Continuous Improvement: Yes or No?

[sjk]

I have always had the thought that as time goes by other players will improve their skills and it is a good plan to try to continuously improve my program to keep ahead. Usually once or twice a year I will test some ideas for improvement (often derived from something I have read here) and if things appear to back-test well I will implement the change.

I usually try to use large data-sets in the development process so I have fallen into the trap of testing against a portion of the same data that was used in development (there is a finite amount of data to work with) so perhaps I should not be suprised if future results differ from back-testing.

I made a couple of changes in the winter, one in March and one in May. After two months of less than satisfactory returns I backed out the last change a month ago and I am now backing out the 2nd last one. Here's hoping I can get things back the way they were without introducing an error; I'm doing the obligatory back-testing now.

I'm starting to wonder if the risk of "improvement" is worth the potential upside. Is the competition getting tougher or not? Hard to tell when I've made some changes that I probably should not have made and with the customary randomness in monthly returns.

Any opinions?

[ratpack]

I remember reading a story in one of Mark Cramer's newsletters about a guy who had a method/system of Handicapping that made him a 6 figure income for several years until it slowly fizzled to nothing but loses year after year.

The moral was that he never changed his method in a game that changes constantly.

Don't know why I keep telling meanless stories

[Jeff P]

I'm ALWAYS tinkering around with new ideas. About 99 percent of the time it turns out to be a dry well. But every once in a while... I actually stumble upon something very obviously significant. And if it's significant enough, I'll bring it into my own handicapping methodology. And if it proves successful there, I'll bring it into JCapper. I read a post sometime last year, I think it was made by GameTheory. The gist of it was that there are still watershed discoveries waiting to be made in handicapping. I believe in my heart that this is true. The nuggets are there. You just have to be williing to put in the work to find them.

-jp

.

[Dave Schwartz]

Jeff,

watershed discoveries

What does that mean?


Dave

[Jeff P]

Jeff,


Quote:
watershed discoveries



What does that mean?


Dave
Suppose I take a look at my Q2 2005 database for the tracks that I play. A mythical $2.00 bet on every starter in every race yields the following results:
Data Window Settings:
999 Divisor
Surface: (ALL*) Distance: (All*) (From Index File: F:\2005\Q2_2005\pL_profile_SAVED_071205.txt)
Data Summary Win Place Show
Mutuel Totals 82808.00 82065.50 79919.30
Bet -107624.00-107624.00-107624.00
Gain -24816.00 -25558.50 -27704.70

Wins 6863 13637 20146
Plays 53812 53812 53812
PCT .1275 .2534 .3744

ROI 0.7694 0.7625 0.7426
Avg Mut 12.07 6.02 3.97


By: Odds Rank

Rank Gain Bet Roi Wins Plays Pct
1 -2554.60 14240.00 0.8206 2455 7120 .3448
2 -2245.90 13430.00 0.8328 1447 6715 .2155
3 -2443.30 13504.00 0.8191 1039 6752 .1539
4 -2795.30 13496.00 0.7929 736 6748 .1091
5 -3314.90 13376.00 0.7522 488 6688 .0730
6 -2598.90 12388.00 0.7902 339 6194 .0547
7 -3699.80 10068.00 0.6325 164 5034 .0326
8 -2090.40 7126.00 0.7067 100 3563 .0281
9 -1454.20 4656.00 0.6877 51 2328 .0219
10 -863.60 2926.00 0.7049 28 1463 .0191
11 -462.10 1446.00 0.6804 10 723 .0138
12 -179.60 752.00 0.7612 5 376 .0133
13 -136.00 136.00 0.0000 0 68 .0000
14 34.60 68.00 1.5088 1 34 .0294
15 -2.00 2.00 0.0000 0 1 .0000
16 -2.00 2.00 0.0000 0 1 .0000
17 -2.00 2.00 0.0000 0 1 .0000
18 -2.00 2.00 0.0000 0 1 .0000
19 -4.00 4.00 0.0000 0 2 .0000
Pretty typical and about what I'd expect.

