I will try to answer all the questions that have been asked so far. A few were on how the model would have done in past years. I summarized that at the end of the message. As you can see, it would have done well in some years and not as well in others but overall it would have performed very well as the top choice ran first or second in 5 of the 7 years.

More specifically to David’s question, as you can see, the model would have pointed to Invisible Ink (and Monarchos) as a great value play.

Jdub asked about how Perfect Drifts win% compares to previous top choices. Things could change after the post draw but right now he looks solid and at the likely post-time odds should be a great value.

HP asked about EOD and wondered why he was not rated higher. The model takes a lot of factors into account and crunches lots of numbers and it is not always easy to predict what it will spit out and why but EOD has several obvious knocks that I know the model picks up on. He ran fast early as a 2yo, which is a negative, he always reacts to efforts, which is a big negative for a Derby horse and he is coming in off a new top at a grueling distance (I am not sure a Derby-distance prep in March is the best way to prep for the Derby itself). His breeding is also a little suspect for the Derby distance despite the flashy names in his pedigree and his high yearling price. Finally, he comes into the race as one of the slower horses so he needs to jump up to win and given all the other factors that seems very unlikely. So he gets ranked in the middle of the pack with about 7 other horses (Blue Burner through War Emblem in the list) that all have very little chance to win but about a 15% chance to hit the board. One of these may very well do just that but most will be at odds that will not be worth a bet. EOD may sneak up into the playable range but I doubt it. Still Godolphin tends to send them out ready and he makes me a little nervous. However, they did not even bother to send any of their other horses to the Derby (they usually bring a couple) so you know he has not beaten much this year in his preps and the stable rider is going to fly all the way out to the US to ride the Oaks on Friday but does not think it is worth sticking around to ride EOD in the Derby on Saturday. Obviously, none of this latter stuff is factored into the model but it gives me more confidence with regards to trusting the model and leaving him off my tickets.

HP also wondered about Perfect Drift’s and Saarland’s high ranking. Both of these horses have classic, Derby-profile condition lines with a fast (but not too fast) race late in the 3yo year (6-10 range) and a consistently, forward-moving line without any real drastic jumps. Saarland is closer to the ideal with a nice winter break and a pair-up several points better than his 2yo top after some time off. Victory Gallop, Real Quiet and Silver Charm had similar lines. Saarland is a few points slower than several but his overall line, near-term pattern, breeding and running style all indicate that he could make a forward move in the Derby. Perfect drift is not quite as strong in any of these areas but his line is still good as is his breeding and running style and he is much faster than Saarland coming into the race. Of the other really fast horses coming in, he is the only one that looks likely to pair-up (based on the factors considered by the model) and a pair-up probably wins the Derby easily. Thus Perfect Drift is pegged as the most likely winner (note that the model still indicates he has a 77 percent chance of losing) and Saarland is not too far behind.

Jdub asked how post-position is factored into the model. Post position and running style are used to estimate ground loss. For example, a front-runner that draws an inside post will likely save ground. The computer simulates thousands of races and the ground loss is randomized for each race (i.e., a horse may grab the rail in one race and in the next he may be forced wide on the first turn) but, on average over many thousands of races, a front-runner that draws an inside post will save more ground than a closer from an inside post and a lot more ground than a presser who is stuck outside in post 20.

Jdub also asked if TG numbers are factored into the model. The answer is yes. They are a key part of the model. In fact, they are the starting point for all the calculations.

I hope that answers your questions. If not or if it just raises more questions, feel free to post away.

Chris

**********
Below are what would have been the model’s top 4 picks over the last 7 Derbies.

Each horse’s name is followed by their actual order of finish followed by their actual post-time odds followed by their model-calculated break-even odds. In theory, you look to play a horse with a reasonable chance of winning (I would say at least 10-20 percent) and with the tote odds significantly higher than the break-even odds (I would say at least 50 percent higher for horses at 6-1 model odds or less and at least 100 percent higher for horses over 6-1 model odds).

2001
Monarchos, 1, 10.5, 3.5
Point Given, 5, 1.8, 5.7
Invisible Ink, 2, 55, 7.2
Congaree, 3, 7.2, 7.3

2000
Fusaichi Pegasus, 1, 2.3, 4.5
The Deputy, 14, 4.6, 4.6
Wheelaway, 5, 20.8, 5.4
Aptitude, 2, 11.8, 6.5

1999
Menifee, 2, 7, 4.7
Stephen Got Even, 14, 5.1, 5.0
Adonis, 17, 18.7, 6.2
Excellent Meeting, 5, 4.8, 8.3

Charismatic 1, 31.3, 31.7 (missed that one by a mile, he was ranked tenth by the model but note how close his tote odds were to the break-even odds...)


1998
Real Quiet, 1, 8.4, 1.3
Halory Hunter, 4, 6.6, 4.6
Indian Charlie, 3, 2.7, 12.8
Victory Gallop, 2, 15, 13.5

1997
Silver Charm, 1, 4, 3.6
Captain Blodgett, 2, 3.1, 5.6
Hello, 8, 9.6, 5.9
Phantom on Tour, 6, 19.5, 6.2

1996
Unbridled Song, 5, 3.5, 5.1
Grindstone, 1, 5.9e, 6.1
Alyrob , 8, 7.2, 8.0
Zarb's Magic, 13, 25.7, 8.4

1995
Timber Country, 3, 3.4, 3.7
Talkin Man, 11, 4, 3.9
Thunder Gulch, 1, 24.5, 6.7
Jumron, 4, 5.6, 8.6