Cramer Class Drop Jockey Switch

[oller21]

Hello everybody. I've been browsing this board for some time now and you all appear to be very knowledgeable cappers. I hope to be able to make some money at this difficult game and any help you could provide would be appreciated. I wanted to know if any of you were familiar with Mark Cramer's Class Drop Jockey switch angle and do you know if it's profitable?

[headhawg]

tryingtowin,

If what you mean by "profitable" is playing it blindly in all situations it probably isn't although I don't have any research stats to support that opinion. I use the angle as part of my overall handicapping process (as well as the Class Drop/Track Switch angle).

However, the horse still has to meet the requirements on figs, form, etc, but a switch to a leading jock OR a jock that has won with the horse before plus a drop in class is a sign of postive trainer intent.

HH

[BMeadow]

From a survey printed in the February 2001 issue of Meadow's Racing Monthly, involving 197,194 horses who were switching to a higher-rated jockey (using a rating method that involved jockeys' win and in-the-money percentages over the past 365 days) and dropping in class (defined as dropping at least 20% in purse value from last race):

Win percentage: 14.3%
ROI: 0.78 (loss of 22 cents per dollar)

Just one more logical idea that doesn't work.

[GameTheory]

I know Cramer has criticized Meadow's research methods. You can read an example of that here:

http://www.gambleinparadise.com/articlex383.html

I must say I agree with him. Meadow seems to test ideas out of the context they were intended for or with parameters of his own invention, and then declare flatly that "this doesn't work". I don't know the specifics of the test for the class drop/jockey switch angle, but there are many obvious scenarios where you would want to exclude certain races from the test. (Like, was the original jockey even available? Is this a shipper? Is this horse coming back from a long layoff? Is "higher-rated" jockey mean *any* higher rated jockey, even if they are rated almost equally? How big a drop are we talking about?)

Tests like this will have a tendency to get you to throw away good ideas that signal positive trainer intent in their proper and logical context. For instance, a positive jockey switch could actually be to a lower rated jockey if that jockey happens to be one that this particular trainer likes to give live mounts to.

Also, as been pointed out here before, Barry Meadow criticizing other authors in his newsletter while he is selling his own product is a bit of a conflict of interest. Since the newsletter is defunct, that point is now moot, but something to keep in mind when looking at the old issues...

[BMeadow]

I agree with GameTheory that all handicapping involves context. For instance, sometimes fast workouts may be a positive (e.g., the horse is training strongly), but other times they're a negative (the trainer always works his horses fast, and has entered his $200,000 yearling for an $8,000 claiming price).

However, if I state that some theory yields positive results, I'm going to have to do a lot better than testing 200 races, or giving some examples. Even if I'm a nice man who sincerely believes in what I'm writing.

I'd love to have a dollar for every player who read some idea in a handicapping book, played real money on it, then wound up with a very light wallet as a result.

Handicapping writers generally have shown almost a disdain for real statistical research. I don't see where pointing out their errors--in methodology, in interpretation, in mathematics--is a bad thing.

In any case, the best handicappers know when to use a particular angle and when to ignore it, and why another factor is crucial in one case but meaningless in another.

[GameTheory]

What I find worse than not attempting "real statistical research" is attempting it and doing it wrong. Garbage in, garbage out.

I am not going to argue with you because I'm ignorant of what you've published in your newsletter for the most part. But the snippets I've seen cited or that people have shown me seem to me to be misguided, as if your focus is just on discrediting ideas/people rather than trying to find out the truth of what handicapping ideas have value (in some context) and which don't. Your studies seem to have far too wide a frame of reference which naturally leads to mediocre results no matter what the method/angle.

I'm not just picking on you -- there is flawed research everywhere in all sorts of fields, much of it done by trained statisticians. It's a tricky business...

[trying2win]

Question for Barry,

In some systems, in addition to the usual rules, the author will specific an odds range to bet...e.g:

--Don't bet on a horse less than 4-1 or more than 15-1. Was an odds range rule taken into effect when your researchers did a test for a system?


--Another comment: Actually I've seen some commercial systems that were not that bad, but with some modifications to some of the rules, adding an odds range (if there wasn't one) and applying the system to certain racetracks, some positive results were attained.

