Player Peformance Charting -- Let's Get it Going Again

[Karl K]

Well, I will be happy to inaugurate our new forum with its very first thread.

Let's get Player Peformance Charting going again!!

For more information, see:

http://www.playerchart.freewebsites.com/

Last time the effort was abortive, as some who promised they would do it didn't deliver.

Some criteria:

--I would like to do it strictly for senior USMNT games starting when qualifying begins in earnest early next year. If we could do it for every qualifier, that would be great.

--I would like to get a cadre of about 25 to 30 game charters, so if somebody can't do it for one game, another person can sub in if necessary.

--If you say you can chart, you HAVE to come through. This is very labor intensive. But really, you will learn a lot about the game, and about the players.

--To participate you must have videotape machine and tape the game. Ideally you should have a pause and rewind remote control so you can review events on the field efficiently.

--You must input data into a spreadsheet (Excel is best, but Lotus or Works will do). You must follow the event designations precisely in your data entry.

--Last time we did it, each charter was responsible for a 15 minute chunk of the game. We can increase that or decrease it depending on willingness and number of participants. The more people we have, the shorter the time period you are responsible for, and the more efficient the process.

--I will not chart. My role will be to gather up all the data, run it through the Access database, create PDF files, and post them on the web site.

--Anybody a slick Access programmer out there? I am OK at it, but there are probably very useful enhancements that could be done.

--Those who chart games have first dibs on writing analytical articles for the web site. I hereby anoint myself editor. As editor, I only ask that your analyses focus on aspects of player peformance revealed (or not revealed) by the player charting process. Other than that, only civility is required.

Those who are interested, go to my PM page and e-mail me off line. I will send you a bunch of additional material you can familiarize yourself with.

I would also suggest that you practice charting a few times before we do it for the USMNT. MLS games are perfect for this. I would be happy to review your data and give you some pointers.

That's it for now.

[Nutmeg]

Karl,

I suck with Access, but I can put together a VB application. If this is going to grow or if you think this will ever move in a web-based entry application, you may want to rethink the access design and go with something else. Again I could help there.

As I said in the PM, I can help out with the charting, too.

[microbrew]

I could write Perl scripts, but I don't have an cgi account that's open to the web. at would bypass the .pdf conversion, and perhaps make the database access interactive.


As for charting- that will wait until I finally get a VCR or Tivo.

[beineke]

A few thoughts about charting ...

(1) "+" notation

I might have mentioned this before, but the one thing I don't like about the current charting notation is that "+" can mean "successful" or "especially good." So a header won is "Hd+," which means a very good header won is "Hd++" -- this is potentially a little confusing.

I'm wondering if we can substitute "c" for successful ("completed"), so that "+" can be unambiguous.

(2) Side-of-field notes

As I recall, our charting doesn't have any notation to say whether the play is on a touchline, or in the center of the field. Am I remembering correctly? If so, is that something we want to add?

(3) MLS Goals

Using the Multimedia archive at mlsnet, I'm planning to chart some of this season's MLS goals (to start with, all goals scored by the Quakes). Karl, would you be interested in hosting that data?

[Karl K]

Originally posted by beineke
A few thoughts about charting ...

(1) "+" notation

I might have mentioned this before, but the one thing I don't like about the current charting notation is that "+" can mean "successful" or "especially good." So a header won is "Hd+," which means a very good header won is "Hd++" -- this is potentially a little confusing.

I'm wondering if we can substitute "c" for successful ("completed"), so that "+" can be unambiguous.

(2) Side-of-field notes

As I recall, our charting doesn't have any notation to say whether the play is on a touchline, or in the center of the field. Am I remembering correctly? If so, is that something we want to add?

(3) MLS Goals

Using the Multimedia archive at mlsnet, I'm planning to chart some of this season's MLS goals (to start with, all goals scored by the Quakes). Karl, would you be interested in hosting that data?

Beineke, I will review the database over the next week to see if those modifications are possible. I think they are.

And yes, I owuld be happy to host the MLS data.

[Karl K]

There are some new folks to stats board, so I wanted to bump up this thread in an effort to get some more folks interested.

Take at look at our site...link in the first post.

[Karl K]

Some additonal background on this subject.

For 2002, we tried to do what I called "Player Performance Charting" for J/K 2002. The web site I put together to publish the information, along with some seminal work done by Ric Miller, is still up.

See:

http://www.playerchart.freewebsites.com/

The process of doing player performance is very labor intensive, which is why it was so hard to get follow through. Which was a shame, because I had built an entire methodology, and a Microsoft Access database to gather up all the information, and to create some interesting reports.

