Money Mgt for spot plays

[teddy]

I have 50 spot plays that I use. Ideally I bet $100 on each one when it shows up. The issue is the same horse can show up 10 times due to overlap. Should I wager 1000 on this horse (obviously not) or some other figure. What if one shows up twice? Or should I just wager $100 on each horse regardless of frequency. Im sure others have this same issue. I bet to place and strike at .95 roi and use the rebate to push over...TMI maybe. Maybe there is one of the managers here that can at least give me a confidence to bet size ratio.

[GameTheory]

I have 50 spot plays that I use. Ideally I bet $100 on each one when it shows up. The issue is the same horse can show up 10 times due to overlap. Should I wager 1000 on this horse (obviously not) or some other figure. What if one shows up twice? Or should I just wager $100 on each horse regardless of frequency. Im sure others have this same issue. I bet to place and strike at .95 roi and use the rebate to push over...TMI maybe. Maybe there is one of the managers here that can at least give me a confidence to bet size ratio.Well, technically you should bet each spot independently (since that is how you presumably verified them in the first place), and so yes you should bet $1000 if a horse fits 10 spot plays or else you risk losing even if the spot plays are winning (because what you are really doing is betting on one play and skipping 9 of them, or only betting $10 on each, or something like that). And if the fact that the overlap exists also means that horse is a better play than if it only fits one profile (common sense says it does, but reality is often otherwise), then you're deliberately betting less on the best plays and you're going to end up losing.

What you really need to do is generate some conditional stats so you can judge each play individually. You'd probably find that some of your spot plays are actually losers if you remove the overlaps, i.e. if you have spots A,B,C, & D; you might find that A is an overall winner, but if you take out the As that are also Bs, it is a loser and it is really B that is the profit generator and just happens to correlate with A often.

In absence of such analysis, I'd lower the base bet amount to maybe $20 or whatever amount would make it so your average bet remains about the level you want, but you bet more on some and less on others depending on the overlap.

[ALL CIRCUITS GO]

I love the explanation.. easy to understand.. thanks.. I had a similar situation, not at $100 level, but had the same idea, and doped it out myself exactly the way you explained it! Each spot is independent, so each spot gets its own wager.. 1x to 10x....

[teddy]

That makes my base wager drop on mynon overlap plays. Total handle would plummet . I need to look at what overlaps regularly. Some will because of last fraction being a big factor. Currently avg handle has been slightly more on the overlaps when pool sizer permits.

[Robert Goren]

You might try the SQ root method. Take the SQRT of the times your spot plays comes with the same horse and then multiply it by your bet size. That way you bet more but your bankroll doesn't take major hit when it doesn't cash.

[Jeff P]

I used to ask myself this same question all the time.

It took a while, but eventually I decided to handle it in a manner similar to what GT posted:What you really need to do is generate some conditional stats so you can judge each play individually. You'd probably find that some of your spot plays are actually losers if you remove the overlaps, i.e. if you have spots A,B,C, & D; you might find that A is an overall winner, but if you take out the As that are also Bs, it is a loser and it is really B that is the profit generator and just happens to correlate with A often.
Expanding on this a little...

Map out the permutations involved and treat intersections of models as individual models.

Example:

You have three UDMs or Spot Plays. They are named A, B, and C.

Permutations involving UDM intersections can be listed out in the following manner:

A
A+B
A+C
A+B+C
B
B+C
C

From here, suppose you wanted to see past history/what happened in races where a horse was flagged by both A and B.

Armed with a database, the player could run the following query:

Select * from historytable
where ((rules_for_ A) and (rules_for_B))

Results returned by the above query will give you a good idea of what to expect when a horse qualifies for both A and B (the intersection of A and B.)

And of course you could (given enough time) run queries for any and all possible permutations involving your individual models - and get a reasonable idea of what to expect from each of the possible intersections.

You don't necessarily need a database. You can do something similar with a good set of manual records (although it takes longer to get the questions answered.)

One interesting thing I've discovered is that answers from this type of R&D aren't always what I expected going in.

Depending on what goes into models A, B, and C (above) each might be good when evaluated individually. However, specific intersections (for example A + C) might be weaker than A all by itself.


-jp

.

[GameTheory]

One interesting thing I've discovered is that answers from this type of R&D aren't always what I expected going inI've found that they are NEVER what I would expect going in, so I stopped expecting and just see what it tells me.

[teddy]

In the past I took the simple method of getting an roi for each play based on number of times it appears. Easy to do and I did gleam that if they showed up about 10 times they paid 2.40 to win usually. Bridgejumper plays.

[dansan]

2.40 to win PASS

[cj]

Over time, I learned to set up my plays so that there are no overlaps.

[Robert Fischer]

if you want to repeat wagers on overlapping spot plays, it has to be bankroll dependent.

If the horse is a "play" for me, on any of my spot plays, the first time i wager with it, I already bet as much as I safely can.

there may be twists to this of course, for example - let's say you have this horse overlapping for a lot of plays and in the "big picture" you can now see that this horse has a greater hit rate, you may increase the wager.

