recency rating

[Rick]

I was looking through some of my old racing newsletters and discovered an article about a recency rating that combines races and workouts. Since workouts seem to be more important these days in determining fitness, I thought I'd check it out. The author calculates a rating for the last 3 and last 4 "moves" (either races or workouts) and takes the best of the two. The calculation multiplies race furlongs by 1.3 and adds workout furlongs before dividing by days. I tried this for 3 "moves" for a small sample and got some results that were somewhat positive (about 0.86 return on $1). Since most traditional recency guidlines seem to be totally useless these days, I thought I'd mention this since someone may want to test it over a larger sample.

Since I haven't posted much lately and some have asked why I also thought I'd explain here. I've been doing a lot of research lately and seem to have improved my bottom line by using LESS factors. Since I've only tested about 1200 races so far I hesitate to declare victory, but so far the method produces 30.74% wins and 2.73% profit playing EVERY race. And playing only about 2 races per card results in 25% wins and 33% profit (eliminating ML favorites and using a form filter).

Also, I've been doing some research on the stock market since it looks like we may be due soon for a very big move upward. There are some very good research papers available on the internet these days and the quality of the research is much better than was available even a few years ago. If anyone's interested in that subject, I'd be more than happy to discuss it in the appropriate forum.

[so.cal.fan]

I'm interested, Rick......post us some stock tips in Off Topic!
Your research angle sounds interesting. I'm taking a break from betting to freshen up.........doing a lot of reading......seems most of it is here on board.

[turfspec]

Rick,

In the thread "The last piece...FORM" started by Larry Hamilton on 8/22, I posted a simple method I've used for about 10yrs. to qualify lay-off horses on form.I derived it from some research I did as a qualifier NOT as a handicapping method, but if you get chance to check it out, I would be interested in any feedback you might have as to how it compares to what you posted here. I haven't kept track of its' efficiency lately but it seems to be performing fine.

I agree about the fewer factors resulting in higher ROI & also avg. mutuel. In my own experience I went from 12-15 factors 15 yrs. ago to 5-7 factors and for the last two years am down to two factors I make numbers for - Class & Pace. I've found a direct correlation between # of factors and a rise in avg. mutuel and ROI. My win % has dropped by about 3 percentage pts. but my ROI has increased dramatically. I can live with the trade-off. Best guess is the more factors you use the closer you get to the crowd.

Rob

[Rick]

turfspec,

That's exactly what I'm finding. I'm down to two main factors now and a third used as a filter. It's possible that I may incorporate the third factor in the main rating, but that will require a lot of testing and it's not always possible to just use a simple multiplier for some factors.

Here's my general philosophy about form. I see three ways to measure form. The first is based on the horse's performance, which is the way most people look at it. The second is by looking at the odds, which may incoporate other information beyond the past performances. And the third, which I like the best, is by trying to discern what the connections think of the horse's chances based on where they place the horse. This alternative usually produces the best results money-wise. Of course, the catch is that it's also the hardest to interpret, since a full analysis would involve subjective factors. But, even using a fixed set of rules and realizing that it may be imperfect can bring some major rewards. That's my two cents anyway.


socal,

I don't have any specific stock tips, but I'll try to get together some of the ideas and resources that I've discovered and post it in the Off Topic section. If you have some specific questions though, you can always contact me directly.

[Rick]

turfspec,

I looked at your post under "Form" and it seems like a good way of evaluating layoff horses. But, what I'd really like is something that rates all horses so that's why I'm interested in also using races as part of the equation.

Another thing I've looked at with workouts involved finding the horse with the most recent 4th back workout. At some tracks, mostly east coast, this can be a nice little longshot angle. At some other tracks, mostly west coast, it's awful. That's one of the big problems with workout data. It seems to have totally different meanings in different locations. You guys with the 100 trillion race databases covering every track in the world won't find anything significant, but it's there on an individual track basis. Anyway, in some places, 4th back workout (or maybe 3rd back workout) can reveal some layoff winners and first time starters that you might have otherwise overlooked. Those who have a cluster of recent workouts usually don't have a recent race and I think most would agree that those 10 day comebackers do worse than average these days moneywise.

[Hosshead]

Rick, You mean you can get 2 plays per track/per day with a 33%Roi ?...... So you can play 10 tracks a day, $100 bets, and make $660/day? If you only played 20 days a month, that's still $158,400/year. Even $50 bets would haul in $79,200/yr !!

