The Cramer Big Win

[CBedo]

I've been building a new database using MongoDB with the intent being to build a better analytical framework to do research on the data. In doing so, I've been playing with some queries just to see ease of querying and how much I can cut down query times. Since it's been a little slow in the handicapping forum of late, I thought I'd use some of my testing to discuss some old Cramer work.

In Cramer's 2002 Hidden Probabilities: Hard Core Research for X-Rated Horseplayers, the first chapter was dedicated to research on what happens when a horse wins big (by 5 lengths or more). His hypothesis was that with a few exclusions, betting horses that had won big, might show a positive result. His major exclusions were big winners from maiden claiming races, and big winners that were then laid off for 45 days or more.

Note: This is not a production db, it's still in development, and I'm still tinkering with the parsing and the structure, so there are only about 150,000 races stored currently, and in some of those, the information is from histories, not charts, so the information set isn't complete. This is why, as you'll see, when I pull a sample of 1000 races, that there might only be 900 or so in the sample, as some data to complete the query just isn't there, so I drop the race.

First, here's 10 1000 race samples of what happens when you bet a horse than won by 5 lengths or more back the next race.
BETS: 918 WIN%: 0.252 RETURN: -0.169
BETS: 870 WIN%: 0.244 RETURN: -0.162
BETS: 879 WIN%: 0.204 RETURN: -0.166
BETS: 885 WIN%: 0.246 RETURN: -0.054
BETS: 890 WIN%: 0.224 RETURN: -0.200
BETS: 863 WIN%: 0.226 RETURN: -0.262
BETS: 869 WIN%: 0.239 RETURN: -0.090
BETS: 876 WIN%: 0.245 RETURN: -0.158
BETS: 828 WIN%: 0.255 RETURN: -0.015
BETS: 877 WIN%: 0.235 RETURN: -0.175
Second, here's the same samples, excluding maiden claiming races.
BETS: 723 WIN%: 0.281 RETURN: -0.126
BETS: 717 WIN%: 0.250 RETURN: -0.162
BETS: 641 WIN%: 0.229 RETURN: -0.231
BETS: 644 WIN%: 0.272 RETURN: -0.041
BETS: 661 WIN%: 0.234 RETURN: -0.255
BETS: 666 WIN%: 0.249 RETURN: -0.177
BETS: 696 WIN%: 0.251 RETURN: -0.110
BETS: 698 WIN%: 0.255 RETURN: -0.191
BETS: 612 WIN%: 0.284 RETURN: 0.020
BETS: 683 WIN%: 0.255 RETURN: -0.095
Third, here's all classes, but excluding 45+ day layoffs.
BETS: 754 WIN%: 0.256 RETURN: -0.176
BETS: 705 WIN%: 0.257 RETURN: -0.145
BETS: 741 WIN%: 0.209 RETURN: -0.179
BETS: 704 WIN%: 0.251 RETURN: -0.064
BETS: 767 WIN%: 0.229 RETURN: -0.253
BETS: 723 WIN%: 0.230 RETURN: -0.280
BETS: 740 WIN%: 0.257 RETURN: -0.025
BETS: 730 WIN%: 0.259 RETURN: -0.110
BETS: 716 WIN%: 0.254 RETURN: -0.019
BETS: 725 WIN%: 0.247 RETURN: -0.130
Lastly, here's the results, excluding both layoffs and maiden claiming.
BETS: 592 WIN%: 0.289 RETURN: -0.106
BETS: 576 WIN%: 0.260 RETURN: -0.154
BETS: 529 WIN%: 0.238 RETURN: -0.206
BETS: 519 WIN%: 0.270 RETURN: -0.096
BETS: 572 WIN%: 0.243 RETURN: -0.238
BETS: 555 WIN%: 0.256 RETURN: -0.182
BETS: 595 WIN%: 0.266 RETURN: -0.066
BETS: 580 WIN%: 0.267 RETURN: -0.141
BETS: 526 WIN%: 0.285 RETURN: 0.033
BETS: 562 WIN%: 0.267 RETURN: -0.050
While it's pretty clear that the exclusions do raise the win rate, it's somewhat less clear what happens to return. In a number of the samples, the ROI actually gets worse, not better--unfortunately, not the positive automatic bet that Cramer seemed to find in 2002.

Thoughts?

Note: Those familiar with Cramer's book probably will note that I left out one of his other exclusions--off tracks. As I stated before, this db is still in the development phase, and I just changed the way I store track conditions. As such, I need to rebuild the db to include that information in the queries.

[Overlay]

I would take a big win as a positive indicator, but only to the degree supported by long-term winning percentages and impact values, and in conjunction with other performance aspects (particularly those that might appear negative on the surface, but which would have their effect mitigated by other less obvious positive aspects of the horse's record (the "plus factor" variety), in order to increase the chances of finding acceptable value).