But suppose during my research and constant tinkering I discover a set of horses that can be easily identified and categorized using a very simple and straightforward set of rules. And suppose those rules can be encapsulated into one single filter. For the sake of argument I'll call that single filter CANTRUN to indicate horses that I don't think can run. Looking at CANTRUN horses only, in my Q2 2005 database I have:
Data Window Settings:
999 Divisor
Filters Applied: CANTRUN
Surface: (ALL*) Distance: (All*) (From Index File: F:\2005\Q2_2005\pL_profile_SAVED_071205.txt)


Data Summary Win Place Show
Mutuel Totals 8663.90 8203.50 7893.70
Bet -14810.00 -14810.00 -14810.00
Gain -6146.10 -6606.50 -6916.30

Wins 347 802 1357
Plays 7405 7405 7405
PCT .0469 .1083 .1833

ROI 0.5850 0.5539 0.5330
Avg Mut 24.97 10.23 5.82


By: Odds Rank

Rank Gain Bet Roi Wins Plays Pct
1 -110.00 282.00 0.6099 29 141 .2057
2 -74.20 504.00 0.8528 46 252 .1825
3 -306.10 842.00 0.6365 45 421 .1069
4 -405.00 1206.00 0.6642 51 603 .0846
5 -701.30 1698.00 0.5870 47 849 .0554
6 -473.10 2204.00 0.7853 56 1102 .0508
7 -1159.30 2428.00 0.5225 29 1214 .0239
8 -850.20 2002.00 0.5753 22 1001 .0220
9 -661.20 1476.00 0.5520 12 738 .0163
10 -577.40 1084.00 0.4673 7 542 .0129
11 -402.30 658.00 0.3886 3 329 .0091
12 -324.00 324.00 0.0000 0 162 .0000
13 -64.00 64.00 0.0000 0 32 .0000
14 -38.00 38.00 0.0000 0 19 .0000
15 0.00 0.00 0.0000 0 0 .0000
16 0.00 0.00 0.0000 0 0 .0000
17 0.00 0.00 0.0000 0 0 .0000
18 0.00 0.00 0.0000 0 0 .0000
19 0.00 0.00 0.0000 0 0 .0000

Much to my delight, every quarter in my database that I look at yields very similar results for CANTRUNs. Suddenly, eliminating CANTRUN horses from consideration before I bet starts to approach my own definitions of what a watershed discovery actually might be. By simply avoiding CANTRUNs, the set of horses that I do bet on, no matter what method I use to select them, HAS to improve.

-jp

.

[sjk]

No one has offered an answer to the question that I was trying to ask so I'll assume I was unclear and try again.

Is the competition getting tougher and if so at what rate? For example, if you're making 10% now and continue with your current methods would you expect that to deteriorate to 9%, 8%, 7% in succeeding years or would the rate of decay be more or less?

Someone who has used the same method for several years without much adjustment and who bets enough races to take the randomness out of the equation might have an opinion on this.

[DJofSD]

There's more one than one component to consider when asking the question 'is it getting tougher?' I perceive the question as it's discussed so far is focusing on either the skill level of the bettors (and how well the tools are working) and/or the number of bettors involved. That's just one part of the game.

By that I mean field sizes. If the average number of horses in a race is decreasing, especially for non-maiden events, I would expect it to get harder.

But that's just a gut level feeling.

DJofSD

[kenwoodallpromos]

If the public is getting better or more information to use, things could get tougher with 1 system; but if success is meet by meet or at certain times, you may just need to tweak it for that.
Yoy cannot say things are getting tougher without knowing why your system is not working as well. If favorite payoffs are getting worse than the win % needs to go up and some payoffs on non-favorites are getting better. That's the way it always is with para-mutual betting.

[sjk]

The reason that I would expect the competition to get tougher is expected improvement in the software that people are using and the ability of successful players to branch out to many tracks. Computer power and software capability advances from year to year in most applications. I would expect that serious handicapping software developers are also making progress.

