Statistical "Corrections"

Some player statistics on the league site contain a number of mistakes. Especially affected are magicJack players. Please see the end of this topic for other examples.

 

I saw this too. Yael Averbuch had one assist through the last game of the season, and when I went to look at the year end stats, she had 2. But I noticed it was given on one of Marta's goals in a subsequent game against Atlanta. Can't go back and check the video, but I'll believe it.

As far as mJ goes, definitely a possibility that they went back and added those assists. Though Washington's 7 assists are the only thing that stands out looking at their team stats, which puts her one behind Sinclair for the league lead, I think they're pretty solid. The league schedule says mJ had 20 assists, the league stats say 21--that's close enough for me. League schedule says mJ has 262 shots, league stats says 257. I mean, just to use another team as an example, Boston's league stats show 169 which equals the amount tallied on the league schedule, but Boston actually keeps legit stats. Again since it's magicJack and the league has been left to calculating this stuff themselves from visiting team's boxscores (I don't even know who else would've kept them for mJ homegames), I think it's pretty good.

Since these stats are put up by the league anyways (as opposed to the teams themselves), I trust them more.

 

reviewed... magicjack films? what films? must have been an away game.

 

So for those who are familiar with the statistic "Wins Above Replacement," I put together a pseudo-WAR stat to rank the forwards this season.

Very simply, this is a stat which measures a player's offensive performance on the field in terms of wins. Think of it as an indicator of how well they played when they were in the game. Some played alot more minutes than others, some were super subs and some missed most of the season due to the WWC. Thus some extrapolating had to be done.

Just to give brief description of how I arrived at this stat. I calculated the average goals, assists, and shots per 90 minutes for all non-starters/league leaders (so no Alex Morgans or Casey Nogueiras). I then extrapolated this to what an average "replacement" players performance would have over 18 games. I then did the same for almost all the players (because very few played in 18 games), so that I could have a standard sample size to compare to. I calculated the Goals created for each player by simply taking their projected goals over 18 games + projected assists divided by two (because a player is only half responsible for a goal if on an assist). The average goals per win this season was 2.3, so I figured 1 win=2.3 goals created. The average goals created for replacement players was 5.2 = 2.2 wins. So the individual players' wins above replacement are just their win value - replacement level win value (2.2). Thus if a player has a 0, it doesn't mean they didn't contribute at all, it means they contributed at replacement level.

So for example, Christine Sinclair had 10 goals and 8 assists in 1191 minutes. If we extend that to 18 games, it comes out to 13.6 goals and 10.9 assists. 13.6 + (10.9/2) = 19 Goals created. 19/2.3 = 8.3 wins, a WAR of 6.1.

Just to preempt any hysterics, yes some players ranked very high in only limited playing time. This is partly a factor of extrapolation of their playing time over 18 games, but I've left it as is (as opposed to making it relative to playing time) because at the same time shows how much of an impact that player had when she was on the field.

There are alot of things that aren't factored into this stat (yet) like defense, a player's endurance, home/away games, etc. That's because 1) I haven't figured out how to factor those things in yet or 2) the league doesn't provide stats on that particular area, particularly defense (i.e., tackles, takeaways and turnovers). That is why I just used it to compare the forwards.

I took out players who: 1) were listed as forwards but really weren't, 2) did not have at least 200 minutes played 3) were injured (namely Lindsay Tarpley).

Let me know what you guys think/How I can improve this statistic:

Rank PlayerWAR
1 Christine Sinclair 6.08
2 Lisa De Vanna 5.73
3 Abby Wambach 5.38
4 Marta 5.19
5 Alex Morgan 3.42
6 Veronica Boquete 3.27
7 Tasha Kai 3.22
8 Laura del Rio 3.03
9 Christen Press 2.60
10 Lianne Sanderson 2.53
11 Casey Nogueira 1.33
12 Adriana 1.18
13 Megan Rapinoe 1.10
14 Katie Schoepfer 1.02
15 Kelley O'Hara 0.82
16 Eniola Aluko 0.76
17 Lauren Cheney 0.58
18 Gemma Davison 0.57
19 Ella Masar 0.55
20 Kelly Smith 0.29
21 Nikki Washington 0.13
22 Amy Rodriguez 0.00
23 Meghan Lenczyk -0.14
24 Laura Kalmari -0.22
25 Frida Magnusdottir -0.40
26 Kandace Wilson -2.20
27 Kiersten Dallstream -2.20
28 Kristina Larsen -2.20

(Put this here because I couldn't find a "stats" thread, let me know if it should be moved)

 

box score for the championship game had the goals scoring summary in the wrong halves. in fact, it indicated that at the end of regulation, Philly led 1-0. It would be great if in the offseason the WPS could "upgrade" its reporting of official box scores to the ones used by most division 3 colleges. The MLS has the same issues. Box scores should not be typed up manually. They should run on a simple, efficient software package, called statcrew:

http://automatedscorebook.com/report/soccer/index.shtml

http://automatedscorebook.com/report/soccer/game/ncaa box score.pdf

 

The WPS already uses a (presumably expensive) commercial statistics package from STATS LLC. Box scores are generated by the software, not typed in manually. Usually the problem is that incorrect information is keyed in by the statistician during the game and the mistakes get rolled up into the stats. I'm not sure what the problem was for this particular box score since the goals seem to have the correct times recorded, unless perhaps the software requires you to enter goals and goal scorers separately, in which case perhaps the goal was keyed into the wrong period.

With the decentralization of the league and budget cutting, I'm not sure who is now in charge of recording statistics at games. Does the league still pay STATS LLC to do it? Or, is each team responsible for the task? (And if so, have teams been hiring pros to do it, or is it something that they've just had the interns do?)