-- With a lot of research and testing, I've even developed some of my own methods or angles with some ideas gathered here at the Pace Advantage site (I prefer using my own methods and angles). For example, I've found that some of the self-developed angles work well at a track like Mountaineer Park, but are losers at several other tracks. Over time, I've become a believer in the use of track-specific methods or angles.

T2W

[BMeadow]

<<In some systems, in addition to the usual rules, the author will specific an odds range to bet...e.g:

--Don't bet on a horse less than 4-1 or more than 15-1. Was an odds range rule taken into effect when your researchers did a test for a system?

--Another comment: Actually I've seen some commercial systems that were not that bad, but with some modifications to some of the rules, adding an odds range (if there wasn't one) and applying the system to certain racetracks, some positive results were attained.>>

In many of our tests, we did specify odds ranges. A complete list of every system tested is in our Index To Articles (Real System Workouts) under Meadow's Racing Monthly on the website trpublishing.com.

Sometimes a system might contain a useful idea. However, you have to be careful about checking it on 10 racetracks, finding it worked on 1, and saying it's a good system for that track--just as you might test it for each day of the week and found it was profitable on Fridays only.

My advice: be careful about jumping to conclusions.

On another subject, I can't very well respond to Game Theory's comment because he says he "is ignorant of what you've published in your newsletter." I can't review a movie if I haven't seen it, no matter who tells me what about it.

[alysheba88]

How people can find problem with Barry's work against some of the "huckters" out there is beyond me.

The hucksters peddle their "Holy Grail" systems because they know there is a segment of the public looking for and believing in them

[Light]

If there is a profitable system that was printed in a book, most handicappers would have heard about it by now. And if there was a profitable system that did catch on with handicappers it would surely turn into a losing ROI with it's popularity.

The only times I've benefitted from these nuggets of gold is when I elaborated on a system/angle and put my own personal spin on it. Whenever I read the Bris handicapping champions methods they allways say "I use this" or "I use that".So what.So does everyone else. What they don't tell you is HOW they use it.

[The Skeptic]

A little off topic but directed towards Barry Meadow Research.

I recently exchanged some communication with a gentleman who regularily used to post on this board. He sells a good product re: Dr Z if that offers any clues . I don't want to get off topic whether Z works or not anymore (it doesn't just for the record).

He mentioned his software works because of the "Favorite Longshot Bias". Red flags red flags. He then mentioned Barry's research pertaining to Fav/LS bias being flawed when defending his Z system software but offered no substance. I also seem to recall others critiquing it a number of years ago. Now Barry...just for the record I do respect your work! He said the research that was done was flawed. I told it's simple....there either is or isn't a Favorite Longshot Bias? If you're selling this product and defending it with this "F/LS bias" is there or is there not a true favorite Longshot bias? He's under the firm belief it's there and I'm inclined to trust the Meadow Research. At the same time as perpetrated in Betting at the Racetrack by Dr Z (whose orig work is also respected) it claimed a profit on extreme favorites and it's obvious there isn't.

Without getting into a Z debate because there isn't one....Barry would you mind elaborating on the following in context of your past published research:
1) the results of your Fav/LS Bias Research
2) the controversary that had surrounded it
3) comments as a whole

Thank you.

[BMeadow]

Results of 200,000 races published in our January 2000 issue of Meadow's Racing Monthly:

Betting all favorites--ROI was 0.81
Betting all 2nd choices--ROI was 0.82
Betting all 3rd choices--ROI was 0.80
Betting all 4th choices--ROI was 0.79
Betting all 5th choices--ROI was 0.79
Betting all 6th choices--ROI was 0.78

It was only on 8th and higher choices (23-1 and up) where longshots were overbet. Extremem favorites (2-5 and below) are slightly underbet, but still do not exceed an 0.86 ROI. Thus, the old favorites-longshot bias (which WAS true at the time I did Money Secrets At The Racetrack, based on surveys of more than 50,000 favorites) is no longer true.

