Friday, February 22, 2013

Sell Seth Davis

Every year, right around the start of the conference season, Seth Davis writes a piece for Sports Illustrated in which he treats basketball teams like stocks, and advises us whether to "buy," "sell," or "hold" our imaginary holdings in various teams. This year he published that column on January 3d. I decided to see how he is doing as of now.

Not well.

I went through and gave his advice a grade of "good," "bad," or "meh" for each of the 41 teams he rated. The result: Davis had 10 "good" picks, 13 "meh" picks, and 18 "bad" picks. You would expect a monkey to get just under 14 "good" picks just picking at random. Davis's performance of 10 was more than one standard deviation below expectations. In other words, a monkey would have gotten more "good" picks about 85% of the time.

Moreover, in 11 of the 41 picks (more than 25%) Davis was as wrong as he could be—advising to sell to a stock he should have put a buy rating on, or vice versa. According to my simulations, a monkey picking randomly would generally make just 8 such terrible picks, and Davis was again more than a standard deviation off the mean. The monkey would make 10 or fewer or screw-ups like this more than 80% of the time.

Even worse, you would expect Davis to do significantly better than a monkey because the sample isn't random: he is choosing which teams to opine on, and he is presumably choosing to give advice on the teams he has his strongest, most well-founded opinions on.

The raw data is below, with Davis's terrible picks in bold. For example, he advised to "sell" Michigan State, Wisconsin, and Miami but "buy" Kentucky, North Carolina, and San Diego State. I guess you could argue there's still time for those predictions to come true, but ... Ouch.

Preliminary conclusion: Don't let Seth Davis fill out your tourney brackets for you, because you can't trust this guy with even five bucks when it comes to college basketball. And, to be safe, don't let him invest your money either.

Team Record on 1/3 AP then Davis Rec Record now AP  now result Correct advice
Arizona 12- 0 3 Hold 22-4 12 Meh Sell
Baylor 8-4 - Hold 16-10 - Bad Sell
Butler 11-2 17 Buy 22-5 15 Meh Hold
Cincy 13-1 14 Sell 19-8 - Good Sell
Creighton 13-1 16 Buy 22-6 - Bad Sell
Duke 13-0 1 Buy 22-3 6 Meh Hold
Florida 9-2 13 Hold 21-4 5 Meh Buy
Georgetown 10-1 15 Buy 20-4 11 Good Buy
Illinois 13-2 11 Sell 20-8 - Good Sell
Indiana 13-1 5 Buy 24-3 1 Good Buy
Kansas 11-1 6 Buy 22-4 9 Meh Hold
K State 11-2 25 Hold 21-5 13 Meh Buy
Kentucky 9-4 - Buy 18-8 - Bad Sell
Louisville 13-1 4 Buy 21-5 10 Meh Hold
Maryland 12-1 - Sell 18-8 - Good Sell
Miami 9-3 - Sell 22-3 2 Bad Buy
Memphis 9-3 - Hold 23-3 21 Bad Buy
Michigan 13-0 2 Hold 22-4 7 Good Hold
MSU 11-3 18 Sell 22-5 4 Bad Buy
Minnesota 10-2 9 Hold 18-9 - Bad Sell
Missouri 10-2 12 Buy 19-7 - Bad Sell
New Mex. 13-2 20 Hold 22-4 16 Meh Buy
UNC 10-3 - Buy 18-8 - Bad Sell
NC State 11-2 23 Buy 19-7 - Bad Sell
Notre Dame 12-1 21 Buy 21-6 25 Meh Hold
OSU 11-2 8 Sell 19-7 18 Meh Hold
OK State 10-2 22 Sell 19-6 14 Bad Hold
Oregon 11-2 - Buy 21-6 23 Meh Hold
Pitt 12-2 24 Buy 20-7 20 Meh Hold
SD State 12-2 19 Buy 19-7 - Bad Sell
Syracuse 13-1 7 Sell 22-4 8 Bad Hold
Tennessee 8-3 - Sell 15-10 - Good Sell
Texas 8-5 - Buy 12-14 - Bad Sell
UCLA 10-3 - Buy 19-7 - Bad Hold
Uconn 10-3 - Sell 18-7 - Bad Hold
UNLV 11-2 - Hold 20-7 - Good Hold
VCU 11-3 - Buy 21-6 24 Good Buy
Wichita St. 13-1 - Buy plus! 23-5 - Bad Hold
Wisconsin 9-4 - Sell 19-8 19 Bad Buy
Xavier 7-6 Sell 15-10 - Meh Hold
Wyoming 13-0 Sell 18-8 - Good Sell

Here's a link to the spreadsheet I used in case you're interested. I probably screwed a bunch of stuff up. If I can remember, next year I will make random picks corresponding to Davis's, and we'll see who comes out ahead.

3 comments:

  1. Hilarious write up. Thanks for putting it together.

    It would be fun to see who wins a Davis vs. Gottlieb clash in a competition to give the worst advice.

    ReplyDelete
  2. Thanks Azzmo. I'd take Gottlieb in a heartbeat. He's a jerk, but at least he knows something about the game.

    ReplyDelete
  3. In case anyone cares, I've followed up on this in my spare time. At the regular season's end, Davis's picks improved somewhat from their terrible performance at mid-season. But still no better than random numbers.

    ReplyDelete