Topic: RAWK Round Table Hull City AFC 3-1 LFC

[The Woolster]

I thought the whole point of PDO was to measure how you are running vs the mean- ie up to now has your team got lucky or unlucky .... Are you in a false position or not ?

Therefore of course it regresses because all PDO measures would be at the mean given enough time
Ours being high will mostly mean we are due to regress ie we've run good up to now

I wouldn't say thats the point of it, I'd say its a measure of how efficient a teams scoring and saving ability has been, its just that it has been found that the correlation for year on year PDO is low, so it is used as a measure of luck.

But just because the average team regresses, does not mean all teams regress. Like I said, Utd under Ferguson had a high PDO year on year I believe. Is that because they were lucky every season, or did they have an efficient attacking system with efficient strikers, a good defence and a good shot stopping goalkeeper, all whilst spending a lot of time in the lead?

Whilst we are not at Utds level yet and our PDO may well drop, I think we are good enough that it will remain above average all season.

I've remembered reading an article about PDO and its regression depending on team strength, I'll see if I can find it.

[The Woolster]

This is all true, though as far as I can tell, the more the model takes shot quality into account, the better we look. MCofA's, for instance, which is entirely based on an expected goals model that heavily relies on shot quality, has us as more or less even with the other top 4 hopefuls. The worrying thing is that while predictiveness of offensive shot quality is high, the same isn't really true of shot quality conceded. Anyway, leaving that last piece aside, I think we're actually looking a littler better that your account makes it seem, we're in the top 2 or 3 for shot quality, both offensively and defensively, and a lot of models haven't really incorporated that data yet. I'd say we're top 4 offensively, and mid table defensively. Definitely needs improvement, as you say.

Didn't see this post, so apologies for repeating a lot of what you said!

[ElstonGunn]

I wouldn't say thats the point of it, I'd say its a measure of how efficient a teams scoring and saving ability has been, its just that it has been found that the correlation for year on year PDO is low, so it is used as a measure of luck.

But just because the average team regresses, does not mean all teams regress. Like I said, Utd under Ferguson had a high PDO year on year I believe. Is that because they were lucky every season, or did they have an efficient attacking system with efficient strikers, a good defence and a good shot stopping goalkeeper, all whilst spending a lot of time in the lead?

Whilst we are not at Utds level yet and our PDO may well drop, I think we are good enough that it will remain above average all season.

I've remembered reading an article about PDO and its regression depending on team strength, I'll see if I can find it.

I'm struggling to find the post now, but someone found that PDO "skill" varied to the tune of only  about 6-8 goals a year at the extremes. The survivor bias point is a good one though, and I'm not sure if it's been accounted for.

[The Woolster]

I've remembered reading an article about PDO and its regression depending on team strength, I'll see if I can find it.

This is what I was refering to http://11tegen11.net/2013/01/06/what-is-a-normal-pdo/

[Timbo's Goals]

Jesus wept.

 

[The Woolster]

I'm struggling to find the post now, but someone found that PDO "skill" varied to the tune of only  about 6-8 goals a year at the extremes. The survivor bias point is a good one though, and I'm not sure if it's been accounted for.

It was defo James Grayson.

[JackWard33]

I'm struggling to find the post now, but someone found that PDO "skill" varied to the tune of only  about 6-8 goals a year at the extremes. The survivor bias point is a good one though, and I'm not sure if it's been accounted for.

Presumably that's indicative of the way the game works generally. We think skill makes the difference because it looks good but actually mistakes, weak links in the team, the overall level of the team and luck are bigger influences

[JTK]

These threads have turned from people talking sensibly about football into people just chatting bollocks if you ask me.

Are you genuinely using stats to calculate luck? Kinell

[GrkStav]

It may not be very interesting, but it's sensible and it's not bollocks.

There is a super-tanker's load of bollocks being chatted in other threads, with usually zero evidence and scant logic behind it.

[JTK]

It may not be very interesting, but it's sensible and it's not bollocks.

There is a super-tanker's load of bollocks being chatted in other threads, with usually zero evidence and scant logic behind it.

Using stats to calculate 'luck' is the least sensible & most illogical thing I've seen on rawk today - including a man being very possessive over what goes in his fridge.

Do you know what luck is? Nevermind me, if you're happy then carry on.

[JackWard33]

These threads have turned from people talking sensibly about football into people just chatting bollocks if you ask me.

Are you genuinely using stats to calculate luck? Kinell

Of course - luck is completely measurable. When it happens can't be predicted but it's overall effect on football can be measured

[JTK]

Of course - luck is completely measurable. When it happens can't be predicted but it's overall effect on football can be measured

I would view it as not only unnecessary, but also impossible to judge what is luck and what isn't. Any tiny factor could affect a football match, from a player tripping over, what a striker has for breakfast, whether the referee missed his opticians appointment. That could be teams being 'lucky' or unlucky.

There's some great points in this thread but some stats just seem silly to me. That said, if you find it interesting then I don't want to stop you.

[The Woolster]

Of course tiny factors can affect football matches, that is why stats people look through to the underlying factors where there are far more observations, which then reduces the luck factor. Instead of goals, you look at shots, you look at shots on target, you look at shot quality, you look at how the chance was created, and you don't look at 1 game, you look at hundreds or even thousands of games. Then you can get an idea of which teams may have been lucky or not.

As for their use, they help you understand what is happening at the basic level of the game, and a lot of people looking at them will be using them for prediction, ie betting or fantasy football.

[The Woolster]

With good timing, two articles published today which help explain why our shot ratios might be low and PDO high

http://www.statsbomb.com/2013/12/score-effects/

http://thebubblegame.wordpress.com/2013/12/04/expected-pdo/