As for the Dr. Z system, it was based on something called the Harville Formulas, which stated that if a horse went off at odds of so-and-so, it should finish second or third a certain number of times based on those win odds. Sounds good--but in real life it didn't pan out.

To test it, I went through 1,000 races with an odds-on favorite (the type of horse most likely to be a Dr. Z selection and totaled up how often these horses finished second and/or third. Groups were separated into 4-5 and up, 3-5 to 4-5, etc. Every group showed that the Harville Formulas vastly overestimated how often these horses would actually finish in the money. The differences ranged from 3.5% short on 4-5 shots to place to 11.2% short on 1-2 shots to place.

Simply put, while the theory was good, it didn't work in practice--mainly due to the assumption that all 3-5 shots are alike, which they're not. Some run the same Beyer every time, are likely to repeat it, and pay little to place and show as a result. Others are win-or-out type who if they don't win, get nothing--and thus are not bet so heavily to place and show, and do not finish in the money as often as their win odds might indicate.

There is no "controversy" over the Dr. Z method. It simply doesn't work.

[Charlie Judge]

I looked at all races in the past 4 years and calculated the win, place and show probabilities for all odds ranges. A pdf of some of the results is at this link:

http://track-judge.com/itmodds.pdf

If I can find the Harville formulas, I will post a comparison.

CAJ

[BMeadow]

Harville Place Formula: (assumes 17.1% takeout)
0.319 + 0.559 (Wi / W)
-----------
(Pi / P)

Harville Show Formula (assumes 17.1% takeout)
0.543 + 0.369 (Wi/W)
-------
(Si / S)

Wi/W, Pi/P, and Si/S represent the ratios between the individual horse's best in a particular pool vs. the total bet in those pools.

[BMeadow]

Due to formatting problems, though this was correct in my screen, it didn't come out correctly on the Pace Advantage page.

The denominators should be Pi/P and Si/S, in the formula, to the numerator Wi + W in each case.

So place is 0.319 + 0.559 x the ratio
And show is 0.54 + 0.369 x the ratio

[Dick Schmidt]

Whoa! A system from the past! When I first met Mark Cramer back in the early 80's he was trying to get some of his students from the handicapping class he was teaching at Los Angeles City College to spend a Saturday researching this angle. I made one of the most momentous decisions of my life when I stuck up my hand and said I would help. Of course I didn't know it at the time.

We met at Mark's place and started wading through a pile of Racing Forms. Like the experiment run by Barry, we just looked for class drops and jockey improvements based on a list Mark provided. It was an earlier time, and the system worked better than Barry found, but it still showed a loss.

Working alone in a corner, I added a column to the worksheet we were using: I noted the name of the trainer involved. After a while I found some that had no pattern, others that showed a nice ROI. When I showed this to Mark, he was impressed enough to invite me to join a group that was trying to use a computer to handicap with. Led by Mark and Dick Mitchell, the group never made much money, but we learned a lot and my career in racing was launched.

If you want to try this old timer of a system, I suggest that you try the following:

1) Keep your records by trainer. Some guys are just so screwed up that there is little pattern to be discerned. They enter horses at any old class (some only want to work a couple days a week) and seem to hire the first jockey they can find without a mount. Others are signaling a real effort. In modern racing, with short fields, a trainer can almost always find a race at the "right" class for a horse, so this may well work even better today than it did back when trainers were forced to enter any race close to the right class if they wanted to get their horses on the track.

2) Remember that your analysis of the jockey talent ladder may differ from that of the trainer. I remember seeing trainers interviewed and saying they were really pleased to "move up" to a jockey I thought wasn't as good. There are a lot of reasons for jockey changes, not all of them well thought out or even planned.

3) Negative patterns can be as profitable as positive ones. Some trainers never drop a horse unless it's hurt and he wants to unload it. None the less, lots of players see the drop and the horse is well bet. This can leave a lot of money laying out there on the track.

Good luck,

Dick


Experience is a wonderful thing. It enables you to recognize a mistake when you make it again.

