[Nutmeg]

Just a quick question - what statistics do you believe are worth tracking in soccer?

Here are a couple I would like to see, but haven't:

- When a team/player has possession, in what 3rd of the field do most of their touches occur? For example, what % of the time does Claudio Reyna get the ball in the defensive 3rd vs. the middle and offensive 3rd. Tough to track, but it would be interesting to chart to see how often outside backs really get forward, how often forwards really track back, etc.

- "Dangerous opportunity (DO)" sequences. Who is involved, how the play develops, etc. Again, this is tough to track because it comes into the question of making a subjective assertion as to what is or isn't "dangerous," but any attempt to track this in my opinion would be a step in the right direction. If tracked consistently, soon you would be able to see which players are most often involved in the DO sequences of a team. Pretty interesting stuff if you ask me.

- This goes hand in hand with tracking dangerous opportunities, but I would like to see, over a series of games, what is the average # of touches before a goal or a "DO" is created. How often does possession lead to a DO, how much counterattack is too much? Somehow it would be interesting to track these statistics.

- Adjustment in style of play. This I think would be the toughest to track and wouldn't be a statistic by itself, but it could be understood better by tracking a variety of stats. For example - how the ball is being played (is boomball in effect), substitution patterns, average # of touches in a possession sequence, etc. The cumulation of those stats can give a more accurate description of how a coach adjusts the style of play of a team during the course of a game. Over the long term, it might be interesting to see what is working, and what isn't.

So what other stats would you guys like to see tracked?

[Karl K]

Originally posted by Nutmeg
Just a quick question - what statistics do you believe are worth tracking in soccer?

Here are a couple I would like to see, but haven't:

- When a team/player has possession, in what 3rd of the field do most of their touches occur? For example, what % of the time does Claudio Reyna get the ball in the defensive 3rd vs. the middle and offensive 3rd. Tough to track, but it would be interesting to chart to see how often outside backs really get forward, how often forwards really track back, etc.

- "Dangerous opportunity (DO)" sequences. Who is involved, how the play develops, etc. Again, this is tough to track because it comes into the question of making a subjective assertion as to what is or isn't "dangerous," but any attempt to track this in my opinion would be a step in the right direction. If tracked consistently, soon you would be able to see which players are most often involved in the DO sequences of a team. Pretty interesting stuff if you ask me.

- This goes hand in hand with tracking dangerous opportunities, but I would like to see, over a series of games, what is the average # of touches before a goal or a "DO" is created. How often does possession lead to a DO, how much counterattack is too much? Somehow it would be interesting to track these statistics.

- Adjustment in style of play. This I think would be the toughest to track and wouldn't be a statistic by itself, but it could be understood better by tracking a variety of stats. For example - how the ball is being played (is boomball in effect), substitution patterns, average # of touches in a possession sequence, etc. The cumulation of those stats can give a more accurate description of how a coach adjusts the style of play of a team during the course of a game. Over the long term, it might be interesting to see what is working, and what isn't.

So what other stats would you guys like to see tracked?

Player peformance charting can easily do the first.

The second I think can be deduced from player performance charting. However, the difference between a great through ball and one that a keeper scoops up just as the front runner gets there can be slim. I bet, though, that most DOs will be apparent. Is a DO ALWAYS the result of the last pass before the shot??

For the third, if you assign the sequence a DO status, then it's should be straightforward to backtrack when the event sequence begins.

The fourth I think is a very hard nut to crack.

Other stats.

--On the ground passes, completed and not completed, vs. balls in the air, completed or not completed. That should be straightforward to track.

--Length of on the groundpasses -- under 10 yards over 10 yards I said in my monster post on he the big thread, I beleive, but have zero evidence to back this up, that teams who complete a good percentage of longer (+10 yards) on the ground passes are teams that advance the ball dangerously and under control, and are likely to be more successful over the long run.

--Turnovers, turnovers, turnovers. Won vs lost, especially by area of the field.

[Andy_B]

From a quick glance through my Opta Stats Yearbooks, some of what you guys are proposing is already covered, specifically stat #1 from Nutmeg and the last 3 stats requested by Karl.

Some of the others seem a bit more sublime in nature and might be tough to track and report.

Andy

[superdave]

I would be interested in "first pass after a ball is won." As in, which direction does the ball go. Because to me, what distinguishes DMs is what they do with the ball after they win it.

[Richie]

You can't keep track of everything, and who wants to. I keep track of what is important to me.

I do keep track of certain things.

Some things can be kept by one person other things you need a nother person.

The amount of corner kicks your team gets is a telling statistic. A lot of corner kicks means your team is putting a lot of offensive pressure on the oppononet. Which doe not have a lot to do with how many goals you scored off corners.

