Thankfully, No One Pays Attention

Thankfully, no one pays attention to my annual Kentucky Derby picks, because if they did, they would have read this prediction that I made Friday:

If I had to pick a last-place finisher (a bet they would never actually offer at the track because people involved with horse racing understand better than most that people respond to incentives), it would be Mine That Bird.

And then they would read this headline in The Boston Globe‘s Sunday’s sports page:


But it gets worse before it gets better. I named five horses that I did like. One of these horses finished dead last out of 18 horses, another finished next to last!

The other three finished respectably: third, fourth, and tenth.

In light of this showing, will there be any demand for my Preakness picks? I think so. You see, when it comes to predictions, there is as much value in someone who can predict as poorly as I do as there is in someone who predicts the truth. You just take the bad predictor’s picks and do the opposite.

Fred T.

Look at it this way: Your predictions for the weekend were basically as good as the local weatherman's.


How did Paul Krugman's picks do?


In the Washington Post, Andrew Beyer said Mine that Bird should have been 200-1. His rapid improvement in speed figure (105 from 81) was unprecedented. The beauty of sport is the unpredictability. Ever so often these extremely improbable events happen. It's why I keep watching.


Busted - I totally noticed! I thought it was awesome, actually; it just further highlighted his outside chances!


Take solace, Steven. Your two big picks (I forget their names), 20-1 long shots at your writing, were both better than 10-1 bets, respectively, at the start of the race!

Either you had 'em valued correctly, or the racing market was listening.


I took your "bet this one last" pick and ran with it to a nice payday.


Maybe you should suggest the opposite of what your program indicates (because something like this happened last time as well, if I remember correctly?)

chris markl

lol I did follow your advice and bet on Papa Clems and General Quarters, don't feel bad though, just some two dollar bets.

I read through the comments on your original post and kept thinking that because your model did so poorly in this once instance, doesn't mean your model is full of error. The question is, if you applied your model to a random sampling of races would you win money? BTW wasnt the track was sloppy on saturday, I would imagine this enters some randomness or atleast needs to be accounted for.

So yes Id love to see your preakness picks.


Your horseracing predictions reminded me of the "Dogs of the Dow". It also serves the purpose of contrarian bettors who will simply bet on the horses that run counter to your picks. So, it does serve a purpose. Will you do the same thing for stocks?


I did notice Steven - but did not post a comment to the original article as i felt like pouring oil into a burning flame. I appreciate the fact that you owned up to the bad pick unlike certain stock market experts last year....

Jamey Stegmaier

I attended the Derby and put down $20 on Papa Clem, per your advice. He raced well, and I would go with your predictions in the future (although, apparently the same jockey has won two out of the last three years...maybe he knows what he's doing), but I must say that I had the feeling during the race that any horse could have one. Everyone at the track seemed sure that their pick was correct, but when Mine that Bird won, you could have heard a pin drop in the infield.


I am really interested in your model - would you mind posting it for readers? Bleg request.

Ex-post results are not a good indicator of ex-ante probabilities.


Incorrect predictions also tend to be cheaper for consumers. That's a nice bonus...


"You see, when it comes to predictions, there is as much value in someone who can predict as poorly as I do as there is in someone who predicts the truth. You just take the bad predictor's picks and do the opposite."

So, are you telling us to do the opposite of what you tell us to do?

Tim Cullen

Can you give us any analysis of what went wrong with your modeling? Were there incorrect assumptions? Someting as mundane as a data entry error? Or, is your model really just not applicable to horse racing?

Trevor L

A question: If I were to put $10 on a longshot (such as Mine That Bird) every year for the Derby, would I make a profit eventually? I'm wondering this not only in light of Saturday's Derby but also another derby in which the longshot (maybe it was longshots) came in at the top.


In the live trading pits, taking the other side of someone who consistently loses is a legitimate strategy. (Source: Sam Seidan). Another example is how so many amateur investors buy a stock at exactly the wrong time only to see it drop moments later. One should ask themselves, what are professionals doing that isn't so obvious on the surface? Whether it's insider knowledge, statistical analysis, or a good gut instinct, some people are making a living through betting (stock market or horses). What do they know that we don't?


I am currently reading The Black Swan by Taleb and immediately noticed how poignant this post is to the thesis of the book! The book has really changed my thinking in future prediction -- especially as it relates to the stock market! I especially like the quote (I am paraphrasing here as I don't have the book with me): randomness is nothing more than "unknowledge". Since the future is truly unknown and thus an "unknowledge" to us, a person's guess (regardless of who he/she is) is as good as one from somebody else.

Mojo Bone

During that extra-long trot between the race and the roses, I kept looking under the horse for the rockets the jockey must have lit off, coming out of the fourth turn. That horse was having as much fun as the jockey.


I understand models can be wrong in specific instances since we are looking at expected payoffs.

That isn't your fault, but it's things like this that make me wonder about about the models economists use to set national policy, especially in times of crisis. I mean, what if despite economists best efforts, our economy is in one of these outlier cases and the model does as well as your model did!