Having said that it is logical to assume that this is happening, it is not my impression that it is happening as rapidly or to the extent that I might have expected, if at all. As I said before, I have made changes over the years, some that I think were extremely positive and others that were not so, so it is hard for me to separate out what is happening on my end from the external influences.

[ryesteve]

the ability of successful players to branch out to many tracks.
This is a very good point. Before the days of simulcasting and internet wagering, how many sharp players would you expect to find at Charles Town on a thursday night? Nowadays it's not uncommon for me to see a horse I like at 7-1 or 8-1 at one of these small pool tracks... but all of a sudden, someone who knows just enough to be dangerous puts a hundred or two on the horse, and he's now 3-1. At that price, I can't make any money. Sure, sometimes the money goes the other way, but the increase in odds when someone's betting another horse isn't nearly as dramatic as the decrease when they jump on yours.

[midnight]

I believe that the software and information revolutions have pretty much ran their courses.

What's going to make the sport tougher to beat is the fact that it isn't drawing enough new players. Racing has to fix its drugging problem before the young adults will get involved over the alternatives of sports betting, poker, and casino/reservation/riverboat gambling. Some jurisidictions are moving in the right direction in cleaning up their dirt, but a lot of states/tracks are doing nothing whatsoever.

It'd also help if the powers that be (Magna, CDSN, etc.) would stop fighting each other and work together to make it EASIER for the fans, instead of throwing roadblocks at them, but that's been hashed and rehashed here already.

[mainardi]

Since I'm in the handicapping software biz, I may seem just a tad biased, but I think that while the software and information REVOLUTIONS may have run their courses, it's the EVOLUTION (of continuous improvement) that's going to make the difference. Enough said about software.

Drawing enough new players isn't as much of a problem as drawing in new money. While it couldn't hurt, we don't NEED the younger crowd and their $7 an hour and all the Friday beer nights at Hollywood Park. What's needed is to find a way to get those with their disposable incomes to join us.

Those of us that still work for a living can do things like organizing a company "day-at-the-races". Tracks could "break tradition" and move first post to 3:30 on weekdays, put in some lights, and spend some money advertising on afternoon drive-time radio. It would help the business if some tracks moved their dark days to Wednesday-Thursday. How many more people would go back on a Monday afternoon after work if they had a great time on a Sunday? Instead, they have a great time, and then the doors are locked for two days! Not only that, but then TVG and HRTV would have something meaningful to show people EVERY DAY!!! More on that later...

As for the "drugging problem", if you ask 100 (or 1000) young adults for their top three reasons for not going to the track, I'm guessing that their top answers are things like "don't retirees do that?", "there's nothing to do between races", and "I don't know any of those rider guys". Even hard core guys like me don't worry much about the drugs, because "everybody's doing it", which evens the playing field.

As for the "Magna/CDSN wars", it's really about who can see what over the airwaves. Nobody complains that the NFL is on so many channels, or that you can watch baseball (not that I do) just about every night, and the list goes on. If Magna and CDSN want to have a testosterone contest, that's okay with me. What they're failing to do is expand their on-air products to the shills (HRTV and TVG) that broadcast for them. It would be like the NFL saying that nobody can watch the Bears, just because they're not that good. I have no doubt that if the broadcasters made the effort to ensure that there was racing action coming at you every five minutes -- ever been to a big Vegas race book? -- the sport would have more people coming, and along with it more money.

[Vegas711]

Maybe small field sizes have made the game a little more difficult. It is not the advancement of the computer age but rather the loss of poor players to other games thatis making it a little bit more difficult to win with also HIGHER TAKE OUTS.

Why you can rule out advancements in computers look at the % of winning favorites been the same before beyer was born and will remain the same long after he checks out. Horses hardly never run the same race they either make small changes race to race or big changes, no program will ever get this right.It is a grey area that for the most part is " FEEL" in the same way that a golfer has the right feel for a chip shot.

Forget about inprovements concentrate more on playing Sharp and developing a feel for when you should bet 1 rating for 1 race and another rating for a different race.I wasted 10 years trying to find this magic rating it does not exist. It was not till I focused on developing a feel for what rating to use that my game elevated itself.