Longshots

[traynor]

An interesting fringe benefit of the research I am doing currently involves the relationship between "speed" (however defined, including average pace, O-Fin, and similar designations) and mutuel prices.

When "all speed ratings" are lumped together, much is obscured. When those speed ratings are sorted and ranked, even more is obscured. I suggest that you may find it interesting to study entries that won that were not in the top 3 "speed" (average pace, O-Fin, whatever method you use to evaluate "speed"). Ignore both odds and morning line (many of which are little more than derivatives of real or perceived "speed").

It would be pointless for me to post racks of numbers, because I use different data formats than most, and generate track-specific data. I dislike sweeping generalities that may seem useful, but only if you bet on each and every one of the next 10,000 or so races at every track every day.

What I am suggesting is that if you cull out the top 3 "speed" entries, it is MUCH easier to locate (and identify) the specific attributes of entries that are both likely to win and to pay a decent return when they do. Those attributes vary from track to track, with a great deal of overlap.

[traynor]

An example, current at The Meadows, trot races:

0.07 TW-11 TM-147 TR-0.07 FW-647 FM-5109 FR-0.13 [41] [48] [57] [23] 0.59

What the numbers mean:
Filters: Horse must have wins 10% or better, must have had odds in its last race between 3/1 and 6/1, must have been up close at the wire in its last race--within 2 lengths of the winner, and must be driven by a driver who is NOT the top rated driver in the race.

"Speed" ratings (however good they may seem) and "pace" ratings (however good they may seem) are ignored completely.

Of the 147 True Matches (TM-147), only 11 won (TW-11). That is 0.07, folks. 7% winners. 93% losers--including those with "the best" speed and pace ratings.

The point is that a horse that may have the best speed rating, average pace rating, pace rating, O-Fin rating, or whatever else rating (including various measures of class and current form), and that looks fairly good on paper, should be an automatic toss out in trot races at The Meadows.

[mrroyboy]

This is exactly what I am trying to do Tray. What figures to use is the problem.

[traynor]

This is exactly what I am trying to do Tray. What figures to use is the problem.

It doesn't take much to find some good tossout recipes. Pass betting on a few of those, and it can improve your return, with very little effort. The thing that really interests me is that it is often an entry that looks good on paper that has such defects. That is why I suggested looking (first) at entries that were in the top 3 speed ratings (or pace ratings). They usually take the most money, and are easy tossouts. That leaves the "real contenders"--with a lot better prices. And they are a lot easier to find.

[HWIG]

An interesting fringe benefit of the research I am doing currently involves the relationship between "speed" (however defined, including average pace, O-Fin, and similar designations) and mutuel prices.

When "all speed ratings" are lumped together, much is obscured. When those speed ratings are sorted and ranked, even more is obscured. I suggest that you may find it interesting to study entries that won that were not in the top 3 "speed" (average pace, O-Fin, whatever method you use to evaluate "speed"). Ignore both odds and morning line (many of which are little more than derivatives of real or perceived "speed").

It would be pointless for me to post racks of numbers, because I use different data formats than most, and generate track-specific data. I dislike sweeping generalities that may seem useful, but only if you bet on each and every one of the next 10,000 or so races at every track every day.

What I am suggesting is that if you cull out the top 3 "speed" entries, it is MUCH easier to locate (and identify) the specific attributes of entries that are both likely to win and to pay a decent return when they do. Those attributes vary from track to track, with a great deal of overlap.

The general betting public bets time and time alone for the most part. The horse with the best time will likely be the favorite.
In order to get a decent price to have to minimize the time factor.

I do find that time is a prime factor in picking winners in lifetime races, NW1, NW2, NW3, NW4 & NW5. These horses have not been classified yet, so the only thing going is form and time. I handicapped these races by first determining the contender in the race, then used the best time for each horse in the past thirty days. I picked a ton of winners that way.

[traynor]

The general betting public bets time and time alone for the most part. The horse with the best time will likely be the favorite.
In order to get a decent price to have to minimize the time factor.

I do find that time is a prime factor in picking winners in lifetime races, NW1, NW2, NW3, NW4 & NW5. These horses have not been classified yet, so the only thing going is form and time. I handicapped these races by first determining the contender in the race, then used the best time for each horse in the past thirty days. I picked a ton of winners that way.

I agree (although I don't use the best time in 30 days). There are times when a relatively simplistic speed (or pace) rating is predictive. What I suggest is that it may be profitable to note the exceptions--the situations in which speed ratings/pace ratings are NOT predictive of the race winner.

You make an interesting point in your sequence of analysis, that is at the core of some very sophisticated software--first determine the contenders. If that step is weak or deficient, the speed ratings/pace ratings are much less useful.

The applications I use combine the steps. Rather than first selecting contenders, then rating speed/pace/whatever, the entries are analyzed for predictive attributes in a single step. It eliminates a LOT of confusion and second guessing. Perhaps "holistic" would be a more appropriate way to categorize the process, rather than sequential. Many of the attributes are analyzed on the basis of "gaps" (as in "fuzzy logic") rather than simple ranking. It works. Well.