Base odds line

[chickenhead]

I've been trying to come up with a clever approach to making a first pass odds line. Something simple, mechanical, and broad.
I wanted to see if anyone thought this was a good idea, or had tried it before, or can see an obvious flaw.

Since speed figures are, or try to be, a kind of apples to apples comparison, seems like they might be of good use here. My thought was to compare every PP line to every other PP line, and record results.

For example, so you had a race, 3 runners (A and B and C) with 10 lines each. Compare each of A's lines with B's and C's first line, record results. Compare each of A's lines with B's second line and C's first line, record results, and so on. For 3 horses 10 lines you get 10^3 combinations, or 1000. For 10 horses 10 lines each you get 10^10 combinations, or 10 billion combinations.

Sounds a bit crazy at first, but the more I think about it the more I like it, the numbers are big enough that one or two bad races don't swamp out the rest, like they could with just an average.

This would give an odds line for every horse at every position.

Anyone care to comment? Remember this is just for a baseline that would be adjusted later to reflect todays race set-up.

[OTM Al]

I have dabbled with odds lines but have not jumped in yet. One thing I did try once in a similar vein was to simply use the average of the past 3 speed figs. That's a whole lot simpler than what you suggest and I think the average taking isn't so bad. First of all, its recent form. Going 10 deep may well get you into last year for many horses. Also, its not such a bad thing to have a clunker averaged in there. Tells you there is some cyclic activity in there and the horse could well throw a clunker. This of course would just be the base. Would want to give or subtract points for such things as surface change, level change etc. I just haven't convinced myself yet of the best way to do it, but as a simple first pass, an average I think works just fine.

[cj]

Its a decent idea for a rough starting point, but what do you do when a horse has one start and its higher than all the other horses have ever run? Is he 1-20? How about turf to dirt, sprint to route, etc.? Do you want a line on a horse who has 8 of 10 races that were run two years ago? It happens. What about first time starters? Trainer changes? First Lasix?

I really do like the idea, but you are going to get some crazy looking odds lines. Email me if you want to talk in private, I've tried many, many things in the odds line making process!

[chickenhead]

OTM -- the thing I don't like about averages is that the one really bad race or one really good race can "make up for" all the others. Maybe an average rank of the last three would be better, I don't know if that makes sense or not, but say for instance one really bad race in an average would make all of the other races look bad too. But with a rank it has less influence, a horse can't finish worse than last. I think both ways (rank or average) have their good points and bad points.

CJ-- I sent you an email, but I think you hit the nail on the head so far as the biggest problem, which would be different number of PP lines. I was thinking mainly of older claimers, but definately this approach would have to be limited to older horses. The other factors (recency, changes, etc.) I think could be attacked as adjustments to this line.

At the end of the day I don't know of how much use it would be. Might be just as well to start with an evenly distributed odds line and go from there. I don't know how people do it.

[Charlie Judge]

I usually use the median of the past 3 starts.


As pointed out above an outlier can pull an average high or low, but the median is the value that stands in the middle. If the last start of the horse is abnormally fast or slow, the horse will most likely regress back to the median, rather than continue to improve or worsen, unless you are dealing with a developing 2 to 3 year old or an older hurt horse.

The speed figure series 74 77 93 is more likely to go to 77 next race out than to 95. The average of this series is 81 but the median is 77 .

When a horse has a bad or exceptionally good race, he is more likely to run back to his median number.

caj

[bettheoverlay]

I dislike medians and averages, played around with lots of varieties for lots of years. I'm pretty much a last race handicapper nowadays, believing that current condition trumps all numbers and theories. I print out the last 4 races only to see if a form cycle is involved. I have fallen in love with simplicity and set bettable odds on no more than 3 horses, usually only 1.

[chickenhead]

I think the best I can say in defense of what I outlined is it is a kind of long term picture. It does weigh all races equal, it doesn't account for recency, but, since everyone seems to be focused on last 3 races (for good reason) it might be a good thing to take back class into account, at least for a starting point.

I doubt I'll get any bites, but if any of you wizards want to try out what I outlined, I'd be curious to see the odds line it comes up with for a few races. With my skills it would take me a month at this point to do one race.

[sjk]

Chickenhead,

One would need to create a few tricks to complete the calculation as you have outlined it, and since it is pretty easy to construct examples (as per the posts above) where it would lead to nonsensical results, it is hard to work up the energy to do so.

I do believe that what you have thought of is a good starting point for making a line. You can certainly refine your process to develop something that does work.

[chickenhead]

thank you for the feedback.