I wouldn't expect any single factor (especially one so apparent as a big win) to show a positive ROI in subsequent races over time as a standalone factor, or even with additional qualifications. As far as last-race winners go, I'd be inclined to give more credit to a horse that won its last race by a narrow margin after a neck-and-neck stretch run.

[Bruddah]

Have you looked to see the outcome, if the next out is up in class or distance?

[Dave Schwartz]

Chris,

Save yourself the effort. We've had "Big Win" in since William Quirin defined it in (I believe) his second book. It simply has never shown anything to recommend.


Dave

[CBedo]

Chris,

Save yourself the effort. We've had "Big Win" in since William Quirin defined it in (I believe) his second book. It simply has never shown anything to recommend.No worries Dave. The purpose of the exercise was more to see how easily I could write the query (and/or code), than it was about the actual query. I just thought since I had done it anyway, and there hasn't been much actual handicapping discussion lately in this forum, I'd post the results to see if anyone had anything to add, or to look for further research, or even just for fun.

[CBedo]

Have you looked to see the outcome, if the next out is up in class or distance?Dealing with class changes is one of the many reasons I redesigned the database. Defining a class drop or hike is interesting. I am still working out the kinks in my new design. An example would be a filly in a statebred $7500 claiming race for non winners of 3 races in the last 18 months moving into an open $5000 claiming race limited to fillies and mares, at a different track. Is that a drop, a hike or roughly the same? (I know this is exaggerated, but you get the point).

As far as distance goes, that's easy and I just reran it with the same samples but only when the big winner was stretching out the next race. 7 of the 10 samples were worse, not better.

[misscashalot]

Chris,

Save yourself the effort. We've had "Big Win" in since William Quirin defined it in (I believe) his second book. It simply has never shown anything to recommend.
Dave

That's assuming Quirin is the be-all and end-all. I agree with Overlay in using added qualifiers. They lie hidden in plain sight and should be employed. Maybe change from 1 turn to 2 turns, or separating the 4 age groups. There are many other obvious factors that can be added and subtracted until you realize maximum efficiency.

Anyway, getting back to Q, the game constantly changes and old methods that lay on the shelf, given up as archaic and therefore less than worthless, should be resurrected if for only briefly, to see if they have any value in todays game. There are also less tangible reasons to do this.

[CBedo]

I think running simple queries like that can often give you things to think about, even if they don't directly lend themselves to increased hit rate or profitability. Most of these types of discussions aren't presenting anything new, but that doesn't mean they are worthless either.

For example, when running those queries, it made me think about "is a horse than wins by 5 (as per Cramer) a better win candidate next out than a horse than only wins by a length?" Or in general, does the winning margin impact the win rate next out?
WIN LENGTHS: 0.00 WIN%: 0.208
WIN LENGTHS: 0.50 WIN%: 0.220
WIN LENGTHS: 1.00 WIN%: 0.232
WIN LENGTHS: 1.50 WIN%: 0.226
WIN LENGTHS: 2.00 WIN%: 0.238
WIN LENGTHS: 2.50 WIN%: 0.240
WIN LENGTHS: 3.00 WIN%: 0.228
WIN LENGTHS: 3.50 WIN%: 0.241
WIN LENGTHS: 4.00 WIN%: 0.239
WIN LENGTHS: 4.50 WIN%: 0.248
WIN LENGTHS: 5.00 WIN%: 0.251
WIN LENGTHS: 5.50 WIN%: 0.251
WIN LENGTHS: 6.00 WIN%: 0.253
WIN LENGTHS: 6.50 WIN%: 0.264
WIN LENGTHS: 7.00 WIN%: 0.255
WIN LENGTHS: 7.50 WIN%: 0.254
WIN LENGTHS: 8.00 WIN%: 0.271
WIN LENGTHS: 8.50 WIN%: 0.277
WIN LENGTHS: 9.00 WIN%: 0.281
WIN LENGTHS: 9.50 WIN%: 0.279
WIN LENGTHS: 10.00 WIN%: 0.292
WIN LENGTHS: 10.50 WIN%: 0.274
WIN LENGTHS: 11.00 WIN%: 0.269
WIN LENGTHS: 11.50 WIN%: 0.270
WIN LENGTHS: 12.00 WIN%: 0.283
WIN LENGTHS: 12.50 WIN%: 0.269
WIN LENGTHS: 13.00 WIN%: 0.265
WIN LENGTHS: 13.50 WIN%: 0.262
WIN LENGTHS: 14.00 WIN%: 0.258
WIN LENGTHS: 14.50 WIN%: 0.256
WIN LENGTHS: 15.00 WIN%: 0.235

[trying2win]

CBEDO:

Thanks for sharing the interesting results.