For my specific style of play, i find that the first time a spot play points out a potential wager for me, I then go and make an accurate estimate of the horses strike rates, estimate the pools, and see what the "best" play(s) are likely to be. I find that once I do that I have an accurate view. However; because of the nature of some of my spot plays that happen to be based on physicality or trips, -if I "discover" these as I go along, my estimates of the horse's strike rates occasionally change significantly. But in the end I make one play or pass on the horse, using the cumulative insight.

a variation to this theme that I do find occurs frequently for me, is rather than multiple wager triggers, i see multiple wager TYPES. I then have to consider whether to use only one(1) wager type(or wager pool), or to "OVERLAP". What I end up doing, is estimating everything and finding the best solution(which may end up being 1 OR several wager type(s)!), and my secret recipe for the overlap here, is to treat it as ONE big single wager in terms of TOTAL bankroll allotment, although the different amounts for different wager types on that same horse will vary significantly.
make any sense?

[highnote]

The size of the bet is dependent on the size of your bankroll, the probability of winning, the payoff and the pool size.

It makes no sense to bet ten $100 bets into a $5000 pool if the horse is 20-1 at post time and you think it should be 10-1. If you bet it, you're going to knock it down to 4-1 -- below your estimated odds.

Similarly, if the horse is 2-1 and you think it should be 5-1 then you probably shouldn't bet.

I suspect you probably do some of this in your head already -- like estimating the pool size. You can probably tell how much you can safely bet without affecting the odds.

To figure out the optimal bet is complicated. It requires software to do the calculations. Without software the best you can do is estimate.

My rule of thumb is to bet 4% of my bankroll. But that is a gross estimation. Sometimes it can be very wrong -- resulting in too high of a bet or too low of a bet.

You might google mean variance analysis. Some stock market investors size their investments using this technique.

[teddy]

What is the most you can bet into a $5000 place pool without effecting the odds by more than 1%.. Considering the horse is the fav and is even money. and has 40% of the place pool wagers on him.

Im guessing $40

[Robert Fischer]

The size of the bet is dependent on the size of your bankroll, the probability of winning, the payoff and the pool size.

It makes no sense to bet ten $100 bets into a $5000 pool if the horse is 20-1 at post time and you think it should be 10-1. If you bet it, you're going to knock it down to 4-1 -- below your estimated odds.

Similarly, if the horse is 2-1 and you think it should be 5-1 then you probably shouldn't bet.

I suspect you probably do some of this in your head already -- like estimating the pool size. You can probably tell how much you can safely bet without affecting the odds.

To figure out the optimal bet is complicated. It requires software to do the calculations. Without software the best you can do is estimate.

My rule of thumb is to bet 4% of my bankroll. But that is a gross estimation. Sometimes it can be very wrong -- resulting in too high of a bet or too low of a bet.

You might google mean variance analysis. Some stock market investors size their investments using this technique.

good post.

You can probably tell that we agree on much of the factors that go into sizing the bet.

Added "variance analysis" to my weekend to-do list. not sure if incorporate any of that or not, I'll have to look into it and either refresh or learn it for the first time, to know.

- I happen to use the factors we discussed and then use a maximum-probable-consecutive-loss-formula to determine the wager size. When making just one type of bet, I get pretty familiar with the wager sizes over a given bankroll, so I don't have to plug in the variables to excel every time and sometimes I just estimate it.

[Robert Fischer]

What is the most you can bet into a $5000 place pool without effecting the odds by more than 1%??.. Considering the horse is the fav and is even money. and has 40% of the place pool wagers on him. ??

Im guessing $40

bumping this for you. didn't mean to bury a fresh question.

as far as answering it - good question. kind of having a brain freeze lol @ the moment(coffee plz!), but 1% sounds like you may be splitting hairs(whoever tackles this may prove me wrong here!), things like "breakage" come to mind here where 1% may or may not be a little too fine... Either way I'm interested in a refresher course on the math behind POOL IMPACT.

[BIG49010]

Ed Bain had something about this in his book, off the top of my head, I believe.

If you have multiple angles for horse "a", and only a single for horse "b", I think he gravitates to the more obscure play which would be "b".

I have played my angles this way, and it has worked out for me.

[pondman]

I have 50 spot plays that I use. Ideally I bet $100 on each one when it shows up. The issue is the same horse can show up 10 times due to overlap.

Are these 50 spot variables independent of each other? Are all of the spot plays positive independently?

I'm a strict spot player. I don't have this problem. I have more of an issue with two horses showing up in the same race-- and then I'm more concerned with the odds..

[jdhanover]

What is the most you can bet into a $5000 place pool without effecting the odds by more than 1%.. Considering the horse is the fav and is even money. and has 40% of the place pool wagers on him.

Im guessing $40

Have to make some other assumptions. If takeout is 17% and the other horse that places had 20% of the pool ($1000), then I think you could bet $70 and the change is 1%.

Note that is the other horse has only $500 (10%) on it, then your bet needs to drop to $60 to hold to -1%