THE BOOK--Playing The Percentages In Baseball

I actually just read this a few weeks ago. Not a fan. The authors relied heavily on regression, like when they determined Iraq is the biggest overachiever based on a regression equation involving population, GDP & # of games played, but ignoring strength of schedule.

I also found the chapter on penalty kicks bizarre. They concluded that “penalties do not matter” (their words). Why? Because the % of games won by the home team or by the betting favorite is the same in games with a penalty kick as they are in games without a penalty kick. Mind you they don’t bother to look at which team had the penalty. Just whether or not the game featured a penalty from either side.

Maybe I’m missing something, but isn’t this like saying home runs don’t matter because home teams have the same win% regardless of whether or not the game contains a home run, regardless of which team hits it. Is that useful?

One would think they would at least tell us that home teams win 50% of the time they don’t get a penalty kick and 65% of the time they do get a penalty kick (or whatever the real numbers may be). That seems like such an obvious fact to share. Nope. I have no clue the win impact of an awarded penalty kick.

For me, the book was underwhelming. None of the statistics presented seemed all that convincing or compelling.

I just finished that two days ago, and was going to review it.  Scooped by the Globe again! 

Anthony/1, I haven’t read it, so I have no idea, but were they talking about matches with penalties awarded in regulation time when they said that, or matches decided on penalties?  I would think the latter would be the more interesting topic to address since the fairness vs. randomness of a shootout is a big debate issue in soccer, and that line sounds like something that would be referring to a study of penalty shootouts compared to matches decided in regulation or extra time.

Kincaid: they were definitely talking about matches with penalties awarded in regulation time.

I had a glance at this book in Foyle’s in London last summer. I did not buy it, because it’s neither fish nor fowl. It’s not a performance-analysis book and it’s not really about the use of performance analysis in football. It seemed instead to be a book by a couple of blokes who thought they’d like to ‘Moneyball’ English football, but discovered that there aren’t enough statistics kept to do it. (I could have told them that.)

Kuper’s been writing columns about football for the Financial Times for some years now, and some bright spark of a book editor somewhere no doubt approached him to do a book on it, so he threw together some of his old columns and revised or deepened some information with Szymanski’s help.

Instead, I’d suggest people see if ‘Fink Tank’ is still online at timesonline.co.uk Daniel Finkelstein used data collected by some chaps at Warwick University to analyse football, and came up with some interesting insights.

#1: what’s a bit weird is that they discount Iraq’s ranking due to its strength of schedule, then take it all back with their closing sentence in the chapter.

"Maybe I’m missing something, but isn’t this like saying home runs don’t matter because home teams have the same win% regardless of whether or not the game contains a home run, regardless of which team hits it. Is that useful? “

A difference there is that we tend to assume that home runs are reflective of ability, whereas the context for the penalties question is the assumption that they are not and instead introduce too much chance into the outcome of the match (due to being too randomly awarded or that they have too large an impact on a match given the success rate on penalties and the avg score of a game).

If they found that in matches w/o penalties, the visitor won 27.23% of matches, but that in matches with penalties they won 37% (or whatever pct would make the wins/losses even), you might infer evidence that penalties make the outcome of the match a crapshoot and are thus problematic.  At that point, you’d probably want to look deeper and compare more specific circumstances.

Fra, thanks for that reference, it looks interesting.  For those who are interested, I’ve attempted to link the archive page for the column to my handle.