more on good chances that are missed rather than previous thoughts of shots on goal/target being any number of random chances.

"I understand the point of the post and whilst it's very true in the real world, the FM world runs on stats and percentages and the numbers are far more related to general performance than real life. The match engine isn't in depth enough to cover the variety of possibilities such as the examples in the OP."

I'm not so sure CP. Take an example: for my midfielders I check how many key passes they make. Now the likes of Fabregas could make loads of superb passes, but if my striker has poor anticipation and off-the-ball ratings, they won't be recorded, and Fabragas will be the one looking mediocre in the post-match stats. I might make a misjudgement and buy a new midfielder when I should be replacing the forward (well probably not if I really had Fabby, but you get the drift).

So I still maintain to those who whinge about the stats, watch the game - watch what your players do.

It is the classic argument dividing positivist and critical thinkers. Whilst I am all for FM employing the statistical analyses of Charles Reep et al as grounding for the ME, reducing football to pure stats misses a mass of complexity. Reep, for example, famously stated that goals come from a very few amount of passes, suggesting the more passes in a move the less chance of scoring. His work was used to inform the horrendous direct ball game in the UK of the Eighties. Graham Taylor was apparently a fan.

What he failed to critically analyse was the ability of the top sides. Many sides will not have enough technically good players in the squad to play competent possession football, thus the direct ball game will produce more goals as it inevitably lends itself to less passing mistakes. However, with a technically superior team there is no need to try this as you can rely on the players to break down the opposition by the sheer amount of possession they generate. Thus, if you are managing a technically good side, you should play technically good football, thus sacrificing the extra potential of direct ball goals for the opportunity to craft a greater number of chances whilst denying the opposition the ball and the chance to play effective, direct ball play. The statistical approach fails to take that into consideration. Reep's conclusions were that a team should always play direct football.

One of the major failings of top players becoming managers is that they expect their teams to be able to play football in the manner that they were accustomed in their prime. Bryan Robson is a perfect example. He refuses to sacrifice his vision of how the game should be played despite not having the quality of player, in relation to the level, to do so. A team that can play attractive football at Championship level still needs to play ugly at Premiership level. Many managers, especially those who were once good players, cannot grasp that simple fact of life. Once they build a squad capable of playing pretty football at the top level, then their ideas are likely to work. Until then, their vision is going to undermine their results. Likewise, the ugly football manager (think Sam Allerdyce) needs to be able to adapt his style to the quality of players he starts to pick up as his league performances get better. Failing to do so will see performance hit a plateau beyond which he cannot climb.

As always, it is important to strike a balance between stats and a critical appreciation of what exactly is going on. Allerdyce perhaps leans too far one way, Redknapp another, which is why they have never quite made the step into the big time. The best managers are always challenging conventional thinking and moving one step further than the already known. To do that they will take advantage of all elements of football, ranging from statistical performance to intuition and gut feeling. Ferguson famously 'knew' something was happening when Man Utd equalised against Bayern and left his team in a 4-3-3 formation for two more minutes, and we know what happened next. We love sport because of its unpredictability, yet we strive for predictable behaviour in our lives and society. These two tendencies collide in FM which is why people suffer so much angst. Its also what makes it such an addictive game, and an exceptional managerial, in all senses of the word, simulation.

I rarely trust the stats in FM anyway. Too often they don't seem to correlate with what I've seen in the match. I prefer to trust my own eyes and my own judgement.

It is the classic argument dividing positivist and critical thinkers.

You know it's gonna be good when it starts with this!

Tim Vickery is brilliant. He's a regular commentator on World Soccer Daily podcasts, and is incredibly encyclopedic in his range of knowledge. They call him "The Vickopedia" on the show in fact.

He makes a good point too. It constantly irritates me how even the commentators misjudge a play, and I think that in the end this is part of the ultimate beauty of the game. No one knows if Player X 'meant it', or if Player Y 'should have been there', so we constantly have to insert ourselves into their minds, 'as if' we knew. Then there's FM.

As someone else said - sometimes the stats that are shown on FM are blatantly WRONG.

eg In my last friendly last night.

@ 20mins - my starting MR gets injured - so I substitute him.

