Measuring -- How do you do?

[DeltaLover]

Originally Posted by green80

I developed a computer program that uses bris files to generate plays. I usually get about 2-3 plays on a 10 race card. (mostly chalk horses) I have handicapped 72 races and showing a positive roi.(highest price horse was 11.00) How many races would one need to handicap to have any confidence in this system? Is there any formula that would apply to give any confidence level?

The best formula would be to start betting with real money.

[Ray2000]

Originally Posted by green80

I developed a computer program that uses bris files to generate plays. I usually get about 2-3 plays on a 10 race card. (mostly chalk horses) I have handicapped 72 races and showing a positive roi.(highest price horse was 11.00) How many races would one need to handicap to have any confidence in this system? Is there any formula that would apply to give any confidence level?

green80

I'll refer you to my last reply in the Harness Forum
http://www.paceadvantage.com/forum/s...d.php?t=105679
reply #7

The link to "Archie" may be what you're looking for.

Other approaches all involve "Z" scores in some manner or another.



Attached is Excel spread sheet which will give the Z score of your test run.

The sample run in column A, has 1000 races and has a Z score of 0.0267.
These are real numbers for a run I did in 2011

You need a score below 0.05 (95% Confidence level) to start believing in the system.

If you delete results in rows 900-1000 Z becomes .017 (depends on the variance)
If you delete results in rows 800-1000 Z becomes .037
If you delete results in rows 700-1000 Z becomes .060 (700 races on the Example run returns is shaky)

[nat1223]

Big thing to remember in this game is that each race is unique and you need to have a good understanding of what is winning at that time. Keeping track of your picks and studying them is a start. However, what works today will likely not work tomorrow or it may work for a few weeks and then crash and burn. I have seen handicappers with large databases come to find out that examining certain consistent factors and running tests find that the ROI goes well below 2.00. This game is difficult because you cant find some angle and think you are going to profit over the long haul by playing it day in and day out without examining the particulars to each race. People who profit at this game do so by being able to make adjustments quite frequently. People buy computer software programs and bet races based on what the software spits back as top horses without examining the race in depth. Big mistake in horse racing. This may be a bit off topic so i apologize for that!

[Capper Al]

Originally Posted by TrifectaMike

http://www.paceadvantage.com/forum/s...light=bullshit

Do similar as in the above thread and update the Binomial on a daily or weekly basis.

Mike

Thanks. It applies.

[Capper Al]

Originally Posted by Tom

If you are tweaking every 3-4 cards, you will never know what worked and what didn't. You might tweak yourself out of the black and into the red.
When you make a tweak, go back and see what it did to your last 100 races of so.

That's a serious problem. It makes interpreting the short run more important.

[jasperson]

[QUOTE=Capper Al]Sartin or Ainslie said something like the average system last 2.2 days before a handicapper is off trying another method. Well, the question that I have is how does one judge how well they are picking them? This is not an ROI measure or how much profit analysis. Just simply, how well did that system or method of picking them do for you?

What the average handicapper needs is a quick method for measuring systems with only a few races. Most of us don't have the huge databases that Dave has available. With our tweaking every 2.2 days (as the joke goes), how can we tell?

'Win Percent' or 'In The Money' or 'Impact Value' or UDR (Universal Driver Rating as used in harness racing) might head the list. I find problems with them all especially when using a few races. Let's say for the discussion a few races is 30 races. (2.2 days might be 3 cards of ten races each.)

The problem with 'Win Percent' and 'Impact Value' with on a few races is that a system's second or third pick might radically out perform their top pick in a sample of 30 races. For instance, out of 30 races one might hit as follows: 3 wins, 9 places, and 6 shows. A 10% hit overlooks the point that the ITM hit was 60%. Is this system a keeper that just needs a little tweaking? This is why I favor ITM or UDR approaches for quick analysis.