If the avg. odds is say around 5/1, I wonder what the length of your losing streaks are? Sounds like you've been working hard on this, and you're finally seeing the light at the end of the tunnel. (and it's not a train). Good Luck with it!-- Hoss

[Rick]

hosshead,

I'm only playing one track right now and I doubt if those results would be achievable everywhere. I wouldn't want to play 10 tracks a day anyway. I've already got an adequate income from other sources without having to work anymore and want to spend some time on other interests. In the past I've usually played 2-3 tracks per day and found it to be a tolerable level though, so maybe I'll look for a couple of other tracks to play later on.

It's really important to look for tracks that have qualities that are consistent with your style of play. Some will look for early speed tracks for example. I prefer to play consistent tracks with a long season. The least desirable thing, at least in my experience, is to play all of the tracks in your local area only. Take Southern California and New York for example. The three major tracks in both areas are so different from one another that you need at least three different handicapping methods. I say "at least" because they also frequently dig up the tracks during the off season and change their pace characteristics so that what worked last year may not work as well this year. It takes a couple of weeks at the start of each meet to identify the changes.

But that's just my opinion. Maybe some who are really good at identifying track bias would prefer to operate in a changing environment. There are lots of books that say how easy that is to do. In about 5 years of research I never found a really reliable method for doing that but I may be missing something obvious. My philosophy is to use whatever works for me regardless of what somebody else says. That gets me in a lot of trouble sometimes with the pace gurus because I can never verify their claims when I test them. My sincere apologies to them though if they can do something that I've never been able to do. I never claimed to be a handicapping genius though.

[Rick]

Hosshead,

I should also point out that I actually play about five races a day rather than just two. Even though the other three don't have a high ROI, they still contribute something to my bottom line since I'm also getting a 3% rebate. Also, I trust the larger sample more to hold up in the future.

Extrapolating ROI from a small sample at one track to all tracks can be hazardous to your wealth no matter how enticing it may sound. Every additional track has to be checked individually and that takes a lot of time. In my experience, a given general approach to handicapping works well at about one third of the tracks overall. That's one reason why there's so much disagreement about what the best approach is. There isn't one that works best everywhere.

[Jaguar]

Rick, looked at recency- among other factors- over the years and found that by the late 80's the old recency criteria faded out, due no doubt to the- by then- widespread use of the hypodermic syringe.

Workout stats were also swallowed up by the chemists' magic.

Trainer stats work great, though, using the right algorithms.

All The best,

Jaguar

[JustMissed]

Originally posted by Rick
Hosshead,

I should also point out that I actually play about five races a day rather than just two. Even though the other three don't have a high ROI, they still contribute something to my bottom line since I'm also getting a 3% rebate. Also, I trust the larger sample more to hold up in the future.

Extrapolating ROI from a small sample at one track to all tracks can be hazardous to your wealth no matter how enticing it may sound. Every additional track has to be checked individually and that takes a lot of time. In my experience, a given general approach to handicapping works well at about one third of the tracks overall. That's one reason why there's so much disagreement about what the best approach is. There isn't one that works best everywhere.

Rick, In this post you say you "play", in an earlier post you said you "test" as follows: "Since I've only tested about 1200 races so far I hesitate to declare victory, but so far the method produces 30.74% wins and 2.73% profit playing EVERY race. And playing only about 2 races per card results in 25% wins and 33% profit (eliminating ML favorites and using a form filter)."

If you are NOT getting 33% profit from actual play you probably should say so.

As you know there is a world of difference between "testing" and actually put your hard cash through the window.

JustMissed

[hurrikane]

Rick,
haven't seen you post in a while.

At one point you were running, I thought, 4 tracks, betting 75% of the races and getting about a 25% ROI. Did that not hold up over time?

Thanks, glad to see you're back.

[Rick]

hurricane,

Not exactly. I was playing 2-3 tracks, mostly 2 with about 4 plays at each per day. ROI was about +27% early on and later slipped to about 20% overall. I used a filter based on some jockey data that worked extremely well for about a year and then lost some significance for whatever reason. I'm sure you've had those things happen too. I got interested in doing some research into using multiple methods in order to play every race. While that produced some interesting insights, I discovered a way to improve on my original rating and spent most of my time doing that. I scaled back my betting to only one track while I was doing that and got to like having the extra time.


JustMissed,

Sorry for the confusion. I played my original method until I had a reasonable amount of confidence in the newer version, then started playing that instead. So, to be more specific, the "testing" was for about 1000 races and the "playing" for everything after that. But that's including every race (I should say every dirt race since I have a different method for turf races) and I only actually play about 53% of those. That would drop to only about 22% for those high ROI races that I mentioned (2 races a day). So, I was "playing" about 4 races a day using the old method until a couple of weeks ago and now I'm "playing" 5 races a day with the new method, and I'm still "testing" the 2 race per card high ROI alternative. IF it's still as good a few months from now, I may play only the two races and look for other tracks to expand my action. OR, I might find a way to incorporate the filter into my rating and improve the overall results which would probably result in 3-4 high ROI plays per card.