This effort was originally inspired by the work of Ric Miller, whose now defunct site, “The Game Behind the Game” was really fascinating and stimulating. Basically, Ric argued that many post-game assessments of player performance were highly impressionistic in nature, and therefore fraught with all the dangers of subjectivity. In an effort to put some precision behind player observations, Ric put together a very interesting shorthand. All of Ric’s shorthand tracked just “ball touch” events. I tried to take his foundation and go beyond in much more detail.

Basically, Player Performance Charting is designed to create a record of player actions during a game. This record, or chart, is created from detailed videotape watching and recording events. These recorded events are then run through a Microsoft Access database to produce the reports.

The charting system I devised is designed to track mostly "on the ball" events in the game -- number of touches taken, passing, dribbling, shooting, crossing, tackling, etc. In addition, we made provisions for certain "off the ball" events such as runs made, shields to retain possession, and situations where defensive pressure and cover is applied.

Besides quantifying such actions, game charters had the option of "qualifying" such events. Was this a good dribbling sequence? Or bad? Was the pass lost an especially bad giveaway or acceptable give the dynamic of the game, such as an ambitious forward pass to a front runner?

From our charting efforts, we wanted to produce three reports. The first, the Game Narrative, is a minute by minute listing of what the players did, in the order in which they did them. It chronicles the type of events in the "event sequence" (a "one touch, then "a completed pass" would be two events in an event sequence). It tells us what area of the field this event sequence occurs (defensive, middle or attacking third; or defending or attacking area). And there is a space for optional commentary on the event sequence.

The second report is the Narrative by Player. This report shows us the event sequences of each player separately; you can look at this report and see, for example, all of DaMarcus Beasley's events over the course of the game.

The third report is Player Summaries, and really the most interesting report. In the summaries we total up for each player (and the team as a whole), key event types -- such as how many touches on the ball did this player get? How many tackles won and lost? How many corners created...or corners conceded? And the like.

I’ll talk about the shortcomings of this system in a minute, but let me say this. The information this system provides can be VERY revealing about an individual player’s performance particularly. Remember back in 2001 when many were hot to have Jeff Cunningham get caps? Well, I charted his performance in a game against the Dallas Burn during the summer of 2001, and I was just shocked at the limitations in Jeff’s game. He won very few balls, checked to ball rarely, played little defense, made few combinations passes. Like the most naïve youth soccer player, all he wanted to do was run off, get the ball at his feet, face up to goal, and dribble.

As in the case of Cunningham, the charting system we created can tell us some key things about a player or a team's performance. For example, it can give us a good indication of how "active" a player is -- is he getting touches on the ball, is he winning tackles, is he completing his passes to teammates? If he is a defender or defensive midfielder, is he winning balls, making good clearances? If he is an attacking player, is he getting good touches in the attacking third or in the attacking area? Are his dribbling sequences good?

The system can answer certain kinds of questions – some of which were the source of endless (some might say monotonous) debate about the qualities of certain players. . For example, does Claudio Reyna spend most of his time in the middle or defensive third retrieving balls...or does he "get into the attack?" Our charts were designed to answer this question.

But because soccer is a game where so much happens off the ball, there are things that the charting system can't tell us. For example, it won't show whether a player has taken up a good position off the ball, or whether the team's defensive "shape" is strong, of if the team created good "depth" on the attack. And a "pass completed" may in fact be a "hospital ball" from Player A that puts teammate Player B under duress, forcing Player B to whack the ball desperately to get out of trouble (though there the game charter has the option of qualifying the nature of this pass).

Over the course of a game, though, these ambiguous situations should be set in the context of a longitudinal performance -- most players are going to get anywhere from 35 to 70 touches per game. Meanwhile, to deal with these situations, and to add richness to the charting, game charters had the option of adding brief comments to qualify events and explain what happened. That's why we always wanted the Player Summary reports to be be viewed in the context of event comments.

I would love to resurrect this effort for the USMNT, particularly in qualifying and the runup to the next world cup.

[Kevin in Louisiana]

I have the Holland and (I think, but I might have taped over it) the Denmark game. I'd be willing to do a ten-minute or so chunk.

[TomEaton]

I love this sort of stuff. Analyzing the data, I mean. Actually collecting it is a ton of not-very-exciting work, which is probably why you've been having trouble getting people to commit to it. There is (or was, at least; I haven't checked back there for a long time) a website called matchanalysis.com that did this sort of work for a fee--you could even send them tapes of your youth team if you wanted--but it was very expensive. To get a whole season's games charted, even for just one team, was cost-prohibitive for me. I'd be willing to pay money for somebody to do this kind of work and send me the results if it was something I thought was reasonably priced. Is there enough of a market for this type of information to consider paying the people charting the games?