The amount of shots a player takes is misleading. Could mean he is great or it could mean he is a bad decision maker. Take his total shot attempts that includes the ones not on goal against his number of goals scored. If it is a pretty high percentage he is a very dangerous player. A player you want on the field.

The number of saves a keeper makes is misleading. Saves are on shots on goal. it does not include how many plays the keeper actually makes. Like the ones when he leaves his goal to get balls just before they become shots on goal.

Superdave and I don't get along politically, but I agree with him about winning the ball and the first pass after they win the ball. A player in general who wins a lot of balls is the first step and then can make the completed first pass is a good player when his team is defending.

Teams that can do that are on their way to becomming a very good team. I actually don't keep that statistic because when it happens you can see it, and when it doesn't happen you can see it and the player doing it good or bad. A lot of statistic just gives you get the feel of the game. I will put his name in my notes good or bad. Then fix the bad one way or the other preferbly on the practice field.

Like you can see the under 17 US National team is lacking in that part of their game. For that I blame their coach more then the players. I think blaming the player when things are not going right is a cop out by the coach for not doing his job on the practice field.

Richie

PS Reelect George W Bush. That's for Super Dave :-) haha

[superdave]

Originally posted by Richie
The number of saves a keeper makes is misleading. Saves are on shots on goal. it does not include how many plays the keeper actually makes. Like the ones when he leaves his goal to get balls just before they become shots on goal.

That's a great point. You know what would be interesting? To track the number of times a GK "wins" the ball. All you'd have to do is count the number of times he gets the ball, when the last team to have possession was the opposition, and it's *NOT* a shot.

In one or two games, I don't think it would mean that much, because some teams like to play long ball, which will give the GK alot of chances to win the ball. But over the course of a season, I think that would tell us aLOT about a GKs style. Is he a pure shotstopper, or a keeper/sweeper?

[JG]

Originally posted by superdave
That's a great point. You know what would be interesting? To track the number of times a GK "wins" the ball. All you'd have to do is count the number of times he gets the ball, when the last team to have possession was the opposition, and it's *NOT* a shot.

I think that's what the MLS catch/punch stat attempts to do. Peter Hirdt claimed in one of his columns that it had a strong correlation with winning.

[sljohn]

Perhaps this stat falls into the "duh" category... what about: scoring off of set pieces (corner kicks, direct and indirect kicks) in either the offensive half or the offensive third of the field.

To track quality of player decision-making, you could also break that down by quick kicks or not.

I also think someone mentioned (in the mothership thread) tracking the success rate of short corner kicks.

[Karl K]

Originally posted by sljohn
Perhaps this stat falls into the "duh" category... what about: scoring off of set pieces (corner kicks, direct and indirect kicks) in either the offensive half or the offensive third of the field.

To track quality of player decision-making, you could also break that down by quick kicks or not.

I also think someone mentioned (in the mothership thread) tracking the success rate of short corner kicks.

One of the interesting things about corners is, of course, that they should be considered "dangerous opportunities."

I think corners are absolutely ripe for the kind of rigorous methodolgy in the category I called "Multi-Game Metrics."

Here is what you would need to have/do in order to make this study interesting and valuable.

--First, you need to have a dataset of successful/unsuccessful corner kicks that is of a statistically valid sample type and size. Perhaps that means X number of corners over X number of MLS seasons.

--Second, you need to run the standard suite of stastical tests on this dataset. Are outcomes random?

--Third, you need to assign a goal scoring probability/value to every corner kick event. In other words, what are the odds that a corner turns into a goal?

--Fourth, now you can delve into the interesting stuff. What is the level of scoring off of corners kicks for, say, Team A, that shows that team A is a succeeding (or failing) at a rate that is statistically significantly different from merely the random/average Team? How much is that worth? And just as important, for teams who want to be successful at PREVENTING corners turning into goals, what is the rate that their opponents fail to do so, and is THAT statisticlaly signficant?

--Fifth, and finally, given the "value" of scoring from or preventing scoring via corner kicks relative to scoring OVERALL, how much time and energy should be spent rehearsing them? If a corner is worth some fraction of a goal, and could represent some statitscally significant fraction of your overall scoring, should you spend MORE training time than "usual" working on them? Should you make the creation of corners a key objective of your run of play offense?

These are very interesting questions, it seems to me, and certainly worth answering with the kind of methodological exactitude common in statistics and OR, but not so common in the world of soccer.

[sljohn]

Well stated, Karl.

Another implication is the potential to adjust defensive tactics, too. For example, if your opponent is shown to be well below average on converting corner kicks, then defenders should be less concerned about giving up a corner than they might otherwise be.