@ 23mins - the sub picks the ball up just inside my half, runs to the byline, cuts inside to the 6 yard box and dinks a lovely cross over onto the foot of my AMC for a tap in goal.

@ 30 mins, I go to check the players stats. I look specifically at the sub (cos he is a young talent and I want to see how he has done apart from this wonder dribble) and I look at his no of dribbles in the game - TOTAL ZERO !!! Wtf? TOTAL ASSISTS - ZERO !!! Wtf? He ends the game at 90 mins with a 5.4 rating - even though on extended highlights I could see that he had caused the opposition trouble several times.

Another example - I looked one time at the opposition's stats to see who was pulling the strings etc, since they had just hit me for 2 goals in 3 mins from their FC. This was after about 60 mins, and I looked at him - and he had recorded 3 passes the whole game - none successfully! Now according to every known wisdom on these forums, he should have either been subbed before that point, or his setting altered to get him far more involved in the game. Now of course the AI may just have done that (changed his settings) - but you can see how stats can lie!!

I find it difficult to be "in the moment" in FM, I see numbers and pixels not players and performance.

I think the op has a very good point and one I've always considered myself in FM. There are many times I look at the player ratings after a game and disagree totally. For example on of my midfielders may finish with a ver average pass completion rate but he so nearly set up numerous really good chances, or he had a couple of really good long range efforts that either hit the post or just missed, yet they simply go down as an innacurate pass or a shot off target and ends up with a decidedly average 6.1 rating when he should be aplauded for his creativity. Personally I like to rate my players with my own mental ratings, but unfortunately the game has to relate morale and player happiness on the ratings given by the computer.

I think the op has a very good point and one I've always considered myself in FM. There are many times I look at the player ratings after a game and disagree totally. For example on of my midfielders may finish with a ver average pass completion rate but he so nearly set up numerous really good chances, or he had a couple of really good long range efforts that either hit the post or just missed, yet they simply go down as an innacurate pass or a shot off target and ends up with a decidedly average 6.1 rating when he should be aplauded for his creativity. Personally I like to rate my players with my own mental ratings, but unfortunately the game has to relate morale and player happiness on the ratings given by the computer.

The problem with what you are saying there is that how many of us watch every match in full? I would suspect that 95-99% of us watch most (if not all) of our matches on either extended or key highlights - therefore we see perhaps 5-10 minutes of actual action during the game - thus we may well be missing the bits that either bring the individual rating up or down to it's actual rating from what we see it to be from the highlights.

eg From your example - you see (in the highlights) that he makes 4-5 great passes / shots that don't end in goals - just near misses - so you (logically) think that he has done well. Now in the "non highlight" bits of the game, he may have missed 15 easy passes, tackles, headers etc - thus bringing his rating right down - but we don't see that so we don't think it happened.

you see (in the highlights) that he makes 4-5 great passes / shots that don't end in goals - just near misses - so you (logically) think that he has done well. Now in the "non highlight" bits of the game, he may have missed 15 easy passes, tackles, headers etc

"The Steven Gerrard Effect", I like to call it

very well noted OP, supporting the appreciation of tim vickery too. he's consistently high quality.

Frankly, there's only one stat that matters in the slightest: the score.

very well noted OP, supporting the appreciation of tim vickery too. he's consistently high quality.

Why thank you very much

I'm not generally interested in South American football, but Vickery is a damn fine blogger - as you say, he never fails to provide fascinating insight.

Also on the BBC Football site, I'm suprised to find Lee Dixon turning into an excellent tactical analyst - definitely loads of tips for us FMers.

The problem with what you are saying there is that how many of us watch every match in full? I would suspect that 95-99% of us watch most (if not all) of our matches on either extended or key highlights - therefore we see perhaps 5-10 minutes of actual action during the game - thus we may well be missing the bits that either bring the individual rating up or down to it's actual rating from what we see it to be from the highlights.

eg From your example - you see (in the highlights) that he makes 4-5 great passes / shots that don't end in goals - just near misses - so you (logically) think that he has done well. Now in the "non highlight" bits of the game, he may have missed 15 easy passes, tackles, headers etc - thus bringing his rating right down - but we don't see that so we don't think it happened.

But I watch every game in full detail (one of the few!) that's why I can make my own ratings and judgements about how well a player really played