wouldn\'t be a great throwdown if someone could do a simple test of the 2 services side by side? i understand it was done many years ago, but we are light years away from that now.
just simply take the top three number horses and see the win percentage?
i somehow doubt that either service would agree to it.
High Roller,
Better yet what do you think the percentages are that people actually make money using them all together.
VERY VERY INTERESTING QUESTION...............
My honest answer is NO! The game is very difficult, no matter what data you have. To play horses full time, players need a good rebate package, the higher the better. I have made my living betting greyhounds for the last 18 years with only about 15% of my wages going towards the ponies. I can tell you if I had to play horses full time I\'m not sure I would make it even with the 13 to 14% I get on rebates.
HR-- Jimbo proposed doing that last year on the Ragozin with an objective study that took the human element out of it as much as possible, and was treated like he was doing something dirty.
I don\'t see how using top 3 numbers would yield accurate results. Part of playing sheets is recognizing certain patterns, using your own developed interpretations. Your method would not take that into account. I\'ll give an example using a race that I had both sets of sheets on for: In one case, the relative #s after adjusting for sheet differences (since Thorograph #s are lower than Ragozin #s) on the winner showed an explosive pattern entering a race on one product, which I\'ll call A. On the other product (B), there was greater improvement in the horses\'s last race, but it didn\'t show the upcoming explosive move, since the number was significantly better in the earlier race that pinpointed the move on A sheets. In this case, using my interpretations, Product A wins this race, while product B would have yielded a higher #, but may not have pointed the user to the winner. Other times B would beat A in a similar race-once again subject to user interpretation.
About the only way you could have an acid test would be for Jerry to square off mano a mano with one of the Lens, each giving a selection before the races for a statically significant # of races. That\'s about the only real way to test the products, having the head guy from each interpreting their own sheets.
i guess since you never worked on wall st. you don\'t understand the term, \"acid test\", its just a quick simple look.
let\'s say the results might be 63% with one service 75% with the other etc. it gives you some idea, what you are saying is up to each unique individual.
let\'s also be honest both products are good, its just that jerry is like the turk in the godfather movie gunning down the old don.
sheets in the minds of many are a ledgend, jerry will always be the upstart, i feel with in 5 years jerry will leave them in the dust, the old crowd is withering away and not using modern methods either in theory or technology.
Yep, never knew acid test was short look.
First of all, as followers of this site know, I\'ve publicly challenged Len to a handicapping contest about 10 times, including challenging both him and Andy from the stage at the DRF Expo, with the contest to be hosted by DRF on their website. Len has never responded.
As far as a purely mechanical test goes, of course there are going to be individual situations where handicappers would not just use number power (and also ones where different handicappers using the SAME data would disagree, it happens all the time). But we all agree that faster is better, and that ALL THINGS BEING EQUAL, the faster horse will win. If you get a large enough sampling, the factors you describe will even out-- there will be roughly the same number of them favoring one side or another. And doing it mechanically eliminates the human element-- one person can be a better handicapper than the other, or hot, or bring inside information to the table, etc.
So bottom line, if you did the right kind of study over a large sampling, it would be indicative. The one Jimbo proposed-- based on an idea I had put forward-- was to take the last 3 of each horse, throw out the worst one, and average the other two. That would give a general picture of how fast each horse is. There are lots of ways you could then do the study (there are some statistics guys out there, I\'m sure they have ideas), and we all might learn something. Like whether a \"right\' figure is right.
From my point of view, I\'d be going up against an outfit that thinks \"right\" means Borrego ran a better figure running 10th in the Classic than winning the Gold Cup by a block. I like my chances.
Thats exactly why they don\'t want to get involved , both of them know they have serious problems . During the past several years , TG players in the big Vegas contest have been all over it might be an kind of indicator or bellweather sorts ...
Jerry,
Your methodology of just throwing out the worst case of last three (ie using best two of last three in effect) is flawed greatly statistically way too general. That would say any data prior to last three races is meaningless and worthless. You really should use Std Dev of last 10 then 5 then last 3 to see trend of horse development. This takes out any bias/anomoly that may contaminate last 3 race data (only).
Anyone using data knows that a sample of best 2 of last three only is useless, especially if the distnces and surfaces change. Take BGC as an example, you had to trust that you had to ignore the last race. Most Handicappers don\'t or can\'t. I know a few good handicappers who did and cashed big.