I can recall reading about Mark Cramer's 'Big Win' theory in an edition of his C & X Report from a few years back. In it, Mark was showing a good profit in his tests. So, I'm guessing after this 'Big Win' angle became public knowledge from these publications, that that might have had an negative effect on results since then.

T2W

[Tom]

andicap and I had an exchange about this here a few years ago....I'll see if I can find it.

Short-term, I was making a nice profit using the idea. I think I had a second rule, but don't remember what it was....I'll see if I can find the thread.

[duckhunter3]

Here are a couple of things to consider.

First take a lookt at Gordon Pine's piece concerning the effect of a Competitive Race Last Time Out.

http://www.netcapper.com/TrackTractsArchive/TrackTractsArchive.htm

He first studied a competitive race last out on an off track, and then finally went on to show stats that a competitive race last out on ANY track portends a probable decline in performance next time out.

Second, for me at least, in low level claiming races, it is reasonable to assume that while it does happen, the odds of a repeat performance to a big win last out (or even a bigger performance) are far less than the odds of a decline in performance (and Pine's numbers agree with that).

So I take the position that while such a performance doesn't DISQUALIFY the horse from consideration, IN MOST CASES (but not all) it is reasonable to downgrade the horse because of that big effort last out. PARTICULARLY IF IT IS ITS BEST PERFORMANCE IN THE PP'S (THE OLD BOUNCE THEORY, WHICH I DEFINITELY BELIEVE IN).

duckhunter3

[Ian Meyers]

I would think that anything that obvious is dramatically overbet. I've never crunched the numbers before but intuitively I would have expected a high win% and 15-20% loss.

I find these kinds of things using HTR all the time, high win %'s and ROIs that erode every year.

[BlueShoe]

May be going against the grain, but I still love these types. Call them the Ainslie big win runners, rather than Cramer. Have the lead at the stretch call and gain ground from there to win going away. As we all should know, they cannot all be played blindly, more handicapping is required, and many of them will be eliminated. The most obvious would be a runner that laid over a dreadful field last out to win easily but with soft pace and speed figures. Distance, surface, recency, workout patterns, pace scenario, connections, and of course todays foes are all factors. Cannot put all this in hard and fast rules or develop a "system". If everything else looks okay, in all of racing my favorite play is a runner coming off of a fast easy win, coming back quickly and stepping up in class. These kinds will often outrun their numbers and beat horses with slightly superior recent figs. Have seen them step up as much as four or five class levels and repeat. By back soon, the quicker the better. Very rare today, but back in 7 days or less has the rubberband coming off of the bankroll.

[duckhunter3]

Cramer Big Win

Blue Shoe added a very important point- jump in class (at least by a good trainer).

All I know is MOST of the time I am glad I didn't bet a horse who was competitive (close to the lead at the 2 Call or finish) last out IN A LOW LEVEL CLAIMING RACE. Too unreliable. Now, some may win, but that doesn't bother me.

Certainly, I keep them for exotics (vertical and horizontal) but no big win bets.

I may be missing a few but I am scoring a lot too.

this could be discussed for eons because there are so many factors to consider- quality of field, trainer, jock, qualilty of last race, recent form otherwise, speed numbers versus the rest of the field, drop or rise in competition since last out, how many with trainer, is there something special (layoff, claim dirt to turf, 1st on turf, yada yada yada). I don't have time to list all the factors.

I am just saying, in LOW LEVEL CLAIMING RACES if last out was a big win, you need to look closer.

duck

[Robert Goren]

The Big Win System has been around forever. I remember reading an article on it in Turf N Sport sometime around 1964 or 1965. Like all things, it works once in awhile especially early in the year with 3 year olds colts. It is not the secret to making a living betting the races however.

[Stillriledup]

Overlay makes a good point about wanting to give more credit to horses who win by short margins rather than horses who win big. I've found that i love to back horses who show they can 'fight' with other horses and show desire.

I'd bet a lot of horses who win by 5 or more are wire to wire winners who went unchallenged all the way in the 1 path and 'got brave' when no other runner got near them.

I'd love to see this 5+length sample split up and just take the horses who won by 5 or more from off the pace. I'd bet there are very few winners who came from behind and won by 5 or more, but those runners, to me, are more impressive than a horse who got the lead and just kept going.

[CBedo]

I'd love to see this 5+length sample split up and just take the horses who won by 5 or more from off the pace. I'd bet there are very few winners who came from behind and won by 5 or more, but those runners, to me, are more impressive than a horse who got the lead and just kept going.I took a base sample of all the horses than won by 5 or more lengths last race and bet them back (didn't exclude layoffs or maiden claiming). I then split the sample into three subsamples: the horse was ahead at the 1st call, the horse was between 0 & 2 lengths behind, and the horse was behind more than 2 lengths at the first call. I then did the same for the second call. Here's what I came up with:

1st Call Position -- Win%
On lead -- 0.236
> 0 to <=2 lengths behind -- 0.236
> 2 lengths behind -- 0.219

2nd Call Position -- Win%
On lead -- 0.238
> 0 to <=2 lengths behind -- 0.236
> 2 lengths behind -- 0.195

[clore1030]

May be going against the grain, but I still love these types. Call them the Ainslie big win runners, rather than Cramer.

