Snowballs statistics

[Hoop Blah]

You are wrong. Based on the fact as you see them we could argue that the table itself is invalid!! Reading v Sheffield Wednesday was played on a different day to Wolves v Sheffield Wednesday, with different teams, each team on a different run, different weather, different referees, possibly different times of day. THEREFORE it is "totally unfair" to compare the matches.

Just for the record Snowball, nobody is putting forward the league table as a work of statistical analysis.

It's a table of how many points a club has accumlated by playing each of the other teams in the league twice. Thats all it is, everyone knows the competition when they enter, everyone knows how it's judged.

My advice to you is, when you're in a hole, stop digging!

[Royal Rother]

Digging a hole? That's pretty funny actually. Don't you get it yet - he might not be the genius that is Dr H but Snowball actually knows what he's talking about, and his posts are pretty interesting and entertaining as far as stats go.

The tedium of real or might-be statisticians arguing amongst themselves about samples and populations is about as bad as it gets, but what Snowball says makes sense to most reasonably intelligent and balanced minds. (It's probably only the thick or the self-proclaimed super-intelligent who have a problem with Snowball's posts - no-one else cares about the absolute minutiae I suspect.)

I think it's fun and it's interesting but if anyone carefully reads and attempts to digest any more than a slice of the info contained in his rapidly-expanding post count they are probably a bit sad.

I really don't even care whether a few of the stats produced are slightly off-kilter or open to debate, those picking fights with the best new arrival on the Team Board in a long long time are the ones looking a bit small-minded to me.

[Hoop Blah]

Digging a hole? That's pretty funny actually. Don't you get it yet - he might not be the genius that is Dr H but Snowball actually knows what he's talking about, and his posts are pretty interesting and entertaining as far as stats go.

The tedium of real or might-be statisticians arguing amongst themselves about samples and populations is about as bad as it gets, but what Snowball says makes sense to most reasonably intelligent and balanced minds. (It's probably only the thick or the self-proclaimed super-intelligent who have a problem with Snowball's posts - no-one else cares about the absolute minutiae I suspect.)

I think it's fun and it's interesting but if anyone carefully reads and attempts to digest any more than a slice of the info contained in his rapidly-expanding post count they are probably a bit sad.

I really don't even care whether a few of the stats produced are slightly off-kilter or open to debate, those picking fights with the best new arrival on the Team Board in a long long time are the ones looking a bit small-minded to me.

Sorry, but his application of his statistics, which I've said I occasionally find interesting, in a football environment is just painful to read.

Football just isn't a game that can be broken down in such a way, especially with the stats Snowball has used to back up his well buried opinions.

[kirkrich]

I like Snowball's stats, they liven up an otherwise dull non football day. Any stats can be manipulted and are to an extent a load of old rubbish! But there is no harm in them and I find them to be informative, if sometimes rather bizarre! Keep it up Snowball (where do you find the time) and let the nerds and geeks argue amongst themselves over a game of warhammer!

[Plymouth_Royal]

Digging a hole? That's pretty funny actually. Don't you get it yet - he might not be the genius that is Dr H but Snowball actually knows what he's talking about, and his posts are pretty interesting and entertaining as far as stats go.

The tedium of real or might-be statisticians arguing amongst themselves about samples and populations is about as bad as it gets, but what Snowball says makes sense to most reasonably intelligent and balanced minds. (It's probably only the thick or the self-proclaimed super-intelligent who have a problem with Snowball's posts - no-one else cares about the absolute minutiae I suspect.)

I think it's fun and it's interesting but if anyone carefully reads and attempts to digest any more than a slice of the info contained in his rapidly-expanding post count they are probably a bit sad.

I really don't even care whether a few of the stats produced are slightly off-kilter or open to debate, those picking fights with the best new arrival on the Team Board in a long long time are the ones looking a bit small-minded to me.

Sorry, but his application of his statistics, which I've said I occasionally find interesting, in a football environment is just painful to read.