[GameTheory]

Originally posted by BMeadow
Results of 200,000 races published in our January 2000 issue of Meadow's Racing Monthly:

Betting all favorites--ROI was 0.81
Betting all 2nd choices--ROI was 0.82
Betting all 3rd choices--ROI was 0.80
Betting all 4th choices--ROI was 0.79
Betting all 5th choices--ROI was 0.79
Betting all 6th choices--ROI was 0.78

It was only on 8th and higher choices (23-1 and up) where longshots were overbet. Extremem favorites (2-5 and below) are slightly underbet, but still do not exceed an 0.86 ROI. Thus, the old favorites-longshot bias (which WAS true at the time I did Money Secrets At The Racetrack, based on surveys of more than 50,000 favorites) is no longer true.


Doesn't the above show a bias? With no bias, the returns at each slot should be equal (although breakage skews things a bit). And why leave out the 8th and above? If they are overbet, then there is a bias. With no bias, nothing is overbet when looking at odds alone.

What was the "old" favorite-longshot bias that is no longer true? There certainly still is some amount of bias. Taking a recent sample of 100,000 races (all tracks, all races), if we look at the returns from the lowest odds horse in each field compared to the highest, we get:

ROI of favorites (lowest odds in race): .851
ROI of anti-favorites (highest odds in race): .580
ROI of random selection in each race: .764

That's a favorite-longshot bias. All it means is that bettors could re-allocate some of their money from the longer odds horses to the shorter and the market would become more efficient (the longshots would become longer, and the favorites would become shorter).

[Charlie Judge]

Looking at the data, a log function fits pretty well. Since I do not have individual win, place, and show pool data, I really can't calculate the Harville data for back-fitting.

Some equations:

Win probability = 0.4113-0.3683*log10(odds)

WP (place)probability = 0.6101-0.4312*log10(odds)

WPS (show) probability = 0.7392-0.4021*log10(odds)

It is interesting to look at the graphs of 2nd place finish and 3rd place finish vs odds. At low odds, horses have low probabilities of finishing 2nd or 3rd, since the win probability is so high.

The graphs are at:

www.track-judge.com/itmlog.pdf

CAJ

[The Skeptic]

Barry,

Your thorough reply is greatly appreciated.

You must be aware of the people who criticized a flaw in your research? Would you mind expounding on where they may coming from.....

My comments:

a) Gametheory: I still see a bias
b) As my originally post said....it doesn't work
c) Z and Hausch did go a step further and claim extreme favorites could be wagered on profitably to wi. Maybe that was the controversy

I successfully employed the use of the Z System back from 1985 to about 1989. The only difference was I demanded a show payout of $2.80 (sometimes $2.60). The only problem was this was an era of non simulcasting and the opportunities were limited. I used to have my little Z Computer but it finally tanked out about a year ago due to lack of use. The reason it doesn't work in addition to Barry's analysis is if there is a perceived overlay in the place and show pool - the racetrack markets have become so effiecient that in the final minute of wagering that inefficiency is now efficient. That $2.60 Show wager is now $2.20. For the fun of it about 5 years ago I tested it using the BRIS tote for Harness Racing and the Dogs. I was demanding a 50% edge.....with 1 minute to post a horse would meet the 50% edge (I required 50% hoping if others were looking it'll still be 20% at post time) requirement and a guy at the track I suppose goes to the window and places his $50 show bet, I place mine and damnit it's an underlay - $2.10. A person's time is definatley better spent elsewhere looking for money.

My point to all this: Here is a person whose still marketing the software. My impression is he still believes it's a viable tool. He's basing it's defense on the favorite longshot bias. When I challenged him he claimed the Meadows research was flawed. Being I didn't have the research in front of my I simply stated "there either is or there isn't" - it's really that simple. It's clear that he should have provided a better defense.

[Light]

In Robert Rowe's book "How to win at Horseracing"(1990),he prints a study of Favorites through 4th choice to win. He further breaks down favorites as "natural"and "created". Natural are M/L fav's bet as fav's immediately and stay that way. Created are not the M/L fav's but become favorites through betting.He further breaks down the favorites as "high" odds favorites and "low and average" favorites.Different win percentage for these 2 categories. Overall created favorites showed a 23% loss and naturals showed only a 14% loss.He further breaks it up with Natural and created favorites win percentage by finish position of their
last start.

