After thrilling us with their AL predictions yesterday, ESPN released their NL team capsules today. They're all linked from the MLB Index (although the Rockies' link is broken; their preview is here). I don't really care about the actual previews at all, but the win predictions for each team from ESPN's analysts sure are fun.
These previews have predictions from Jayson Stark, Tim Kurkjian, Buster Olney, Keith Law, and Steve Phillips. I'll get to the more interesting specific predictions next week, but first some more general stuff.
Average Win Predction
This is not complicated. If all 30 teams play a full season, the average team will win 81 games. I don't think this requires any further explanation.
Kurkjian: 82.1
Stark: 82
Olney: 81.3
Phillips: 81.1
Law: 81
Congratulations to Keith Law. The only ESPN analyst who can add. Give that man a raise.
That money should probably come out of the paychecks of Kurkjian and Stark. I would like to see the process these individuals go through when making their predictions. I am pretty sure they look at the list of teams, arbitrarily assign a win total to each one, and that's that.
Here is my question: what's the point? Clearly, the guys who don't consider numbers at all are not good at this. Would it be that hard to add up your predictions and make sure that they're, uh, possible? Also: do they not have editors? If they do, do they not know how to add?
PECOTA Correlation
Law: 0.93
Kurkjian: 0.81
Olney: 0.80
Phillips: 0.80
Stark: 0.78
This is not rocket science. Law's predictions will most likely do very well. He clearly at least put some thought into this. The others will do about as well as yours or mine would do if we arbitrarily picked numbers for each team. I find it amusing that ESPN trots out these predictions like they mean something. They do not.
Standard Deviation
Phillips: 10.7
Olney: 10.4
Kurkjian: 10.1
Stark: 9.7
Law: 9.0
For reference, PECOTA's standard deviation is 8.4.
Phillips is so absurd. Here's my favorite little stat from all of these: he has 14 teams winning 88 or more games. Think about that for a second- that's one team away from half of baseball. Here is my prediction: Steve Phillips' predictions will not fare well in this post at the end of the year.
Thursday, March 27, 2008
Subscribe to:
Post Comments (Atom)
9 comments:
Great research. I love it!
Personally, I'm surprised most of them were even that close. Especially Phillips. (Law is the exception, of course - I knew his would average to 81 even before I scrolled down to read it).
I didnt check the math... but is this average for NL teams or all teams? Because there are interleague games.
All teams. That's completely unclear if you haven't read a previous post...my fault.
LOFL
Thrilling commentary, I know.
this won't surprise anybody but just for a reference, here are the number of 88-win teams from each of the past 10 season (since last expansion)
11,9,10,12,9,11,10,8,9,11
14? I'll take the under?
I'd thought something was fishy! Something about the numbers seemed off, but I hadn't done the math.
Great work, just posted some commentary linking to this post. Nicely done.
Maybe they all think there's going to be massive logjams at the top of each division, necessitating a whole bunch of one game playoffs to determine who goes to the postseason and raising the win totals of a bunch of teams. And in Kurkjian's case, he just thinks there'll be 33 of 'em...
I was actually thinking about that this morning, SBDD. Excellent point. I'll feel quite dumb when every NL team ends up 79-83.
Post a Comment