Dutching Win Bets

[Nitro]

Originally Posted by BCOURTNEY

I don't believe the odds are that volatile at all, I think that perspective at this point (after years of awareness) players not adopting their play to the environment and utilizing technologies or simple worksheets solutions available or they are simply lazy. Yes the tote technology is antiquated and needs upgraded as well.

It’s not that the Tote technology is antiquated at all. The problem is the way players have been so indoctrinated to using just the odds (the Win pool) in their evaluations of what’s going on during a typical betting cycle.

Even the on-going concerns and discussions about the tote changes after the gate opens is at best misconstrued.
For instance, if the tote is updated every 60 secs., and the gate opens - how far would the horses travel in a 6F race to see the final tote update? Even with 30 sec update they would gone over a ¼ mile!

[ubercapper]

Originally Posted by BCOURTNEY

Claim that your estimated cumulative probability of both horses winning is X (Let's say 30%), show that the odds offered when combining both are likely to reflect X (20%), suggest weighted win bets for each. This demonstrates a positive EV bet. Argument done.

Where were you when I needed you on Twitter recently?

Thanks, too!

[ubercapper]

Originally Posted by BCOURTNEY

Claim that your estimated cumulative probability of both horses winning is X (Let's say 30%), show that the odds offered when combining both are likely to reflect X (20%), suggest weighted win bets for each. This demonstrates a positive EV bet. Argument done.

Since my math skills are not as sharp as they once were:


My estimated combined probability of both horses winning was 50%.
The odds of the horses were 7/1 and 9/1 so I think that equals 22.5%.
I believe the weighted win bets should be 55% on the 7/1 shot and 45% on the 9/1 shot.



Did I get that right?

[ultracapper]

Therefore, any statistical model, however well developed, will always be incomplete. An extremely important step in model development, and one that the author believes has been generally overlooked in the literature, is the estimation of the relation of the model's probability estimates to the public's estimates, and the adjustment of the model's estimates to incorporate whatever information can be gleaned from the public's estimates. The public's implied probability estimates generally correspond well with the actual frequencies of winning.

So this is Bill Bentor's way of saying the favorite wins more than the 2nd fav, who wins more than the 3rd fav, and so on and so on. I mean, duh.

[Nitro]

Originally Posted by ultracapper

So this is Bill Bentor's way of saying the favorite wins more than the 2nd fav, who wins more than the 3rd fav, and so on and so on. I mean, duh.

No, what he's actually implying is in the previous paragraph:

Additionally, there will always be a significant amount of 'inside information' in horse racing that cannot be readily included in a statistical model. Trainer's and jockey's intentions, secret workouts, whether the horse ate its breakfast, and the like, will be available to certain parties who will no doubt take advantage of it. Their betting will be reflected in the odds. This presents an obstacle to the model developer with access to published information only. For a statistical model to compete in this environment, it must make full use of the advantages of computer modeling, namely, the ability to make complex calculations on large data sets.

The odds set by the public betting yield a sophisticated estimate of the horses' win probabilities.

Again, I don't like the use of the generic word "public" when it comes to identifying the actual betting population.
.

[InsideThePylons-MW]

Originally Posted by ubercapper

Since my math skills are not as sharp as they once were:


My estimated combined probability of both horses winning was 50%.
The odds of the horses were 7/1 and 9/1 so I think that equals 22.5%.
I believe the weighted win bets should be 55% on the 7/1 shot and 45% on the 9/1 shot.



Did I get that right?

You gave both horses the exact same 25% chance of winning.

If you think that betting $11 to win on the 7-1 and $9 to win on the 9-1 is optimal then just quit now.

Anybody with one ounce of wagering common sense knows that $20 to win on the 9-1 will outperform the above in the long run based on your opinion.

Go ahead and keep betting the most on the horse with the least edge.

[AltonKelsey]

I see where he gives a combined prob of 50% , doesn't break it down .



In that case his bet is fine




Case can be made for cash flow optimization, so knee jerk 'bet it all on the biggest overlay' is short sighted.

[cj]

Originally Posted by AltonKelsey

I see where he gives a combined prob of 50% , doesn't break it down .



In that case his bet is fine




Case can be made for cash flow optimization, so knee jerk 'bet it all on the biggest overlay' is short sighted.

He did earlier in the thread, said he thought each had a 25% chance to win.

[AltonKelsey]

Originally Posted by cj

He did earlier in the thread, said he thought each had a 25% chance to win.

ok, didnt look back that far.


my point about cash flow still holds.



Take an extreme case for illustration.


5-1 and 40-1



5-1 is a 20% overlay and the 40-1 is a 33% overlay (should be 30-1)





You gonna pass on the 5-1 and bet it all on the 40-1 , cashing once out of 30 times ?





Don't think so. Though that's what was suggested , as if anything else made you an idiot.

[MJC922]

The post isn't from the mid-90s so I'm still trying to figure out how anyone is getting the odds they're accepting in the first place. I'll say it again. Make a line and figure out how often horses who are overlays when you actually need to bet on them i.e. 0MTP 1MTP use whatever MTP you want to use but let me know how many of them are still overlaid after the race is over. The more accurate your line is the more often they will end up being underlays. As I recall for me it was right around 40% ended up underlaid when betting at 0MTP. That was using real $2 bets, not 'on paper' or theoretical. Is it any wonder the bottom line will be negative? The degree of overlay didn't matter, use 50% overlays 100% overlays, doesn't matter, same result.

So the wagering piece, Kelly and all the rest, hasn't been viable IMO since the mid 90s. For exchanges it's viable. Someday it may even be semi-viable again for the tote but it sure isn't for most people these days.