Football just isn't a game that can be broken down in such a way, especially with the stats Snowball has used to back up his well buried opinions.

so if football isn't a game that can be broken down in such a way to back up an interesting point, then what the hell are you supposed to use to back up or provide a form of evidence for such an argument?

if your good at manipulating stats in such a way that backs your argument then great and that's what he's bloody good at. I generally think there's an element of jealousy amongst some you in a sense that you cant fish out stats like he does to back up your claims. or maybe some of you feel intimidated when you make a point and then he comes up with a different one that seems more plausible, due to his ability to manipulate stats in such a way.

[Ian Royal]

The point is he isn't very good at it because his stats are easily challenged, when that happens he fails to address any points and just goes off on one.

I've posted up a variety of stats as counter points to him only to be told, essentially, that mine don't count.

He is good at producing stats and he seems to understand the art of selective reporting and manipulation though. Little more than that.

He could bring something refreshing to HNA and provide interesiting stats. But he's failing to do that quite impressively from where I'm standing. And quite a few people who actually agree with his opinion disagree with his method. And a fair few people who disagree with me on most things quite fundamentally also agree that his posts are often very flawed.

[Son of Len]

Goal superiority predicts nothing. It's a reasonable guide to which is the better team, but not a predictor by any means.

Imagine a team that wins half its games 3-0 and loses the other half 1-0. GF=69, GA=23. GD=46. Points = 69. Identical goal diff to Mr Angry's hypothetical team, but 23 points fewer.

Oh I DO so love it when people spout without thinking or checking their facts

The correlation in this year's Championship between goal-superiority and points per game is .9939

That is 99.39% totally predictive

Even if you don't bother to allow for the fact of different numbers of games played and just
look at goal superiority, the prediction is a staggering 91.3%

(data romoved due to space limitations)

I also find Snowball's numbers to be interesting, but I would not call them predictive statistics. To be predictive, there needs to be an independent variable that causes a dependent variable to come out in a specific, predictable manner. Correlation does not equate predictive ability and goal difference and ppg may correlate, but one would be hard-pressed to prove that one is predictive of the other. Most likely, other factors are the independent variables that predict both of these statistics.

In as fluid a sport as football, it is amazingly difficult to find those independent stats. Baseball, on the other hand, is much easier. For instance, if Joe Smith hits a fastball in fair territory 68% of the time and hits a curve ball (in fair territory) only 14% of the time, I will want to throw my curveball pitcher against Joe Smith. Not only are these numbers relatively transferrable from one player to another, but they are based on hundreds and thousands of experiments (pitches) going back to the minor leagues and even college/high school ball. Finding a predictive variable like the pitch that gets hit (or not) in football is awfully tough.

Nonetheless, Snowball does give us things to chew on and the resulting discussion/insultfest is almost always entertaining.

- mfg

[Fezza]

Digging a hole? That's pretty funny actually. Don't you get it yet - he might not be the genius that is Dr H but Snowball actually knows what he's talking about, and his posts are pretty interesting and entertaining as far as stats go.

The tedium of real or might-be statisticians arguing amongst themselves about samples and populations is about as bad as it gets, but what Snowball says makes sense to most reasonably intelligent and balanced minds. (It's probably only the thick or the self-proclaimed super-intelligent who have a problem with Snowball's posts - no-one else cares about the absolute minutiae I suspect.)

I think it's fun and it's interesting but if anyone carefully reads and attempts to digest any more than a slice of the info contained in his rapidly-expanding post count they are probably a bit sad.

I really don't even care whether a few of the stats produced are slightly off-kilter or open to debate, those picking fights with the best new arrival on the Team Board in a long long time are the ones looking a bit small-minded to me.

A bit harsh

I was only bored and looking for a reaction! Therefore, statistically speaking, I am neither thick or super-intelligent, just a bit of a cock!

