Liverpool FC Forum

by **JoeTerp** » Sun Jan 04, 2009 12:51 am

if you read some of the method in the how the stats were measured in American football (in the links that I provided), you would see that they did not just take measurement of how many yards were gained and divide by how many times the player got the ball. A football eye watched the play a few times in order to determine the value that each player contributed to the play.

of the top of my head a model would have to involve some type of scale that was able to account for the difficulty of pass attempted, the other options available (opportunity cost), the level of pressure the passer was under, where on the pitch the passer was, where on the pitch the pass went, how far the ball traveled, controlability of the pass, the danger creation of the pass. Then rate all of these on scales from -10 to +10. Then figure out a way to weight each category apropriately to come up with an overall rating for each individual pass, then this stat could be used in many ways both in average score per pass and total score created through passes

Like I said it would take a long :censored:

time and it would have to be done by the same set of eyes to keep things consistent, and it would never be perfect, but at the same time, nobody will ever draw a perfect circle (even with a compass, (if you go by the most strict mathematical principle of what a "circle" is)), so perfection is not the goal, the goal is to represent the value of player by a universal measure using numbers

by **Sabre** » Sun Jan 04, 2009 12:57 am

I promise I'll read the applications and the links you provided.

of the top of my head a model would have to involve some type of scale that was able to account for the difficulty of pass attempted, the other options available (opportunity cost), the level of pressure the passer was under, where on the pitch the passer was, where on the pitch the pass went, how far the ball traveled, controlability of the pass, the danger creation of the pass.

And we'd still be scratching the surface. What if in a theoretically worse part of the pitch there's weak player. What's the best way to goal. How do you measure the risk of interception of a very good pass, and the risk of getting caught in counterattack. How would you find a numerical method that determines if it's better a long ball, or simply passing back (in american football you have to go always forward gaining yards). It would be very, very, very complex, let alone the time needed and the experts needed.

by **JoeTerp** » Sun Jan 04, 2009 1:02 am

bigmick wrote:In golf they use a lot of stats, driving accuracy, driving distance, sand saves, puts per round etc, and once more they can be decieving. The guy with the least puts per round is quite often the guy who goes in the most bunkers, his escapes from bunkers to three feet making him look like a better putter than the better player who hits greens in regulation and has birdie chances from 15 feet.

I think Golf, they are starting to use avg. length of putts made as the better measure of who is actually putting better, or they will show percentage of putts made from 5-10 feet because those are the real testers that you need to put away with consistency in order to win tournaments.

I don't think stats are meaningless its just important to know what they are missing and not to assume to read more into a stat that what it actually says, like point of the team that gives up the most goals does not automatically have the worst keeper (or even the worst defense for that matter). I am also a much bigger fan of the more complicated stats that weight in different factors like the QB rating in football vs. completion percentage or OPS in baseball vs. batting average.

by **JoeTerp** » Sun Jan 04, 2009 1:14 am

Sabre wrote:I promise I'll read the applications and the links you provided.

of the top of my head a model would have to involve some type of scale that was able to account for the difficulty of pass attempted, the other options available (opportunity cost), the level of pressure the passer was under, where on the pitch the passer was, where on the pitch the pass went, how far the ball traveled, controlability of the pass, the danger creation of the pass.

And we'd still be scratching the surface. What if in a theoretically worse part of the pitch there's weak player. What's the best way to goal. How do you measure the risk of interception of a very good pass, and the risk of getting caught in counterattack. How would you find a numerical method that determines if it's better a long ball, or simply passing back (in american football you have to go always forward gaining yards). It would be very, very, very complex, let alone the time needed and the experts needed.

Clearly a complex quesiton but to give a simple yet pragmatic answer, I would suggest an opportunity cost rating that measures all the other possible choices that passer could have made and chose not to, and in that category account for those factors and then give that category a very high weight when combining categories.

by **JoeTerp** » Sun Jan 04, 2009 1:19 am

And also, there is always going to be more things to factor into when determing what goes into the "passer rating formula" but at some point you have to pick a place to draw the line and it would probably start where the ability to quantify gets really really hazy* And I would error on the side of including too many things, but giving some of the "last to make the cut" a very low weight in the formula.

*its all a bit hazy to start with and thats why its tough to decide when it gets out of hand

I would also think the rating would split itself into two basic categories in skill with the foot and skill the with mind and having all account for the external factors as best as possible.

by **account deleted by request** » Sun Jan 04, 2009 2:14 am

Or easier yet, if you don't like stats ..... don't read them or comment on them. There are plenty of opinions to discuss.

by **JoeTerp** » Sun Jan 04, 2009 2:20 am

s@int wrote:Or easier yet, if you don't like stats ..... don't read them or comment on them. There are plenty of opinions to discuss.

spot on.

by **Owzat** » Sun Jan 04, 2009 11:31 am

Quadrophenia wrote:Only one stat matters. The scoreline, that's the only stat that will be taken into account in years to come.

I agree

Some stats are more popular than others, what I like to call "impact stats" which are short but sweet - number of titles, the odd statistic that says such and such went unbeaten or the first side to win since 3009, no side has won the Premiership having finished fourth the previous season etc

As for one or two points in the thread starter's post, or more specifically some types of stats :

Defence stats - at the end of the day a defence is measured by goals conceded and these are measurable. More significant perhaps in the league where you play 36 of the same sides as every other team - the two different being us play the mancs, the mancs play us etc You could argue luck into the equation, but how many times have we got away with lucky near misses compared to how many times we've had bad luck? ie it probably evens out

Goalkeeper stats - you could apply the defence stats to the keeper, but how much are clean sheets down to the keeper, the defence or even the team as a whole? If the defence stops attacks before it results in a shot then it's not the keeper's doing, if the team dominates possession and minimises shots to make sure a clean sheet is kept then how is that the keeper's doing? Was Cavalieri's clean sheet fully earned? Or were six goals in his previous three appearances more reflective of his ability?