My point is Barry's study is a bit general cause there are many subsets in all those categories. For example in Rowe's book he finds betting "low and average"2nd choices who finished 3rd in their last start made a 10% profit.

[GameTheory]

I have no comment about the Dr. Z system, or if the favorite-longshot bias can be exploited in any way to make money. I was just showing that it exists, plain as day. I found it a little weird that Barry posted research showing a bias while saying the bias was gone.

Which is why I'm wondering what the "old" bias was, which must be what he is referring to.

[BMeadow]

<<Betting all favorites--ROI was 0.81
Betting all 2nd choices--ROI was 0.82
Betting all 3rd choices--ROI was 0.80
Betting all 4th choices--ROI was 0.79
Betting all 5th choices--ROI was 0.79
Betting all 6th choices--ROI was 0.78>>

If you want to be technical, yes, there is a tiny favorites-longshot bias--but it's so slight as to be worthless on a practical level. A different survey of 200,000 nationwide favorites (published in the April 2003 issue of Meadow's Racing Monthly, a different sample from the research published in 2000) showed the ROI of all favorites as 0.817.

Yes, the longest longshots (8th choices and up) are overbet, as I mentioned. Bet every 8-1 and you'll do much better than betting every 80-1 shot. So what?

ROI of favorites (lowest odds in race): .851 (this is 3-4 points higher than my two most recent surveys found)

ROI of anti-favorites (highest odds in race): .580 (yes, the 99-1 shots are way overbet...so?)

ROI of random selection in each race: .764 (which shows that any handicapping is better than no handicapping--or even blindly betting any contender is better than throwing a dart. again, so....?)


<<What was the "old" favorite-longshot bias that is no longer true?>>

A combined study of more than 50,000 favorites, published over several years by five researchers (some by hand, some by computer) that I cited in Money Secrets At The Racetrack showed that the public favorites lost only about 10%, rather than the 18%-19% of today. They won at the same rates as favorites today, but paid a lot more. Whether this is due to higher takeouts, better handicapping, better information, smaller field sizes, or whatever, the bias is gone (unless, of course, you're comparing betting all 2-5 shots with betting all 60-1 shots).

[GameTheory]

If I want to be technical? You were the one spouting off about the lack of "real statistical research". A longshot-favorite bias is defined thusly: The average returns go up as the probability of winning goes up. That is clearly the case.

Like I said, I wasn't claiming the bias was useful, just that it was there. And I've never seen anyone claim the bias was anything but weak -- the pools have always been fairly efficient. It is just a property of the market, more of interest to people doing econometric analysis of the behavior of markets in general than it is to handicappers, where what is important is the single race in front of them at the moment.

[highnote]

GT,
I couldn't have said it better myself.

I don't think I've ever seen a study that said one could make money betting the f/l bias.

Most studies I've seen that use large samples show that you can't make money betting extreme favorites.

Some authors have said that it might be possible to make money betting extreme favs, but they couldn't say for sure because their samples were too small.

Some studies suggest that the f/l bias does not exist in all jurisdictions. I think the bias probably changes from year to year, from place to place. It's a moving target. Why should it stay the same? Technology changes. The means people use to place their bets changes. The industry changes. The surface of the track changes. If everything else can change, why can't the bias?

Why should the dogs have the same f/l bias as the trotters?

I use the f/l bias to adjust the odds. An extreme favorite is probably a little better bet than the public thinks. And a longshot is not as good a bet as the public thinks. So I account for this with a f/l bias adjustment. I use the adjustment as one component of a betting strategy.

To say that the bias exists or it doesn't seems a bit black and white. I think it's bit more colorful.

[kbishop]

If I've learned anything in horse racing, it's that there are few absolutes. When Cramer wrote about his drop/switch angle, he did not place any qualifiers on it. My guess is that if the horse was 0-8 at the track and was coming off of a 150 day layoff, the play might be passed. There are dozens of other scenarios that would give pause to a blind play on a drop/switch angle. My guess is that meadows research on this did exactly that- total all D/S plays, even those those with supporting information that would have made the play less attractive or downright unplayable.

[Tom]

Cramer is one who goes for the spirit of the rule rather than the letter of it.

