Value and Market Forces

[SeeDee]

I've said in the past that I don't know my form from my ar... elbow. But I am a great believer in market forces implying that the shorter the odds the more likely a win. However, as many on this forum have quite rightly pointed out, this (nearly) always leads to low value. MichaelG and others have convinced me that value may exist in the exotics. How to marry these points into a system??

Well, not being a form man and wishing to ride on the back of the "market", I have tried to compare combined win odds of a pair of runners with their Quinella price as late a possible before the start. Using Excel I select upto 12 pairs (not a boxed group) which I dutch bet based on quinella prices. I look for at least 5/1 overall odds and the process takes about 20secs to complete.

This method seems to work for all types of races but has produced some very good days and some very bad. I would like to reduce/eliminate the very bad. Does anyone out there have any suggestions as to how I can eliminate or filter the dud races?

I guess I am looking for logical reasons why a quinella in a race should not reflect the win odds of the horses/dogs involved. Remember I am using knee-jerk, last-minute, calcs and would be looking for suggestions in the area of total pool size, Quinella pool size, field size, etc. as I would prefer the information was already on (currently) the WA Tab screen.

I know this is like asking how long is a piece of string, and I know these issues have been discussed ad nauseum before, but not in the context of Win Odds vs Quinella Odds. Your suggestions don't even need to be specific instructions, just ideas on this comparison. Any thoughts would be greatly appreciated.

[KennyVictor]

I haven't anything concrete to contribute at this stage SeeDee but I am interested in this concept and it's one I have told myself I will persue at some stage. I've got about 2 years worth of quinella divvies downloaded (but not p****d from the webpages into a database yet) against the day I have some theories to test. I'll hope for something to turn up here that's worth testing, it will be the kick start I need to do some useful research.

KV


Now that's clever. I had the word p a r s e d in the stuff I just wrote and when I submitted it I find my pa r s e has automatically had the a r s e removed from it and asterisks put in it's place. I've heard of the ar se dropping out of a market but never out of a p a r s e.

[Bhagwan]

Hi SeeDee,
I have done some research on Quinellas & have noticed that using the paper pre-post market came very close to the eventual quinella divs once I ran my program over the pre-post prices. Amazing

My proram converts the price to odds then does its calculations .

Races with exactly 9 runners presented the best value to risk.

This way , one can lay all their bets at the start of the day.

[michaelg]

Hi, SeeDee.

I have sometimes compared quinella divvies compared with the expected divvy when based on their Win prices, and quite often the actual quin divvy is less than that when calculated from the win prices. This is more true with races of eight or less runners. I have also done the same using Mark Read's prices (approx 140% market) and it seems to be the same. Based on this, I would consider deleting races of eight and less runners. I have also seen huge outsiders run first and second in fields over 16 runners, and the quin divvy pays less than the expected divvy when calculated using their win prices.

If you are waiting until the last moment, I would look at the price of the fave in relation to the size of the field, and also the relationship with the fave and second-fave. If there isn't too much difference between them, and the third fave is nowhere near them in the market I think its a good betting race for quins providing they both don't snare the quinella. But how often does the fave and second fave get the quinella? I could be wrong, but I seem to remember from personal experience about five years ago that the shorter the fave and the bigger the gap between it and the second-fave then the less quinella value there is if the fave runs first or second, which would be generally expected because the market regards it as vastly superior to the rest of the field.

[KennyVictor]

How do you guys work out the expected quinella div from the win prices? I'll show you how I would do it and although it's a bit involved and not perfect I think it would give a good figure.

Say we have a four horse race and I'll keep the market percentage to 100% for simplicity.
Horse 1 is at $2.00 so it has a 50% chance (based on the prices).
Horse 2 is at $5.00 so it has a 20% chance.
Horse 3 is at $4.00 so it has a 25% chance.
Horse 4 is at $20.00 so it has a 5% chance.

