Saturday, August 20, 2011

Crew IQ

Yesterday, Craig Merz wrote the following over at "When healthy, his bullish runs have effectively put him in a position to score. He had half of the Crew’s first eight goals of the campaign, and has the best goals per 90 minutes average in MLS (0.73) for players with more than five appearances."

Renteria has been playing well, but the goal stat mentioned above is misleading.

1st Problem: Measurement system. "per 90 minutes" used as well as the "appearances." Emilio has 10 appearances and 616 minutes. That equates to 6.8, 90 minute games. Hitching appearances to p/90 statistics will take you down the wrong roads. Roger Torres has been in 16 games for Philly but only has 489 minutes. When pulling goal scoring efficiency, or any other efficiency, "appearances" should not be used as a qualifier.

2nd Problem: Not enough data points. 10 appearances or 6.8 games for Renteria does not give you enough data points to draw any real conclusions other then he had a good start. It's actually dangerous to look at soccer efficiency this way. Millions have been paid to poor players because of a few good games.

3rd Problem: Merz includes goals off penalty kick. Renteria has 5 goals but 2 off PK.

Oh PK. Saving and extending careers since 1996.

Renteria's real goal p/90 is actually good (0.44). It's a rate I see often when I look at good forwards. But is it the best?

Here is what you get when looking at players around Emilio Renteria's minutes:

Rate: 0.63 / 714 MINS De Rosario (time with DC only)
Rate: 0.57 / 632 MINS Alan Gordon (time with TFC only)
Rate: 0.56 / 809 MINS Conor Casey (Injured)
Rate: 0.45 / 792 MINS Brian Ching
Rate: 0.44 / 616 MINS Emilio Renteria
Rate: 0.40 / 453 MINS Steve Zakuani (Injured)

That puts Renteria in some good company but you can also see that using low appearances (5) as a qualifier yields exceptions (injuries, recently traded players) which is an indicator that says 5 apps is not enough time to pull any real substance out of.

Regardless, even using Merz system his claim on Renteria isn't accurate anyway. According to Stats, Inc. Paulo Jr. has 5 apps and a p/90 rate of 0.89 for RSL.

Past performance predicts future results.

In 2010 Renteria scored 5 goals in 857 mins or 9.5 games. His rate was an extremely impressive 0.52. Combine last year with this year and we get 16.4 games played and a goal scoring rate of 0.48.

That's very good. Comparing to guys with similar minutes in 2011? Eric Hassli (0.49) / Landon Donovan (0.48). I'd say that's good company.

1 comment:

Unknown said...

your maths skills are beyond impressive, it's becoming more and more intriguing to me. and especially your efficiency in how you approach the numbers. but... i must say, as sure as david seaman's ponytail, saying that "Past performance predicts future results." - I feel like I need a qualifier in there. Like "can aide in". All three of us contribs are in the subjective biz, and maybe that's why we love the game, it's subjectivity.