Unless of course you can bet later than most, in that case you should be fine. Aside from that there are ways to find inherent value, like what a vet knows is going to be built-in value. However what the outsider (the handicapper) knows better than the rest will apply to such a limited number of events these days that IMO it would be questionable that it's worthwhile to pursue unless you're actually working (drawing a paycheck) in the industry.

As for dutching of course it's valid. I did it regularly when I played for a living but it was done typically when something else was being way overbet, in that case I shopped the board for easy money. It would've been in the action bet category, it wouldn't have been a well-planned bet where two horses I was waiting to bet back were in the same race at high odds.

[cj]

Originally Posted by MJC922

The post isn't from the mid-90s so I'm still trying to figure out how anyone is getting the odds they're accepting in the first place. I'll say it again. Make a line and figure out how often horses who are overlays when you actually need to bet on them i.e. 0MTP 1MTP use whatever MTP you want to use but let me know how many of them are still overlaid after the race is over. The more accurate your line is the more often they will end up being underlays. As I recall for me it was right around 40% ended up underlaid when betting at 0MTP. That was using real $2 bets, not 'on paper' or theoretical. Is it any wonder the bottom line will be negative? The degree of overlay didn't matter, use 50% overlays 100% overlays, doesn't matter, same result.

So the wagering piece, Kelly and all the rest, hasn't been viable IMO since the mid 90s. For exchanges it's viable. Someday it may even be semi-viable again for the tote but it sure isn't for most people these days.

Unless of course you can bet later than most, in that case you should be fine. Aside from that there are ways to find inherent value, like what a vet knows is going to be built-in value. However what the outsider (the handicapper) knows better than the rest will apply to such a limited number of events these days that IMO it would be questionable that it's worthwhile to pursue unless you're actually working (drawing a paycheck) in the industry.

As for dutching of course it's valid. I did it regularly when I played for a living but it was done typically when something else was being way overbet, in that case I shopped the board for easy money. It would've been in the action bet category, it wouldn't have been a well-planned bet where two horses I was waiting to bet back were in the same race at high odds.

The other way would be to have a fairly reliable method of predicting the actual final odds. It won't be perfect, but it is doable and better than what you see on the tote board with 0 minutes to post.

[thaskalos]

Some winning players bet for "growth"...so they employ "optimal" methods of betting which maximize the ROI and the long-term growth of the bankroll. Other winning players bet for "income", and they employ suboptimal methods of betting which introduce dutching and hedging techniques into the wagering equation...thus sacrificing some long-term growth for the benefit of reducing some of the more immediate risk that a gambling enterprise must endure. What seems like a "mistake" to some may be readily acceptable to others. We don't all have the same wagering goals...that's why the game offers such a varied menu of betting opportunities.

[MJC922]

Originally Posted by cj

The other way would be to have a fairly reliable method of predicting the actual final odds. It won't be perfect, but it is doable and better than what you see on the tote board with 0 minutes to post.

Possibly. Would have to be tested. I would want to know how closely for example horizontal will pays are to final odds on the final leg or if for example we're hooked to the tote api how closely combined action in all pools at 0 MTP lines up with final odds. Admittedly I've never tested either of those scenarios. I strongly suspect betting later than most is key to doing something on a large enough scale these days to make it worth my while.

I believe what we see happening is a reflection of certain large bettors making the odds very very late and I suspect in most cases trying to discern that beforehand won't be possible.

[BCOURTNEY]

Originally Posted by ubercapper

Since my math skills are not as sharp as they once were:


My estimated combined probability of both horses winning was 50%.
The odds of the horses were 7/1 and 9/1 so I think that equals 22.5%.
I believe the weighted win bets should be 55% on the 7/1 shot and 45% on the 9/1 shot.



Did I get that right?

Assumption: You agree fundamentally with the Kelly formula.
Assumption: The win estimates are 25% for each horse and accurate

Your edges would be:

14.3% on the 7 to 1
16.7% on the 9 to 1

If you are not risk adverse and committed to full Kelly and you choose to invest your edge, you would invest 14.3% of your current bankroll on the 7 to 1 or 16.7% on the 9 to 1 or 7.8% on both. Common variations are partial Kelly approaches which mitigate risk involve some flavor of edge/odds offered e.g. 14.29/7 = 2%, 16.67/9 = 1.8% of bankroll. These types of approaches seek to reduce variance because of the inherit inaccuracy of win estimates. Some players seek to increase churn for rebated play some prefer a sniper style approach some a combination of both.

Some caution should be exercised in betting every slightly positive EV bet as this can lead to issues as well as some bet types are mutually exclusive and the effect can be accounted for by including a reserve rate (more advanced, discussed in mathematical literature on this topic) and the ordering of the bets is considered until the reserve rate is met. Some approaches discourage calculating the reserve rate and executing multiple plays and focus exclusively on the highest EV play

[BCOURTNEY]

Originally Posted by AltonKelsey

ok, didnt look back that far.

my point about cash flow still holds.

Take an extreme case for illustration.

5-1 and 40-1

5-1 is a 20% overlay and the 40-1 is a 33% overlay (should be 30-1)

You gonna pass on the 5-1 and bet it all on the 40-1 , cashing once out of 30 times ?

Don't think so. Though that's what was suggested , as if anything else made you an idiot.

This is an excellent point, time horizons are important as well. For example a player might not want to play a 30000-1 pick 6 ticket with a positive EV due to the fact the combination likelihood of being physically dead before the wager realizes itself. In fact one can argue if the player is reinvesting winnings, there is a time horizon versus value to compounding returns which is not evident nor accounted for in most strategies.