I was tracking two systems on Million day. System A had a 30% win hit rate while system B had a 40%. Drawing any conclusions on 10 races is dangerous. But at the end of any racing day, I simply like to know how well did I pick them? And if I picked them this way or that way, how well would I have done? Before I post a method for measuring a day or two at the races, I'd like to hear what others do for a quick measure. And if one believes that a quick measure could have any value in it at all?[/QUoOTE]
I looked at a system(or method) of betting on last Sat races at 4 tracks and looked vary promising. I enter the results from Thursdays races and it was a bad day. I looking at thurs races there was a lot illogical longshot and turf races at Sar. Should I throw the method out or give it further investigation?
I don't think you can base a opinion on 70 races.

[Capper Al]

I agree that 70 races is too small an number to conclude anything. Three hundred would be my minimum. The question being solved is for a short run indicator. That's what Trifecta Mike pointed out --- the Bayesian approach answers this question for the short run. With your 70 races, you might have questioned (as it appears) how well are you doing? Bayesian approach will tell you how will you're doing within a level of confidence.

[lamboguy]

my record at Saratoga this year is the easiest one of all time to keep track of. i have bet about 75 races and have only hit 1. i did make the 1 race i hit count though. there is another one that i am waiting for if she runs at the Spa, i just hope she is alright if and when she gets in. for the record i could have hit another nice race, but something held me back from playing it.

[JJMartin]

Originally Posted by nat1223

Big thing to remember in this game is that each race is unique and you need to have a good understanding of what is winning at that time. Keeping track of your picks and studying them is a start. However, what works today will likely not work tomorrow or it may work for a few weeks and then crash and burn. I have seen handicappers with large databases come to find out that examining certain consistent factors and running tests find that the ROI goes well below 2.00. This game is difficult because you cant find some angle and think you are going to profit over the long haul by playing it day in and day out without examining the particulars to each race. People who profit at this game do so by being able to make adjustments quite frequently. People buy computer software programs and bet races based on what the software spits back as top horses without examining the race in depth. Big mistake in horse racing. This may be a bit off topic so i apologize for that!

You could examine each race in depth and you'll still lose. In fact you could over analyze it. You get a strike rate of about 60% in the top 3 speed figures consistently. Now if someone knows how to determine which of the 3 is going to be the winner, you got a million dollar system.

[pondman]

Originally Posted by nat1223

People who profit at this game do so by being able to make adjustments quite frequently. !

If you are going to make big profits at this game, then you will need to stay away from the top 3 performance driven contenders. Hopefully, your handicapping develops to the point of being able to see the potential of unproven animals. Because if you can't you won't make it in this game. Adjustments? I don't think you need to make adjustments. You need to really be able to think like a trainer. Is the horse ready to run a big one? Can you get that type of information, stuff the crowd doesn't see? Yes. I think you can on enough races to make it profitable and interesting.

[CincyHorseplayer]

Originally Posted by pondman

If you are going to make big profits at this game, then you will need to stay away from the top 3 performance driven contenders. Hopefully, your handicapping develops to the point of being able to see the potential of unproven animals. Because if you can't you won't make it in this game. Adjustments? I don't think you need to make adjustments. You need to really be able to think like a trainer. Is the horse ready to run a big one? Can you get that type of information, stuff the crowd doesn't see? Yes. I think you can on enough races to make it profitable and interesting.

This is a great point and reason why I have been getting younger and younger and fewer PP's because anticipating a win before it becomes a past performance is something that drives most handicappers crazy.They turn their heads away from it willfully.

[dkithore]

Originally Posted by pondman

If you are going to make big profits at this game, then you will need to stay away from the top 3 performance driven contenders. Hopefully, your handicapping develops to the point of being able to see the potential of unproven animals. Because if you can't you won't make it in this game. Adjustments? I don't think you need to make adjustments. You need to really be able to think like a trainer. Is the horse ready to run a big one? Can you get that type of information, stuff the crowd doesn't see? Yes. I think you can on enough races to make it profitable and interesting.

your point is un-disputable, IMO. And so, the question one needs to ask is, is this horse ready to peak? Are others, the obvious contenders, about to slide the slope or have some flaws (like three wins and out), wrong running style at this track and many other factors, too many to list here. Isn't that the heart of handicapping for value all about? What the public should not like? Do we have the temperament to work hard and wait for the opportunity? That is the crux of the issue.