 

hahaha, there's a guy who video taped several of the home games, and the video is excellent quality, better than the web cast. i forgot who he is.

hahaha x 2, as crazy as dan is, it looks like he is going to be back in 2012. he gets to keep his football team, and the Freedom haters will continue to wish him ill health.



ROY

Fresh off a successful collegiate career at Stanford University, the reigning 2010 MAC Hermann trophy winner has quickly become magicJacks’ go-to for goals. Press has scored eight goals for her South Florida-based team, second most for her club, including the first-ever hat trick scored by a rookie on July 30. Press climbed up the WPS leader board, scoring five goals in the second half of the season and finishing fifth among league leaders in goals, just two behind the PUMA Golden Boot winner.

we may be few, but MJ fans are a live and kicking.

 

I glanced at Statcrew pricing, it's like $160 to $400 for a license. The league should get on board with it and have it as standard across the board. I'm not sure what the point of STATS LLC is, because the format of the box scores being only text based, and error ridden, defeats the purpose of having a box score. Like I said, the NCAA is more professional looking, at the D-3 level, than WPS stats. Although you make a good point about the role of the statistician in making whatever package work, I think the least that statcrew could do would be better than the best case scenario that stats LLC could offer, but IMO, isn't offering. And if they don't want to go to PDF's, statcrew also makes text based box scores but they are in a more readable font and they include shots, shots on goal, and every other stat, per player for everyone in the line-up as well as the formation/position. Statcrew is presented so neat, whereas Stats LLC is coming across like a sloppy mess. Did you download the sample box score links above?

 

Stats are kept by team representatives - STATS is the conduit to getting the information onto the website in near-real time and for official match records. With that said, I know the team statistician for the Flash, and I saw him input the information into the program correctly (considering I was with him at the game). I checked the box score, and found it odd that the stats were showing the goals in the wrong period....must have been an issue with the data once the game was over. I know he creates a copy of the information for his records after the game......nothing seemed out of the ordinary after he closed out of the program.

The Flash utilize 2 people for stats (who also double as the referee liasions / game day coordination team). One keeps track of the stats on the computer, and the other spots for him (and also keeps a separate track of the action in the event there is a technical issue). In addition to that, there is an on-field coordinator who calls up information from the field and/or verifies assists when necessary. After the game, WPS utilizes Match Analysis to double-check accuracy of stats and makes adjustments when needed. So, there is a process to correct stats that are incorrect, but like most everything else, seeing it live and recording it down vs. looking at it on video can be a whole different thing. The guys who do the gameday stats are both USSF referees, so they're usually pretty good about discerning what should go in the stat book (i.e. shot on goal vs. shot off target).

Just my $.02....

Aaron

 

thanks for the input. just covering my bases by saying that it doesn't matter what package you have, that sometimes incompetence will reign supreme, so I understand that having a more professional program wouldn't necessarily equate to perfection if the people don't utilize it properly, you can still get bad results.

I don't see how any Statcrew program would allow goals to tally to the incorrect periods. It is obvious that this had to be some kind of a manual, data entry error. Perhaps the STATS program they use was not set up with capability of showing overtimes and PK's, so they found they had to modify the box score manually (if so, they should be more careful with manual entry to avoid such an error). Either way, I fault the system for having such an obvious loophole, as much as I fault the statisticians for not proofreading & correcting it.

But I hope they upgrade to division 3 programs enxt year for better presentation as well as better accuracy

 

I'm well familiar with Stat Crew from the American football version of their The Automated Scorebook stat recording program. It's a good system. However, I'm guessing that the WPS instead went with STATS LLC for a number of reasons:
- fully integrated with the WPS web site to provide player/team stats
- provides Match Tracker live game reporting
- provides the Fantasy Challenge game

Another factor is that, as I understand it, the box scores on the WPS web site might not be directly generated by the STATS LLC program. At the beginning of WPS, the box scores were only available as PDF files, and it seemed as if the STATS LLC program only generated PDF box scores. There may have been some intermediate step created to convert PDF files into HTML box scores. The PDF versions contained more information than the HTML, so some of the stats were left out in the process. Anyhow, the error in the championship game box score might have been during the PDF-to-HTML conversion.

Thanks for the info, DazzRef. So, individual teams are responsible for hiring people to record stats during games. Being a professionally-run team, the Flash hired two USSF referees to do this. It makes me wonder what other teams more reluctant to spend money on game infrastructure and front office staff might have done. More moonlighting tech support employees?

 

Whoever was doing the stats for Freedom home games in 2010 had 52 of the 56 shots taken in those games as being on goal - including 100% (33 of 33) in the first and third game. After I and others pointed this anomaly out on BS, the stats started looking more normal for following games.

To the WUSA's credit, they hired the Elias Sports Bureau after the 2002 season to review all their first two seasons of tape and correct various errors in the stat keeping.

 

Which led to Abby being credited with a double-double (10 goals, 10 assists) as initially she was only credited with nine assists.

 

It's interesting because the feedback that has been given has been positive - referees in a sense know what to look for in a soccer game and might have a better idea of what is important to record, and how to assist the game day officials before and after the game. I don't think that it's really a function of anything specific - the 2 guys I know have been doing this for the Flash since their initial year in the W League. Could other teams adopt the same model - they could.....but I think it's all about how each team works within its own model. I don't think there's a hard and fast rule for the gameday coordination work that goes on behind the scenes. As long as the training around using the stats software is the same, there shouldn't really be much of an issue with the entering of game stats. If the league decides to review and change any stats, then that's OK - I believe that this is common amongst other sports anyway.

 

I'm still waiting for WPS to correct the first goal in league history. It was incorrectly credited to Alison Falk. It should have been credited as an own goal (Briana Scurry).