your opinions please

[tilson]

Lets say i played 75 races and had 25 winners for 33% and Roi of 200 vs 150 bet.....now lets say that in 5 of those races i had played a 2 to 1 shot FOR 4 BUCK and a 4 to 1 shot for 2 bucks for a total of 80 wagers.
where the double bet won all 5 times for 230 total bucks
what is the correct approach
25 vs 75 for 200 bucks or is it
25 vs 80 for 230 bucks or is it
30 vs 80 for 230 bucks

I mean is 30 winners ok vs 25 actual winners etc etc

[Larry Hamilton]

Since you appear to be structuring the amount to wager on the odds, you haven't given enough data. 5 winning bets at 2/1 or less is not statistically significant. The question you pose cant be correctly answered.

[tilson]

I realise that the sample is small and thus unreliable however if we times every thing by 100 that no longer would be the case and i believe the question would be valid.

[Dave Schwartz]

Tilson,

It is whatever you wish it to be. Personally, I see each bet as a bet. However, if you choose to see the race as an entity, then so be it.

[tilson]

Dave i had a guy tell me it should be 25 of 75 for 200
However i think i tend to agree with you that 30 vs 80 for 230 might be more correct.
The way i see it.....i don't see why one shouldn't take credit for correctly identifying a horse worthy of a doube bet.......surely there will be times when those double bets go south and when they do one must also count them as double loses.
I think it then becomes a factor of the times right on double bets vs the times wrong on double bets etc etc
while the base bets would remain the same under either condition.
The way i see it, it is no differnt then what a card counter does in blackjack.....if he feels he has identified a situation that is favorable he wagers more.....in the long run if he is correct he will prosper.....if he is not right in is evaluation of the situation.....in the long run his ROI will be a reflection of that.

[Lefty]

Larry was right, not enough data. Now if you do extend
this 100 times for real and it holds up, you've found a
marvelous method. However, in real world can't expect
projections to hold up; you must actually garner the data times 100.

[Que]

tilson,

I think you are trying to compare apples to oranges here. Overall, your method of selection still selected 25 winners in 75 races, regardless of your doubled bets. Specifically, its hard to justify using 105 races (25+80) in a 100 race sample. However, your "double bet" method of selection was 5 for 5, in the same 100 race sample. I think the two methods of selection are independent. And who knows, after a season of using both these methods, you might find that you would have done better playing only the "double bet" horses. Finally, your overall selection/money management methodology should use total return to compute its ROI, i.e. your total return divided by your total money wagered. I'm sure blackjack players factor in their double down bets and splits to compute their overall ROI. But as Dave said, it's really "whatever you wish it to be," i.e. it's your money and you are the one assuming all the risk--so you can call it whatever you like.

regards, Que.

[Rick Ransom]

Although what everyone has said here is true about small samples, etc., I think it's a valid question to ask. This is something that comes up in other ways. For, instance, when you win money at the beginning of the year and increase your bankroll, you increase your bet size later in the year. When you total up all of the bets at the end of the year, you can calculate a win % counting all your bets equally. Your ROI however, will be dependent more on the bets made later in the year, because there will be more money bet on them.

Now, the real question is what are you going to do with the information. If you're just going to document that year's results, then the above figures are OK. But, if you're going to try to use that year's data to try to predict how the next year might go, or use it to make money management decisions, you'd better go back and look at the year on a constant bet basis.

When you bet more than one horse in some races, you have an even more complex situation. The purpose for doing this is usually to decrease risk even if ROI suffers, because this allows you to bet more on the race.

It is rather similar to card counting in blackjack, but I've never heard of a blackjack player worrying about which way to count his % of hands won. They just don't care about this number. The best they can do usually is work with average bet sizes and average odds to determine a money management strategy (usually fractional Kelly). It's mathematically possible in their case to do better than this but practically infeasible.

In horse racing, you can't calculate your odds exactly as if you were drawing from a deck of cards. The best you can do is use some combination of your odds and actual odds as an estimate to determine the risk involved. Then you can vary your bet size on each bet in accordance with these odds and your expected ROI. But, your expected ROI is usually different in some odds ranges than others. If you know all of these things and you're sure they're not going to change, then you can use some sophisticated mathematics to determine whether it's better to bet on one horse, two horses, or more in a given race and calculate the amounts to bet on each.

Notice that all of these things involve estimates. Now it is true that sometimes estimates are improved by taking large samples, but where the thing you are estimating may change over time, at some point they may actually get worse by going back too far.

Anyway, I guess my opinion is that it doesn't matter how you count things as long as you are very careful not to fool yourself into over-betting in the future. The safest way to do this is to take the most conservative interpretation and assume that the future won't even be as good as that. You'll never see any experienced player betting full Kelly, always 1/2 or 1/3, mainly because it's so devastating to overestimate your advantage. The same is true for blackjack counters even, because counting errors will reduce their advantage.

In your case, I would count those double bets as if they were single bets in a different race since this would be the most conservative approach and they are such a small proportion of your total bets. But don't assume it will be even that good in the future.