Using Std Dev you can find the range that any horse can figure or project to run in High or low (best case or worst case) Worked great for me in Oct Stakes including BC races. requires that the TG data be accurate accross all tracks and sufaces. I had issues at the time with your Cali figs.
Thats all I am willing to say at this point.
NC Tony
Thats a good point - I\'m having issues with how those races even fit into an overall context of a horse\'s real ability .
Although if it will give players an egde , I\'m all for it .
I think the math and stat guys are getting off point.There is only one question to be addressed. What data is consistently more accurate,TG or RAGS.
TG and Rags are now very far apart too often.The former scale of comparison between both products,3-4 points, has almost VANISHED. Someone is getting it wrong too often,and not even close on occcasion.I have seen differences of up to 10 lengths in a game which often comes down to a difference of a nose on the wire.
You do not have to be a figure maker to take exception to a figure which is obviously controversial, defies common sense, and stands alone from all other credible sources.There are MANY people with strong and credible racing opinions around.On more than one occasion Andy Beyer has admitted to the scientific inexactness of making figs.
As much as anyone, I have sought answers which are surely not apparent from some handicapping contest with a few stabbers using both products.Very tough issue, this one.
To me there are compelling questions for both camps,for example RAGS stated:
Who said that horses \"must\" pair so often and what science is behind that?
JB,I have seen many trainers, including hall of famers,often flabbergasted by the inconsistency of a large population of their stable. Just the opposite of what is on display daily when viewing TG.How can that be?
From me, a TG user to RAGS:
It is difficult that you have not identified that tracks can substantially change speeds during the course of the card. How can you produce accurate figs without factoring these obvious changes in? Do you feel that averaging obviates that issue? Why?
Mike
If the Data \"is accurate\" the method I propose will point it out clearly.
NC Tony
NC Tony,
I agree but in the absence of independent \"audit\", how can you state with reasonable certainty what is accurate.
Mike
Miff-- that\'s the problem, all right. By the way, you should try this some time-- go through a card, and look at the last 5 races on every horse, and see what the percentage was of pairs (5th back to 4th back, 4th back to 3rd back, etc.). I think you will find it is much smaller than you think, all other issues aside.
Tony-- like I said, I\'m sure statistical guys have better ideas for testing than I do-- I would hope so, since they teach college courses in statistics.
But I do think the study I described would have some relevance (and yes, we would not count races run on other surfaces, and make other adjustments). But in general, recent races are more important than those further back, and the other factors should wash out equally for both sides in the long run.
Bottom line, if a reasonably accurate study can be done I would love to see it. We would kill Ragozin, and I\'m convinced that Beyer would beat them, even without ground and weight. Miff is right about the differences getting bigger, and it\'s not because I\'ve changed anything.
JB,
I did not look at the pair thing the way you suggest, but I will. What I did do for app six months was to compare pairs on TG,Rags,Beyer(adjusted)for the same runners.
The results showed the most pairs were on TG with Beyer(adjusted) second and Rags third.One interesting thing I found was that Beyer unadjusted can appear up to 7 lenghts slower than TG but equal after adjustment(mostly happened in TG buried figs).Incidentally, I believe that naked Beyers can be very misleading on occasion.
What this all means is not very clear right now but the more I probe, the closer I feel I am coming to solving this from my viewpoint.
Mike
The comparison of the products is not going to happen, at least not with the consent of Len and the Rags folks. I posted on their web site after the Saratoga meet last year that I would like to do the comparison, but was looking for a Rags user to \"partner with\". I was betting the horses every day back then, and wasn\'t going to purchase two sets of sheets. Plus, with a regular t-graph user and a regular rags user doing the study, there would be less room for skewing the numbers. I, for one, was open to any conclusion that the study produced. I think I am using the right numbers, but if the study proved otherwise, then so be it. Instead of generating any intellectual debate at all on the Rags board, Len took a potshot at my methodology, without proposing any alternative methodology of his own. Some other posters over there gave some unintelligent responses as well, but nobody stepped up and was interested. The Rags marketing strategy seems to be to make believe that Tgraph doesn\'t exist and try to convince their customer base of this as well.
Miff,
What do you mean (so I understand) when you say adjusted Beyer\'s (ie you factor in ground loss somehow?)
NC Tony
NC Tony,
Yes, ground and weight shifts from start to start. It\'s a fairly accurate comparison 90% of the time, the other 10% they disagree by quite a bit.
Mike