Computer Generated Odds Line for the ROTW

[derby1592]

Based on the morning line, Deputy Dash looks like a great value. Sunday Break is the most likely winner but will be over bet if the morning line holds. Fast Decision and Ibn Al Haithmam are two other contenders that may possibly offer some value if they stay at or above their morning line. The rest look like poor plays with the possible exception of Essayons if he floats up significantly above his morning line.

Good luck to all.

Chris

***********
Horse followed by its \"Break-even\" odds based on computer model

Fast Decision, 3.6
Sunday Break, 2.4
Deputy Dash, 3.5
Essayons, 13.4
Ibn Al Haitham, 6.7
Fireballer, > 99/1
Heir D\' Twine, 27.2
Puzzlement, 49.1

Bonus computer odds-line coverage:

\"Break-even\" $2 exacta payouts for Deputy Dash with contenders:

Deputy/Fast, $44
Deputy/Sunday, $49
Deputy/Ibn, $51

Fast/Deputy, $47
Sunday/Deputy, $34
Ibn/Deputy, $78

P.S. Notice the seemingly \"strange\" break-even exacta payouts. That is because although Sunday Break is more likely to win than is Fast Decision, he is actually less likely to run second than is Fast Decision; hence the strange payouts. This is often the case with a favorite. They are seldom good value bets underneath in exactas (or trifectas).

[mholbert]

Based on your odds line, the correct exacta \"fair pays\" are:

deputy/fast $34
deputy/sunday $26
deputy/ibn $56
fast/deputy $34
sunday/deputy $24
ibn/deputy $62

[mholbert]

one more try...

deputy/fast $32
deputy/sunday $24
deputy/ibn $54
fast/deputy $32
sunday/deputy $22
ibn/deputy $60

[derby1592]

Mike,

I am guessing that you are using a poplular formula to estimate the fair exacta payoff based on projected win odds. This is a decent approximation but it is not as accurate as what the model uses. The model actually simulates many thousands of races and keeps track of who runs first, second, third, and fourth in each race. This produces a much better estimate of the fair exacta payoff (IMHO).

Chris

[mholbert]

I would suggest that your logic and your math are flawed.  This is not uncommon among horseplayers.  My team\'s success relies upon, and relishes, mistakes such as these.  

It is interesting that when people talk about unsophisticated money in the pools, they are refering to those that make wagers based upon names or colors.  The far greater opportunity lies in exploiting areas of hueristic instability.

Although I have no idea what our models will construct for this race since these are done with two minutes to post, I can say that initially we differ considerably on two horses.  Our models give significantly less chance to Deputy Dash and significantly greater chance to another horse.  Who\'s model is right?  I know how our models have performed over the last several years, but one race is one race.  However, I can state for certain that your exacta fair pays need some more work.  But again, I encourage you to keep using them.

[Michael D.]

Mike and Chris,

Do your models take into account any trainer or jockey stats?

[derby1592]

Mike,

It is nice to come across someone who is so certain of themselves in this game. I envy you.

I guess I just gave you a little \"mustard\" to add to your team\'s \"relish.\"

I am guessing that the big difference in win odds line in our models may be based on the figures that we are using. I used TG, which has Deputy Dash faster than several other figures going into the race.

[Anonymous User]

I make four horses contenders. Sunday, Fast, Al and Deputy. My guess is the only horse you\'re gonna get your \"break even\" odds on is Deputy. I make him the least likely winner of the four. Does that mean you\'ll bet him at 12-1 if the others go off below \"break even\"? And do you think with Deputy at 12-1 the \"break even odds\" have any relationship to the actual probability of that horse winning the race?  I\'ve seen a couple of these projections you\'ve posted in this regard and I haven\'t seen that relationship. If your posting your personal discipline thats fine, but call it that.

Good Luck,

Tabitha

[derby1592]

Sorry about that, I somehow accidently cut  the last message short.

Regarding the exacta payoffs: using formulas that assume that \"fair\" win odds are highly correlated with \"fair\" exacta odds is OK but inferior to the Monte Carlo approached used by my model. Of course, as with everything else in this game, that opinion is open to debate although the shortcomings of such an assumption are pretty well documented in the book, \"Efficiency of Racetrack Betting Markets (Economic Theory, Econometrics, and Mathematical Economics,\" by Donald B. Hausch, Victor S. Y. Lo, and William T. Ziemba (Dr. Z) et al.

As you said, the value in any model is based on performance in the long term, so if you have a successful model, stick with it. In fact, I would be interested in finding out more about your model.

Chris

\"The whole problem with the world is that fools and fanatics are always so certain of themselves, but wiser people so full of doubts.\" Bertrand Russell

[Anonymous User]

Whoops...always revising even to the last...I don\'t know how T-Graph can attempt to pick the day or two before..lol.

Deputy Dash looks to be the benefit of circumstances and may be a stronger influence than I implied above. The more I look at this race the more difficult it appears.

Double Good Luck,

Tab

[derby1592]

Tabitha,

I don\'t understand what you stated or what you are asking or implying in your last statement.

The \"break even\" odds ARE the \"odds representation\" of the model\'s estimate of the probability of each horse winning. There is subjectivity involoved in the model input (i.e., some of the input reflects my opinion) but the primary inputs are TG figures.

I am not sure what past postings you are talking about so you will have to be more specific.  

If I play a horse in the Peter Pan it will be the one that offers the best value even it if is not the most likely winner (i.e., not Sunday Break). To do anything else would be folly and would ensure a less than optimal ROI in the long term. This is the primary tenant of all the \"value\" handicapping literature out there.

So, if I am tracking with what you are asking/stating, the answer is \"Yes, I would definitely play Deputy Dash at 12-1 if the other 3 main contenders were below their morning line.\" If the model, is accurate (admittedly a big if and I am not claiming that it necessarily is), I would have about a 22 pct chance of cashing a win bet (and a 78 pct chance of losing) but, even so, I will end up way ahead in the long term making such wagers.

I hope that response made some sense.

Thanks

Chris

[mholbert]

sorry for not getting back sooner.  we generally run thursday thru sunday, but with the holiday, ran thru monday this week.

1. i would be extra careful in using ziemba\'s work.  things have changed somewhat in the last 18 years.  there is a great deal of research out there.  racetrack betting offers fertile ground for economic research due to the availability of data.  however, if you want some humor, look at dick mitchell\'s work on \'fair pays\'.

2. i would agree on monte carlo.  back when ziemba\'s work came out, i was using a trs model 100 at the track doing monte carlo simulation.  it that instance, monte carlo was vastly superior.  however, you must remember that it is only as good as the prediction model.  in looking at your exacta fair pays, they looked off.  the one good thing is that they were higher than what they should have been.  i wonder if you could show all the possible combinations and your calculated \'fair pays\'.  i\'m not saying your model is wrong. let\'s rule out a math error.  the numbers just don\'t align with a very large race sample.

3.  on the peter pan.  deputy dash had to be an overlay on most odds lines.  we had him at 15/1.  our model used deputy dash and puzzlement lightly and fast decision more heavily weighted.

4.  i got a chuckle out of your quote.