Why boxing exotics and Dutch books are fundamently flawed

[Dirk Gently]

If you box 3 horse in a Quinella you place three bets: only one of these can collect, two must lose. What you are doing is betting against yourself. Dutch books are the same thing, when you bet more than one horse to win or three horses to place you are accepting that you have just made a losing bet.
However, the psychology of punters mean we like to win often and lack patience so rather than bet six trifectas on six different races we box three horses in one race. In the first instance it is possible that all six can return a dividend but when you box a trifecta you can only collect on one unit.
Don Scott recommended putting multiple trifecta combinations in a race but I suspect the overlays in those days were greater than now and he could afford to waste money on losing combinations.
Please correct me if I'm wrong, it does seem such a simple fact that I'm surprised so many people overlook it every day.

[Chrome Prince]

Fundamentally right Dirk, but the key is value here.
While two out of three must lose (except in a deadheat), the one should pay much more than 2/1 if a horse at good odds is included.
If I like a horse at 6/1, I would not be boxing it with the favourite and second favourite - I'd consider putting in something at odds capable of running a place and something at around 5/1 or greater.
Therefore should two of the three run 1st and second, I'd recoup my investment and achieve possibly better overs than backing my horse straight out.

The same goes for Quadrellas and similar.

<font size=-1>[ This Message was edited by: Chrome Prince on 2004-09-08 12:37 ]</font>

[moeee]

Dr. Dirk.
I reckon the way to get round the box quinella drama is to bet proportionate units on each combination as to the chance the bettor gives it.
Not too hard to do with 3 runners.
Bit of mucking around with trifectas if you have over 3 selections.

[Dirk Gently]

Yes, mooee, the general consensus is that you should back combinations to win the same amount i.e. more on short odds and less on long shots. Incidently I have written an excel formula to calculate what quinella odds should be, quite handy.
However, I still contend that you should back 30 two-horse quins than 10 three-horse quins. By boxing three you are throwing 3 units in the pool and only hoping to collect on one unit. The thing is that it takes three times longer to rate 30 races than ten, so the time-poor punter likes to get as much action per rated race as they can.

[moeee]

Dirk Gently.
Consider your 30 races.
Let them be the next 30 races to be run.
So that's 4 meetings with about 7.5 races each.
Of them 8 are maidens,8 lower class events and maybe 4 are not worth considering for various reasons.
We now have 10 races left to bet on with more time to study each race more deeply and 20 more units to invest.So now we can box 3 in the Quinella.
Better still take our 2 best in each race and invest 2 units,and 1 unit with our third selection into our 2 best.
Quite often our 3rd best selection is at large odds and can give us a large collect.

[Dirk Gently]

I bet Saturdays mostly, last Sat I looked at 5 meetings, discarded 13 races, backed 27 Quins and my only return was a combo returning $102. The favourite normaly pops up in my top two selections so I'm mostly looking for lower returns but its nice when your only collect for a day is a couple of longshots. It just takes a bit of nerve to sit through 20 losers in a row, entertaining mind! Cheers!

[thebookie]

Dirk,

Without giving too much away, can I ask what formula you use to calculate quionella's?

Cheers,

thebookie

[Shaun]

the most commom method is to multiply the win dividend of your 2 runers then devide by 2
example

goulburn race 2 the winner was paying $3.60 the second horse was paying $7.70 unitab
$3.60x$7.70/2=13.86
quinella paid $18.60
so the quinella paid better than expected

you can do the same for an exacta but deviding by 1.2

don't forget this is subject to the price of the fav if it wins and the size of the pool...i also have one for trifectas that is just as accurate if you want it

[Dirk Gently]

What I do is re-rate the odds from a (average) of 118% back to a 100% market. Then you convert the odds of your first horse into it's chance of winning(this assumes the market is usually right) then for your second horse you do the same but take the first horses chance of winning the race from the 100%. Then you multiply the chances of the two to get the probability of them running 1st and 2nd in that order. Do the whole thing again but for the other quinella combination and add the two probabilities and convert back to a price. You can see why I wrote it in excel!
I'm happy to email the equation, ask me at ewan@phaedrus.com.au.

[thebookie]

Thanks guys.

Email sent.