To get the quinella on 1 and 2:
Chance of horse 1 winning is 50% leaving horses 2, 3, and 4 to fight out second. I'll now convert the total of their percentages to 100%. I do this by multiplying their original percentage by 100 / (100 - the %age removed)
Horse 2 has 40% ---- 20% x (100 / (100 - 50))
Horse 3 has 50%
Horse 4 has 10%
therefore horse 2 has a 40% chance of coming second.
Combining Horse 1's 50% chance of winning with horse 2's 40% chance of coming second I get (50% x 40%) = 20%.
This means there is a 20% chance of a 1 2 result, (converted back to a 100% market gives me $5 as the value of this exacta).

Now doing the same for a 2 1 result I get the original 20% chance of horse 2 winning and with the same dodgy manipulation of numbers as above I get a 62.5% chance of horse 1 beating the other two. Combine the 20% with the 62.5% and I get a 12.5% chance of a 2 1 result (this gives me a price of $8 for this exacta).

For the quinella I add these two percentage chances together. 20% + 12.5% gives 32.5% chance of getting the quinella. This equates to about $3.08 as an effective quinella price in a 100% market.

I'd be interested to see what price you guys would give for the 1 2 quinella in my example.

KV

[DR RON]

SeeDee, I have one suggestion for you to help eliminate some quinella bets, look for the combinations that have some factor in common such as both first up or both outside barriers or both apprentice jockeys, you will find one of them can finish in the top two but usually not both. This idea I have used mainly with eliminating trifecta combos but I dont see why it wouldnt work with quinellas either.

[SeeDee]

Thanks very much for the input fellows.

I really should exclude 8 runners or less even though this cuts out all dogs which, surprisingly, provides a lot of good returns AND too many bad ones I suppose. I am not trying to anticipate the market so much as take what ever it happens to be and go with the most under-bet pairs. I simply take the top eight in the betting and divide the product of their win odds into their current quinella odds. I sort them by this number (decreasing) and dutch them. If this offers a certain overall return I post the bets.

This obviates the need to look at individual odds, though I do exclude races where the favourite is less than half of the second favourite. Though maybe I should look at this issue more closely Michael.

Typically this "fiddle-factor" (Quin odds/Win1 odds/ Win2 odds) is between 0.4 and 0.8. Would any of you guys have the means to knock out any useful stats on this? Sadly the details of all but the place-getters are no longer available post-race. But is there any value in looking at, say, the average fiddle-factor of past races by race type, or field size, etc.?


KV, following on from your moderator issue: I wonder how they would deal with the English soccer club ****nal..... or even S****horpe Utd ;-)

Good heavens..... sanitised!

[Wunfluova]

Right on the money KV!

