VALUE

[TenZin]

Which set of player picks is more valuable.
#1 >> 10/17 Winners (59%) with a ROI of 2.27 or
#2 >> 12/33 Winners (36%) with a ROI of 1.79.

[ranchwest]

The combination of higher ROI and percentage is certainly better. However, the sample size is far too small for any conclusions.

[andicap]

Not so fast,

let's say you are betting a flat amount of $100 per race.
33 races, you would have bet $3,300
17 races, you would have bet $1,700

the smaller ROI on the $3300 would produce larger overall revenues than the larger ROI on the $1,700.
If you're looking at the bottom line (the only line I look at), I'd take the 12-33 with 1.79 ROI thank you.

[rrbauer]

Andicap wrote:

the smaller ROI on the $3300 would produce larger overall revenues than the larger ROI on the $1,700.

Comment:
True statement.

Another true statement is that it is the riskier proposition.

[ranchwest]

Originally posted by andicap
Not so fast,

let's say you are betting a flat amount of $100 per race.
33 races, you would have bet $3,300
17 races, you would have bet $1,700

the smaller ROI on the $3300 would produce larger overall revenues than the larger ROI on the $1,700.
If you're looking at the bottom line (the only line I look at), I'd take the 12-33 with 1.79 ROI thank you.

If you want productive return, then double down on the high percentage, which gives (approximately) the same total wager.

This is far too small a sample to indicate that kind of confidence, though.

[Dick Schmidt]

No, this isn't a big enough sample to demonstrate long term performance, but its big enough for me to start betting them both. Who cares which is better, they both produce very good profits. Once you find stuff like this, get the hell out of research mode and head for the windows. Start small and play work-ups with your bets.

Go get 'um,

Dick

[Tom]

Too many times things that are profitable are also fleeting-by the time you have a big enough sample, it is not making profits anymore. Get on the bandwagon while it is working.
When your records show it is not that good anymore, stop betting it.
What good is a 10,000 race sample if you didn't bet any of the races? I would rather have a 10 race sample that I got in on the last 4 or 5 winners. When I see an interesting trainer move, I will bet it the next time I see it occurr, even though I have only seen it happen once before. If I lose on it a couple of time, I stop playing it. More often than not, it happens a few more times then disappears. My old FL records show that the same trainer can vary greatly year to year-say with layoff horses, one year the guy is 40% the next year he is 0%. The next year he is 20%.
Longterm, he is 60% (for sake of arguement) but you lose your shirt in year two, make a little in year three, and pass up year one, the good year, collecting data. My way, I have him year one, get off early in year two, and either forget him in year three, or notice he has a few hits agian, and get back on.

[ranchwest]

I didn't say not to play it. I'm just saying that there are no assurances that such a small sample will pan out.

Part of it depends on how contrived the rules are. For spotplays, the simpler the rule the better. For point systems, you can have a whole lot of rules, but it helps if each individual rule is simple. Complex rules suggest that the rule may be too contrived, in which case I've found that it is more likely to break down.

As Tom mentioned, with trainer rules you usually have to go with small samples because you just can't get much data on an individual trainer on a specific pattern. Once may be enough to trigger you to follow through with the pattern. If it is paying off in double digits, you won't have to hit many to have success.

[andicap]

Originally posted by Tom
. When I see an interesting trainer move, I will bet it the next time I see it occur, even though I have only seen it happen once before. If I lose on it a couple of time, I stop playing it. More often than not, it happens a few more times then it disappears. My old FL records show that the same trainer can vary greatly year to year-say with layoff horses, one year the guy is 40% the next year he is 0%. The next year he is 20%.
Longterm, he is 60% (for sake of argument) but you lose your shirt in year two, make a little in year three, and pass up year one, the good year, collecting data. My way, I have him year one, get off early in year two, and either forget him in year three, or notice he has a few hits agian, and get back on.


Tom, that is some of the best advice I've seen on this board.
That's the problem with these long-term BRIS or DRF stats. It's like flipping a penny. Yeah, in the long-run it will be 50% heads/tails, but in the short run, heads could have a horrible slump. If you flip a penny once a month, you could go months without winning.

(BTW, hope you don't mind me cleaning up your spelling/grammar errors)
:)

[Rick]

My personal rule to start betting is at 30 winners. I don't care if it's a high percentage method and I only have 100 races or a longshot method where it takes 200 races. When I have 30 winners, I setup a small bankroll and start betting it. If it continues to win, I escalate my bets and win a lot. If it doesn't pan out, I haven't lost much. Try it, it works.

[GameTheory]

Rick,

That sounds reasonable for a general method.

But if you're looking at a specific trainer angle for a specific trainer, you'll be lucky to get 30 winners ever.

Trainer angles that are specific enough should be looked at like key races. See it twice, bet it the third time...

[Rick]

GT,

I agree, but you're talking about the rules of your method rather than the criteria you use to begin betting your method. What I mean is that you've surely used the same rules with other trainers in the past and accumulated enough winners and plays to justify betting the method.

[GameTheory]

Originally posted by Rick
GT,

I agree, but you're talking about the rules of your method rather than the criteria you use to begin betting your method. What I mean is that you've surely used the same rules with other trainers in the past and accumulated enough winners and plays to justify betting the method.

An apt and proper abstraction.