[bigga 1]

If i wanted to get involved in this game and understand the whole concept of picking winners and making money at the race track you guys would turn me off.All one has to explain is that this is a game of athletes and if the horse wasn't any good as a two or three year old he/she won't be any damn good as an older horse.This is an individual sport and you have to learn it the hard way and take your knocks like everyone else.This is about observation over time,making proper notes and good record keeping.This game is about the fitness of the athlete involved,pace of the race the bias and of course post position,that can also be a negative or positive depending on run style.Class drops have to be one of the biggest angles in horse racing but has to be accompanied by few things of which i'm not at liberty to say because you are the guys/gals i play against every day.

[LottaKash]

If i wanted to get involved in this game and understand the whole concept of picking winners and making money at the race track you guys would turn me off...

Class drops have to be one of the biggest angles in horse racing but has to be accompanied by few things of which i'm not at liberty to say because you are the guys/gals i play against every day.


Not sure what to make of this post of yours....

Is it, you are so smart, and we are so dumb, or should I say ignorant ?... Class drops are key....No chit Sherlock...

Otherwise, I am just not sure what the intent of your message was....:confused: ..

[pondman]

If I've learned anything in horse racing, it's that there are few absolutes. When Cramer wrote about his drop/switch angle, he did not place any qualifiers on it. My guess is that if the horse was 0-8 at the track and was coming off of a 150 day layoff, the play might be passed. There are dozens of other scenarios that would give pause to a blind play on a drop/switch angle. My guess is that meadows research on this did exactly that- total all D/S plays, even those those with supporting information that would have made the play less attractive or downright unplayable.

In 2013 I played 122 singles that had both a drop in class and a jockey change. However, at no time did I make these bets because of a jockey change. Nor did I rate the jockey as being of higher quality. Truthfully, I would have been happy seeing an unknown morning rider on each of these bets.

It's reckless to take a population and attempt to derive a sample. It's bad science. It's coincidental and means nothing.

Likewise, at times I could play lower odds and an occasional favorite, if the math was there, and it wouldn't damage my ROI over time. I don't, because of short term banking issues, but I could. However it would not have anything at all to do with the morning line. It's a coincidental play.

[bigga 1]

I will share with you guys the only angle that matters in a drop in class and prodouces a very positive ROI is when the owner and the breeder is the same person.This mainly occurs in maiden races....commonsense handicapping.....he didn't pay 100 or 200,000 for the horse so he is the only one that can really afford to drop a horse to whatever level with a very positive intent...as opposed to a horse dropping in class to half or more of what the owner has paid for him (negative) regardless who is riding him.Back in the day when cramer wrote about this angle we didn't have the information we have now as related to the price owners paid for their horses.Again Common Sense plays a big part in the handicapping process...check it out you will be amazed.And if anyone wants to share a solid angle among your peers that would be a great start to sharing of angles and ideas......Thanks

[Longshot6977]

If I've learned anything in horse racing,.....

Am I seeing this correctly, a 10 year thread bump? Wow.:)

I do like the idea of sharing your best solid angles, but I doubt most readers would be willing to do so in a parimutuel game. But I will review some files and post what I have found/used over the years if this thread takes off. However, I don't have backup data, just seat of the pants stuff like:

1. In a MDN/MSW race, look for a horse that broke from the gate better than his last race or so or had good early speed. Doesn't matter if he was beat by 25 lengths, the odds will be better. He is improving/rounding into form.

2. Class drop from MSW to MCL

3. A few toteboard/willpay angles

Of course, use common sense when making any bet.

[Cratos]

If I want to be technical? You were the one spouting off about the lack of "real statistical research". A longshot-favorite bias is defined thusly: The average returns go up as the probability of winning goes up. That is clearly the case.

Like I said, I wasn't claiming the bias was useful, just that it was there. And I've never seen anyone claim the bias was anything but weak -- the pools have always been fairly efficient. It is just a property of the market, more of interest to people doing econometric analysis of the behavior of markets in general than it is to handicappers, where what is important is the single race in front of them at the moment.

Well stated and well thought out!!!