Possession - I like this as an indicative stat, if it's too far from 50-50 then it's probably a mismatch. But at the end of the day it's what you do with it, having 60% of the possession and drawing 0-0 is one example of how little possession means.

Striker goals - one of the hot potato topics for strikers. Goals to appearances ratio is the standard measure, some like to factor out subs appearances and other theoretical influences such as irregularity of appearances. Fair enough, but as a basic measure it is commonly accepted a top striker averages a goal every other game or thereabouts. Owen does that and he can hardly be accused of being injury free or starting every game. Obviously the more starts, minutes on the pitch and runs of games you get, the higher the chance of you maximising your goals output. But short of analysing every minute, frequency of playing etc and coming up with a convaluted rating system to try and level it up, then goals to appearances is the most accepted method.

Compare it to the mode, mean and median averages. Which is the best? Each has a weakness, even the most commonly accepted and most mathematical average - the mean. How can you find the 'perfect' average? You probably can't, if I were to design one it would probably use elements of all three averages, but that would still be somewhat artificial and subjective. So goals to appearances is accepted as the mathematical average of the goalscoring measures, not perfect but applied consistently is the most commonly used.

Comparing positions in the table - most will compare this season with last and see we have more points. In fact we have more points than in any previous Premiership season. Is this a fair comparison? To a degree, I would use that comparison but also compare it with the corresponding results since the order in which you play games can also be a factor.

The point of all that waffle? Interpretation of stats can be a minor or major issue, I guess it depends what you're trying to show/prove/imply/demonstrate. At the end of the day stats prove nothing in terms of who is the best striker, who will win the title etc. But stats are an interesting way of analysing, can show trends and so on. For instance a pre-match analysis of head to heads over a time period won't determine a result, but if we always beat WBA then there's a reason and good cause to suspect that is the more likely result. The head to heads against Stoke would be worthless because a) we have not played them much in the last 20 years and b) because we have slipped up against other sides we should have beaten down the years - Hull, Stoke, Reading, Wolves, Swindon, Barnsley - in fact we have only one 100% winning record and that is over WBA, even sides like Swindon and Wolves who've played only one season in the Premiership have drawn against us.

by **Owzat** » Sun Jan 04, 2009 11:36 am

JoeTerp wrote:
s@int wrote:Or easier yet, if you don't like stats ..... don't read them or comment on them. There are plenty of opinions to discuss.

spot on.

Exactly. Everyone is entitled to post what they like and how they like, if you don't like the posting style then ignore those posts you aren't interested in. The stats I (and others) post aren't there for the sake of it, they are there for those who find them interesting to read. It is an alternate way of looking at things, summarising something in numeric form rather than the bogstandard "Babel was sh1t", "we have the best defence", "we're playing well", "we're not scoring enough goals", "we're on course to win the league" etc.

by **Sabre** » Sun Jan 04, 2009 11:48 am

Owzat wrote:
JoeTerp wrote:
s@int wrote:Or easier yet, if you don't like stats ..... don't read them or comment on them. There are plenty of opinions to discuss.

spot on.

Exactly. Everyone is entitled to post what they like and how they like, if you don't like the posting style then ignore those posts you aren't interested in. The stats I (and others) post aren't there for the sake of it, they are there for those who find them interesting to read. It is an alternate way of looking at things, summarising something in numeric form rather than the bogstandard "Babel was sh1t", "we have the best defence", "we're playing well", "we're not scoring enough goals", "we're on course to win the league" etc.

Nice try, but that's like asking Mick to not post on rotation and Rafa's management if he doesn't like it

Plus, I like stats. I've studied them quite deeply as a math tool. I'll say more, I like people putting the numbers in the board so that those are analysed.

But sorry, it's hard to not comment on the stats, while in the middle of a nice thread of football someone comes and says something like "Mascherano is dull and ineffective and here's the proof", with a stat in which you can see that he doesn't make a lot of assists.

That stupid attempts to prove something blatantly wrong with a stat, have to be commented and dismissed.

Anyway, don't worry, if you don't like those comments, don't read them

by **dawson99** » Sun Jan 04, 2009 11:50 am

The thing with stats is you need the WTF factor, and I don't know how to add that to find out the proper stats. there is also the sh!t happens factor. football is not a science, it's passion, its VERY human.

by **Sabre** » Sun Jan 04, 2009 12:10 pm

There's quite a lot of truth in that Dawson.

Yesterday I read my mate's comments on here BEFORE watching the game.

It might seem according to many that we missed Alonso in the second half... but there are other more important factors.

FA Cup. Disallowed goal (correctly), the stadium is outraged, one big headed guy like me says Oi!!!!! behind the ref while explaining why he disallowed the goal. The pressure increases, the oppo players dive to get free kicks, the ref concedes them. The pressure is big.They created danger out of it.

So the stats did get we concede more free kicks than usual, but the stats do not get FA cup is different, how was the ref, and how was the pressure. And that HUMAN factors count.

Our biggest mistake problem yesterday was not that Alonso was injured, but that we didn't kill the game before

Liverpool FC Forum

Statistics. - How do we interpret them?

Who is online