* *

*Click to enlarge image.*

*[originally posted on PRECOG10 mailing list Wed 2011-06-29]*

Hello,

I have decided that my intent in doing an ARV sessions from now on will be:

*“My intent is to view the photosite represented by the photo associated with the horse with the maximum return-on-investment (ROI) for a win bet on this race.”*

Until now, my intent was to ARV the winner of the race, usually (exception: NHC contests – highest win/place total).

But I got to thinking, what does the concept “winner of the race” mean? Before the race is run, each horse has a *probability* of winning, between 0 and 1. For no starter at any time before the race finish becomes official is that probability either exactly 0 or 1, it is always somewhere in between.

That is, the “winner of the race” is not determined, it is not somewhere out there in the future, for me to somehow discover using ARV, as I have implicitly assumed all along. There are only *probabilities*, of each horse being the winner, between 0 and 1.

Now the public votes with its wallet, to determine its probabilities – the more money bet on the horse, the higher its win probability – as based on logical analysis of past performance information, for the most part. In the long run, this has been shown to be accurate – the probability of a given range of odds (e.g. 2-1 to 3-1, 3-1, 4-1, and so on) decreases with the higher odds ranges uniformly – and it must be this way, as tote betting is a self-correcting process.

But I have ARV, something almost none of them do. **My assumption is that ARV allows me to be more accurate about determining win probability.**

But it is not enough to know the horse that is most likely (has the highest probability of all starters) to win. The horse must also have high enough odds to justify a bet on it – since the probability is always less than 1, there is some probability it will lose, and if the odds are low, the wins may not make up for the losses over time.

It is best to explain this by example. Suppose I have somehow determined that my Targ CRs of 5 and above result in wins 30% of the time. Then for any given race with a Targ CR 5 or up horse, I assign that horse a .30 probability of winning the race. That is, if the race is run 100 times, I expect my ARV horse to win 30 times.

Now suppose the odds on these 100 starters average 2-1. If I bet $1 on each of the 100 races, I will cash 30 of them, with a net profit of $2 each or $60 on those races, but I will lose $70 on the other 70 races, so that my net is a loss of $10. ROI is defined as (net money) / (risked money), so is -.10 (per dollar bet).

But if the odds on these starters averaged 3-1: Net profit on winning races is now $90 on the 30 races, while I lose the same $70, so net overall is a $20 profit. ROI here is +.20.

**Therefore knowing which horse has the highest win probability is not enough to win money. The post time odds must also be factored in.**

Now, I could embark on lengthy data analysis to determine win probabilities for Targ CRs and other ARV parameters (like my HCFs and GFs), and I may do this at some point. But for now, I feel that it is sufficient for me to state my intent as above – that I ARV the horse with the maximum ROI potential – and the universe will do the rest.

It did tonight ðŸ˜‰Â * [see sketch and photosite above]* — (too bad I had $0 cash on him ðŸ˜‰

I feel strongly that my intent has very much to do with my success, and now that I have clarified it, I expect very good, highly profitable results.

— Tom

PS. Here is an easy way to calculate ROI per dollar bet if you know the win probability w and the odds o to 1:

ROI = w * o – (1-w)

or

ROI = w * (1+o) – 1

PPS. By the way, I claim that the only way to compare horse racing ARV results with binary or 1ARV results is by ROI – meaning one should really include the odds you are getting for your (e.g.) baseball picks in the statistics – usually around 10 to 11 on Over/Under (but it varies). For the current 1ARV Public record of .5862 wins (as of tonight’s game – which I got a hit on too ðŸ˜‰ in 29 live predictions, assuming the average odds really are 10/11,

ROI = .5862 * (10/11) – .4138 = +.12

12% profit is *excellent*!

I have calculated my own over the last 3 months: avg odds 8.57 for 10 winners in 61 live predictions (16.4%):

ROI = .164*8.57 – .836 = +.57

57% profit is very nice!

Lincoln might want to comment, he keeps his own stats.

Horse ARV ROIs for all will be up soon ðŸ˜‰

Tom