Well, thanks for that anyway. When I saw the thread title, and then the reference also to Quirin, I was certain that one of the oldest books in my collection, an Ainsley title, was the first to refer to "the big win."

[BlueShoe]

The Big Win System has been around forever. I remember reading an article on it in Turf N Sport sometime around 1964 or 1965.
Goes even farther back than that. Les Conklin wrote a couple of books in the 50's that promoted the concept that horses that were 2nd or better at the stretch call and gained ground to win were strong prospects to repeat.

[CBedo]

It's been great to see how such a simple idea can spawn so many opinions, qualfiers and open discussion. This is why I love this forum. :ThmbUp: :ThmbUp: :ThmbUp:

[classhandicapper]

If you stick with light raced undefeated horses (maybe one loss as a FTS) that drew off through the the stretch to a big win and closed in very fast time I think you'll have a subset of horses that contains some that are capable of running a lot faster than they've revealed to date.

[John]

I got a flier from RPM a while back. They have a software on Mark Cramer's work. I don;t know if the big win [ Cramer's version ] is in that program. May be one of the members has the program.

[ManeMediaMogul]

It's been great to see how such a simple idea can spawn so many opinions, qualfiers and open discussion. This is why I love this forum. :ThmbUp: :ThmbUp: :ThmbUp:

Try your sample with the big win with a speed rating higher than any other entrant's last race speed rating.

I have been using this angle for years with much success.

[ranchwest]

Try your sample with the big win with a speed rating higher than any other entrant's last race speed rating.

I have been using this angle for years with much success.

That's a bit similar to Scott's suggestion.

Scott's big win definition excluded horses on a large lead because the result was never in question.

He also had the qualifier of ability time.

To all: While big win horses may not always be good bets, how often do you want to empty your pockets to bet against them?

[Dave Schwartz]

One of my upcoming episodes of Improving Your Game in 10 minutes or Less takes a very quick look at 6 of Quirin's form angles:

Taxing Drive, 3 calls
Taxing Drive, 2 calls
Stretch Gain
Bid-Hung
Z-Pattern
Big Win

There are some basic statistics from the last 4 years as well as a more comprehensive look at the only factor of the 6 that seems to offer much.

Should be released in the next week or so.

[PICSIX]

One of my upcoming episodes of Improving Your Game in 10 minutes or Less takes a very quick look at 6 of Quirin's form angles:

Taxing Drive, 3 calls
Taxing Drive, 2 calls
Stretch Gain
Bid-Hung
Z-Pattern
Big Win

There are some basic statistics from the last 4 years as well as a more comprehensive look at the only factor of the 6 that seems to offer much.

Should be released in the next week or so.

I'm guessing Z-pattern on dirt only. ;)

[illinoisbred]

As long as we're guessing,I'll take bid but hung. Back in the mid 80's, I tinkered with variations of Quirin's B but H angle. I found that if you insisted on the bid to be into 1st or a length out of 1st,results improved. If you applied pace and speed figures to cull out the imaginary bids in races that were falling apart,results improved. Also,if the bid race was the 3rd-5th off a 90+ day layoff,results improved. I have no idea though if these "improvements" have held up over the last 25 years.

[John]

Also,Remember, that 1 out of 5 last out winners repeat. It is pretty hard to get 5-1 on a repeater. unless you are spot playing certain pattens.

[Dave Schwartz]

See, those would have been my obvious choices.

Like I said, you will be surprised.

[cj]

I've read pretty much all the books. I would say very few, if anything, will show a profit. There are good ideas and certainly the books give a starting point for research, but again, there was no real testing done.

[Tom]

I've found most things I test come up short on their own,l but once you start fining what compliments some other factor and combine them, you can really get some good results.

Wasn't it Rick Ransom who used to come up with a lot of three-factor plays?
Whatever happened to him?

[ranchwest]

See, those would have been my obvious choices.

Like I said, you will be surprised.
So many teasers... I feel like I'm watching CNN.

[Dave Schwartz]

LOL - Think of the ketchup commercial...Anticipation...

[JohnGalt1]

It's been great to see how such a simple idea can spawn so many opinions, qualfiers and open discussion. This is why I love this forum. :ThmbUp: :ThmbUp: :ThmbUp:

Another important exception is a 3+ big win on a wet/fast or sloppy track.

[ranchwest]

LOL - Think of the ketchup commercial...Anticipation...
Sorry, it reminds me of a laxative advertisement.