Jaguar,

Yeah, I'm afraid that there's not much value in any kind of recency rating these days. I'm not good at using trainer data but I'm sure it's better. I have had some good results looking at where the connections place a horse though. There seem to be some optimal patterns for class changes relative to recent performances that are more likely to occur when the horse is in good form. It's not a very exact science though, unfortunately.

[Rick]

One more comment on the "testing" vs "playing" thing. I've tested and played thousands (probably tens of thousands) of races and out-of-sample testing wasn't much more effective at predicting what the future results would be. The reason is that I don't test hundreds or thousands of different ideas and find one that's significant out of those. The ideas that I test are based on previous research and experience about the kinds of things that usually work. They're not based on data mining so they tend to hold up much better in further testing. In fact, I've gotten a lot of ideas about what might improve ROI from research that others have done here. There are some real research gems here if you know what to look for. Most people look at research on single factors and get discouraged if they're not profitable. That's really missing the boat in my opinion. Combine two factors that lose 5-10% each individually and stay away from morning line favorites and you should have no problem making a decent profit. That's my message. It's easier than most people think and you don't have to be a handicapping genius. Please don't think that I'm shooting for the high ROI bragging rights. We already have another guy who's famous for that.

[Rick]

Sorry to be long-winded here but I'll further elaborate. My current method was inspired by a 2000 race sample from about 5 years ago that lost about 5%. I tested a 400 race sample from Santa Anita from a couple of years ago with one of the factors modified to be measured over a different time period and found surprisingly good results. So, when I tested the sample for this year, it was really already an out-of-sample test. That's why I have a lot of confidence in the method. It wasn't just pulled out of a hat. And, it was also very similar to the successful three factor model that I'd been using before. I'm belaboring the point here because it's really important to use what you've learned in the past, even though it might not have been as spectacular as you'd hoped. A little change here or there and bada boom, bada bing!

Anyway, I'm going camping tomorrow so I'll have zero plays per day over the next 3 days. Not very ambitious I guess.

[JustMissed]

Thanks for your reply. Best of luck to you.

JustMissed

[Rick]

A little philosophy about profitable gambling before I leave. It seems that there are at least two (maybe more) reasons that people put so much effort into finding a way to win at games like horse racing (or blackjack, poker, etc.). One is that it is satisfying to do something really well, that is, much better than most people can. Even someone who breaks even at horse racing can consider themselves a rare individual comparatively speaking. And to actually make a living at it is a status symbol of the highest order. In fact, anyone who has any income from other sources is not considered a "true" professional even though they may be able to produce a much higher income from doing other things. It takes real guts to live on a gambling income alone and people admire that.

The other reason is to escape the rat race and be truly independent of the "system". That's a personal triumph and doesn't really depend on the number of dollars earned but more on whether it's enough so that you can escape from being a lifelong wage slave. That was my objective for many years and, ironically, I was lucky enough to escape the rat race before I'd actually achieved the level where I could play at a full-time professional level. So, I got what I really wanted a different way than what I'd been working so hard to achieve. Maybe not really needing something makes you better at it because the next thing that happened is that I became consistently profitable.

So, whatever category you fit into, I wish all of you success. Just remember that if you choose the first path you'll be joining a group of people including billionares who never seem to be happy with what they've achieved. There's always somebody better and you'll start exaggerating your successes and hiding your failures in order to satisfy your ego. Get what you really want, not what you think will impress other people. End of sermon.

[PaceAdvantage]

Great posts Rick...thanks!

[turfspec]

Rick... excellent series of posts and sage advice. It all rings true.

The idea of creating a form rating for all horses in a race incorporating the expanded parameters of odds and trainer placement seems to me to be a daunting task. I can think of at least a half dozen scenarios, off the top of my head, concerning trainer placement alone that would be extremely difficult to quantify and codify. If you are able to do it, I'm sure it would afford you a significant edge. Probably wouldn't be workable on a universal scale but might work "locally" as you suggest in your earlier posts.

Best of luck with your research.

Rob

[hurrikane]

boy Rick,

you take off for a while and come back charging.

Great posts.

Yes, been many times in a system that cools off. I have one (2 params and 1 filter) that over the last 4 yrs alternates profitable yrs..one yr it's 1.45 the next it's flat. Can't figure out why. There is nothing associated with odd or even in the params :D.

Enjoy the camping trip and thanks for sharing your insight.