[Snowball]

JUST LIKE a manager will say, force him outside cos he's crap on his left (or whatever)

PS Ian Royal is full of excrement. He knows well that I wish to only talk of league
stats, and why that is so. He continually throws in cup stats, knowing I will argue
that they shouldn't be there, just so he can say "I told you so."

[Sarah Star]

I just think he's on a very good fishing trip and he's really enjoying it here insulting everyone.

Looking at figures is all right, but you have to take the time to check them yourself every now and again, and sometimes life's just too short for that.

[Hoop Blah]

so if football isn't a game that can be broken down in such a way to back up an interesting point, then what the hell are you supposed to use to back up or provide a form of evidence for such an argument?

You use your opinion, reasoning, understanding of what the game involves and how that player fits and performs in relation to the game, maybe annecdotal evidence if you feel that way inclined and potentially some relevant stats if they are suitable (which in my opinion they rarely are). You use your own eyes, mind and brain to form an opinion and then try and get that over in your posts.

[Fezza]

I just think he's on a very good fishing trip and he's really enjoying it here insulting everyone.

Looking at figures is all right, but you have to take the time to check them yourself every now and again, and sometimes life's just too short for that.

He's actually out by 10 minutes on Long, he gave him 10 extra minutes against Donny!

[Ian Royal]

JUST LIKE a manager will say, force him outside cos he's crap on his left (or whatever)

PS Ian Royal is full of excrement. He knows well that I wish to only talk of league
stats, and why that is so. He continually throws in cup stats, knowing I will argue
that they shouldn't be there, just so he can say "I told you so."

Oo, more abuse there. What makes your opinion that cup games are irrelevant more valid than my opinion that they aren't? I pose an opinion, you disagree and counter with league only stats, I counter including cup stats because I use them to inform my opinion. Where is the discussion? You just ignore it.

You post an opinion, I disagree and post some counter stats or analysis, you ignore it. Where is the debate?

A discussion is about listening as well as talking. I'm taking note of what you're saying. You just seem to be dismissing what I am. Why is that?

[Snowball]

An Example of Sampling

You can take this year's results in order and "sample" them.

The population (34 games) is tiny and a sample of say half of them, randomly selected
should not be expected to be too-representative of the whole 34

But just for fun I put the games in alphabetical order and took a look at the first 17 v the second 17

The first half gave W10 D2 L5= 32 POINTS 30-19 which doubles up to be 64 points 60-38

The second half gave W8 D6 L3 30 POINTS 29-10 which doubles up to 60 points 58-20

W20 D04 L10 60-38 64 Points GD 22 Silly Prediction from A-N
W18 D08 L08 60-30 62 Points GD 30 ACTUAL
W16 D12 L06 58-20 60 Points GD 38 Silly Prediction from N-W


Considering it's a truly horrible randomisation, that's a neat sample and the prediction is close.

[Ian Royal]

An Example of Sampling

You can take this year's results in order and "sample" them.

The population (34 games) is tiny and a sample of say half of them, randomly selected
should not be expected to be too-representative of the whole 34

But just for fun I put the games in alphabetical order and took a look at the first 17 v the second 17

The first half gave W10 D2 L5= 32 POINTS 30-19 which doubles up to be 64 points 60-38

The second half gave W8 D6 L3 30 POINTS 29-10 which doubles up to 60 points 58-20

W20 D04 L10 60-38 64 Points GD 22 Silly Prediction from A-N
W18 D08 L08 60-30 62 Points GD 30 ACTUAL
W16 D12 L06 58-20 60 Points GD 38 Silly Prediction from N-W


Considering it's a truly horrible randomisation, that's a neat sample and the prediction is close.

But there is no cause and effect so there is absolutely no relevance or use to it. Or is that what you are trying to say? I don't really see the point of that.

[Snowball]

so if football isn't a game that can be broken down in such a way to back up an interesting point, then what the hell are you supposed to use to back up or provide a form of evidence for such an argument?