[MJC922]

While it's useful IMO to look at a large sample just to get a baseline idea of the potency for an angle such as the class drop jockey switch, it's like anything else in racing, context matters.

First I suspect the angle probably pre-dates Cramer by several lifetimes, but with that aside, let's talk about what the angle is meant to uncover and that is simply trainer intent... as a sign of trainer intent a good rider gets the call, the trainer doesn't want to take an unnecessary risk of losing the horse via claim and not getting the winners share from some silly mistake that a low% rider might make. This is Ainslie 101 stuff. The ROI is probably all over the map for the past 50 years, I don't understand the fascination with ROI in cases like this, win% will typically go up with the better jock, to what extent it goes up is relevant to setting probabilities, the ROI can be anything, I would argue the ROI is almost meaningless for this angle.

About stats and context, I think back a week or so ago when Linda Rice had a horse coming off the layoff which absolutely figured and the stat was posted on the NYRA feed showing how apparently bad her ROI was off the layoff. The stat that I want to see however is the one showing me more context, i.e. in cases like that particular race, where the horse is a winner if it's ready how often does the horse win? Long story short, the last Rice layoff that I recall looking that obvious won for me, so I opted to bet right into the teeth of the generic stat and won again. I have no problem if people want to believe this was dumb luck. :)

[pondman]

First I suspect the angle probably pre-dates Cramer by several lifetimes, but with that aside, let's talk about what the angle is meant to uncover and that is simply trainer intent... as a sign of trainer intent a good rider gets the call, the trainer doesn't want to take an unnecessary risk of losing the horse via claim and not getting the winners share from some silly mistake that a low% rider might make. This is Ainslie 101 stuff. The ROI is probably all over the map for the past 50 years, I don't understand the fascination with ROI in cases like this, win% will typically go up with the better jock, to what extent it goes up is relevant to setting probabilities, the ROI can be anything, I would argue the ROI is almost meaningless for this angle.

I would completely disagree with all of this. There are meets that are design for drops; They are tracks for the unsuccessful horses. And the profits will be found in the lower tier of the those jockey colony. Many of the good bets come with those little ridden jockeys. Intent can be spotted on a drop, but I don't believe a jockey change is one of those identifiers. It's a coincidence.

Anyone who is telling you that a horse wins XX% of the time because... and therefore horse A has a chance to win at XX% is misleading you. There has to be more to the equation such as: XX% win + XX% lose + uncertainty(error or epsilon or whatever you want to call it.) And the xx% lose + uncertainty in my opinion is about 60%. Anyone who isn't defining this uncertainty in this way is a snake charmer.

[MJC922]

I would completely disagree with all of this. There are meets that are design for drops; They are tracks for the unsuccessful horses. And the profits will be found in the lower tier of the those jockey colony. Many of the good bets come with those little ridden jockeys. Intent can be spotted on a drop, but I don't believe a jockey change is one of those identifiers. It's a coincidence.

Anyone who is telling you that a horse wins XX% of the time because... and therefore horse A has a chance to win at XX% is misleading you. There has to be more to the equation such as: XX% win + XX% lose + uncertainty(error or epsilon or whatever you want to call it.) And the xx% lose + uncertainty in my opinion is about 60%. Anyone who isn't defining this uncertainty in this way is a snake charmer.

Profits are something else altogether. A better rider wins more often, that's step one, probabilities, on that I hope we can agree... whether or not you or I can make money on it is something entirely different. In fact in cases of obvious intent I suspect we're far LESS likely to make money on it.

[Hajck Hillstrom]

We met at Mark's place and started wading through a pile of Racing Forms. Like the experiment run by Barry, we just looked for class drops and jockey improvements based on a list Mark provided. It was an earlier time, and the system worked better than Barry found, but it still showed a loss.

Working alone in a corner, I added a column to the worksheet we were using: I noted the name of the trainer involved. After a while I found some that had no pattern, others that showed a nice ROI. When I showed this to Mark, he was impressed enough to invite me to join a group that was trying to use a computer to handicap with. Led by Mark and Dick Mitchell, the group never made much money, but we learned a lot and my career in racing was launched.

Would have liked to have been fly on that wall ...