You use your opinion, reasoning, understanding of what the game involves and how that player fits and performs in relation to the game, maybe annecdotal evidence if you feel that way inclined and potentially some relevant stats if they are suitable (which in my opinion they rarely are). You use your own eyes, mind and brain to form an opinion and then try and get that over in your posts.

So PLEASE explain why there are such polarised opinions over, for example, Harper.

Are we not all capable of watching a game and thinking a player played well (or didn't)

Actually we are NOT that good, and thousands of experiments have shown that

As Keith Hackett has said, "perception and actuality are not the same."

[Ian Royal]

none of that makes use of a small selection of types of stats anymore useful though.

At least if you're clear in explaining what your opnion is based on and why, it should be shown to cover a myriad of reasons. Rather than, for example, goals scored, assists made and tackles made per game.

Take some of my reasons for thinking Harper is a good player and contributes a lot.

He covers other players (mainly defenders) positions when they are not where they should be or need to be.
He slows down attacks by quickly running to opposition players on the ball
He forces passes that aren't especially creative or dangerous from opposition players by doing the above.
He makes a lot of simple quick passes out to our wings where we are most dangerous and creative.
He (usually) rarely gives the ball away to the opposition cheaply.

Quantitify those with stats if you can!

[West Stand Man]

I also find Snowball's numbers to be interesting, but I would not call them predictive statistics. To be predictive, there needs to be an independent variable that causes a dependent variable to come out in a specific, predictable manner. Correlation does not equate predictive ability and goal difference and ppg may correlate, but one would be hard-pressed to prove that one is predictive of the other. Most likely, other factors are the independent variables that predict both of these statistics.

In as fluid a sport as football, it is amazingly difficult to find those independent stats. Baseball, on the other hand, is much easier. For instance, if Joe Smith hits a fastball in fair territory 68% of the time and hits a curve ball (in fair territory) only 14% of the time, I will want to throw my curveball pitcher against Joe Smith. Not only are these numbers relatively transferrable from one player to another, but they are based on hundreds and thousands of experiments (pitches) going back to the minor leagues and even college/high school ball. Finding a predictive variable like the pitch that gets hit (or not) in football is awfully tough.

Nonetheless, Snowball does give us things to chew on and the resulting discussion/insultfest is almost always entertaining.

- mfg

You are right in saying that these are not predictive, but they can be used to show a pattern. They are not predictive because football is a random activity that cannot be predicted, which i think you have reasonably shown in the latter part of that post.

The patterns are, nonetheless, important and hence why OPTA make good money out of providing stats to clubs. If you can show a pattern and develop a probability of repetition then you have something to work with. So, if a goalkeeper dives to his right for penalties more often that to his left you might then wish to dig into that and see what other patterns occur. Maybe he always dives right in the 2nd half but is less predictable in the first (for instance). That doesn't mean that he will dive right for your penalty - but you may be able to influence him to do so as it is his prefered selection anyway. What makes the raw numbers take on a meaning is the patterns that develop around them.

Equally, linking unrelated stats is mathematically wrong and needs to be highlighted. For instance, just because I can see that player x gets 1 goal for every 100 minutes on the pitch and player y gets 1 goal every 90 minutes doesn't mean that they will, necessarily, be a good partnership. Even if their mate in the team is only bagging one goal every 350 minutes it could be that he is the catalyst they need to get their goals.

In short, stats ARE important but they neeed to be used in context and with the right patterns being identified. Those of you who say they are meaningless are simply wrong. Those of you who set them up as the maidens prayer are also wrong.

As for Snowball, he started well and obviously has the time and intellect to manipulate this data. I would suggest that he goes OTT on the basis for his analysis at times - and that is where you should direct your intelligent critique.

[Snowball]

An Example of Sampling

You can take this year's results in order and "sample" them.

The population (34 games) is tiny and a sample of say half of them, randomly selected
should not be expected to be too-representative of the whole 34

But just for fun I put the games in alphabetical order and took a look at the first 17 v the second 17

The first half gave W10 D2 L5= 32 POINTS 30-19 which doubles up to be 64 points 60-38

The second half gave W8 D6 L3 30 POINTS 29-10 which doubles up to 60 points 58-20

W20 D04 L10 60-38 64 Points GD 22 Silly Prediction from A-N
W18 D08 L08 60-30 62 Points GD 30 ACTUAL
W16 D12 L06 58-20 60 Points GD 38 Silly Prediction from N-W


Considering it's a truly horrible randomisation, that's a neat sample and the prediction is close.

But there is no cause and effect so there is absolutely no relevance or use to it. Or is that what you are trying to say? I don't really see the point of that.

It's an example of sampling PART of a population to predict the whole population
just like we do with opinion polls. This has nothing to do with cause and effect
just as Shane Long's stats are not about cause and effect, they are accurate renditions
of reality that are BETTER than mere spectator perception

And that's because people approach the actual sights on the pitch with prejudice.

We saw Forest "cheating". I didn't see any READING cheating or time-wasting at Sheffield
yet the Wednesday fans were chanting that we were cheats...

Bias, like love is blind, distorts the reality of what we see.



A silly little story. Newport County had a player called Peter Passey ex Birmingham. He had tunnel vision and did everything in a straight line

One day, we were winning comfortably, he burst out of defence and did 3-4 one-twos, found himself in front of the keeper and lashed it home. He was a clogging defender
and at first glance he had turned into Pele.

I used to hang with a few of the team and they told me the truth (according to them)

PP NEVER carried the ball (because, they said, he was too scared to) and "the lads" (for a laugh)
kept giving it back to him as he ran. What we saw as brilliance was him panicking. The rest
of the team were p!ssing themselves with laughter. He scored one other goal in six years at the club

[Snowball]

none of that makes use of a small selection of types of stats anymore useful though.

At least if you're clear in explaining what your opnion is based on and why, it should be shown to cover a myriad of reasons. Rather than, for example, goals scored, assists made and tackles made per game.

Take some of my reasons for thinking Harper is a good player and contributes a lot.

He covers other players (mainly defenders) positions when they are not where they should be or need to be.
He slows down attacks by quickly running to opposition players on the ball
He forces passes that aren't especially creative or dangerous from opposition players by doing the above.
He makes a lot of simple quick passes out to our wings where we are most dangerous and creative.
He (usually) rarely gives the ball away to the opposition cheaply.

Quantitify those with stats if you can!

Ian, Ian, what do you think the above is if not statistics?

Example 1. "He covers other players (mainly defenders) positions when they are not where they should be or need to be."

Snowball asks Ian, "How do you know?" and Ian says, "Because I have seen him do it 13 times this half (or whatever). That is, you consciously or unconsciously DO log behaviour. If he did it the odd time, say once a game, it wouldn't raise up as "something he does." It's TOTALLY quantifiable. Problem is it's time-consuming. It's the sort of thing the professional game-watchers do for clubs. pro-=Zone being a classic example.

Say you tell me, "Harper has a fantastic engine, he will run and run." OK I can see he runs, but Pro-Zone can actually tell me that (for example) he ran 14K in the game whereas other players ran maybe half as much and Seol ran forty-eight metres to pick up his drink.

These are stats. Now they have stats for passes, completed passes and CRUCIAL passes, tackles, missed tackles etc.



The problem with laymen viewing anything is YES they use intuitive statistics, but they are seriously affected (and their perception distorted) by DRAMATIC information.

So, say for example we have two strikers, same starts, same subs, same minutes played. Striker A has scored 125 goals, Striker B has scored 75 goals, but striker A is a Lineker type, loads of simple tap-ins. He gets boring goals. Striker B gets "AMAZING" goals,. a thirty-yarder, an overhead kick, a mazy dribble. Fans tend to over-estimate the flashy player and underestimate the "boring" one. This is just an example. Please don't start saying, the guy who hits from long-range brings something extra etc (that's not the point.)