<TABLE style="WIDTH: 150pt; BORDER-COLLAPSE: collapse" cellSpacing=0 cellPadding=0 width=200 border=0 x:str><COLGROUP><COL style="WIDTH: 48pt" width=64><COL style="WIDTH: 48pt" width=64><COL style="WIDTH: 54pt; mso-width-source: userset; mso-width-alt: 2633" width=72><TBODY><TR style="HEIGHT: 12.75pt" height=17><TD class=xl22 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; WIDTH: 48pt; BORDER-BOTTOM: #d4d0c8; HEIGHT: 12.75pt; BACKGROUND-COLOR: transparent" width=64 height=17>Comb.</TD><TD class=xl23 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; WIDTH: 48pt; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" width=64>Prob.</TD><TD class=xl23 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; WIDTH: 54pt; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" width=72>Q Price</TD></TR><TR style="HEIGHT: 12.75pt" height=17><TD class=xl22 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; HEIGHT: 12.75pt; BACKGROUND-COLOR: transparent" height=17>1-2</TD><TD class=xl25 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" x:num="0.32500000000000001">0.325</TD><TD class=xl24 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" x:num="3.0769230769230766">$3.08</TD></TR><TR style="HEIGHT: 12.75pt" height=17><TD class=xl22 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; HEIGHT: 12.75pt; BACKGROUND-COLOR: transparent" height=17>1-3</TD><TD class=xl25 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" x:num="0.41666666666666663">0.417</TD><TD class=xl24 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" x:num="2.4">$2.40</TD></TR><TR style="HEIGHT: 12.75pt" height=17><TD class=xl22 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; HEIGHT: 12.75pt; BACKGROUND-COLOR: transparent" height=17>1-4</TD><TD class=xl25 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" x:num="7.6315789473684212E-2">0.076</TD><TD class=xl24 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" x:num="13.103448275862069">$13.10</TD></TR><TR style="HEIGHT: 12.75pt" height=17><TD class=xl22 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; HEIGHT: 12.75pt; BACKGROUND-COLOR: transparent" height=17>2-3</TD><TD class=xl25 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" x:num="0.12916666666666665">0.129</TD><TD class=xl24 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" x:num="7.7419354838709689">$7.74</TD></TR><TR style="HEIGHT: 12.75pt" height=17><TD class=xl22 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; HEIGHT: 12.75pt; BACKGROUND-COLOR: transparent" height=17>2-4</TD><TD class=xl25 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" x:num="2.3026315789473686E-2">0.023</TD><TD class=xl24 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" x:num="43.428571428571423">$43.43</TD></TR><TR style="HEIGHT: 12.75pt" height=17><TD class=xl22 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; HEIGHT: 12.75pt; BACKGROUND-COLOR: transparent" height=17>3-4</TD><TD class=xl25 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" x:num="2.9824561403508771E-2">0.030</TD><TD class=xl24 style="BORDER-RIGHT: #d4d0c8; BORDER-TOP: #d4d0c8; BORDER-LEFT: #d4d0c8; BORDER-BOTTOM: #d4d0c8; BACKGROUND-COLOR: transparent" x:num="33.529411764705884">$33.53</TD></TR></TBODY></TABLE>

Using the formula quinella probability equals :

1/(X(Y+1))+1/(Y(X+1))

where X and Y are the odds of the runners concerned. (i.e. decimal 'odds' -1)

Wunfluova

(formula courtesy of Roger Dedman in Commonsense Punting)

[Merriguy]

SeeDee, the details of all the runners in each race are available for months back on the QTab site

[SeeDee]

Thanks Merriguy. That detail is still only for the place-getters though, unless I'm missing something. What I meant was, on the WA Tab site, before the off, you can view ALL the current quinella odds - and I use this. This information (not surprisingly) is not available from history. So I can't retrospectively test my system which compares all pairs of win odds against their respective quinella odds.

The only retrospective testing that seems available is simply with the ultimate 1, 2 place-getters and the winning quinella odds.
As a variation I am looking at the equation from Wunfluova for value check.

[Merriguy]

Follow now. Sorry for the mis-steer; though, incidentally, I do find QTab the easiest to work with, even if I bet somewhewre else.

[KennyVictor]

This information (not surprisingly) is not available from history. So I can't retrospectively test my system which compares all pairs of win odds against their respective quinella odds.



Oh but it is. www.tabracing.com.au Find a race and then click approximate divs if I remember rightly.

Love your soccer names - hee hee.

KV

[Merriguy]

He's right, you know. Good one, KV :)

[SeeDee]

Well done KV. Just have to work out how to manipulate them for Excel. Unfortunately the main screen seems to be a single entity so I can't copy and paste into Excel. But, hey you can't have everything.

Cheers

[Wunfluova]

SeeDee,

What version of Excel are you using? I am using Excel 2003 so I don't know if earlier versions can do this, but all I have to do is right click on the quinella dividend matrix and select 'Export to Excel' and the quinella prices are inserted into an Excel sheet as displayed on the Tab site.

Failing this you still should be able to select the appropriate info for a copy and paste into Excel then select 'match destination formatting' for the same result. (Unless this is also a feature of later versions of Excel)

Wunfluova

[SeeDee]

Hey Wunfluova,

I am using Excel 97 and I don't get that option. Time to upgrade. I can deal with the Quin odds with a bit of effort - BUT the main screen (from which you select "approximate divs") is not Excel 97 friendly. Can you do the same with that screen? If so, I will definitely upgrade.

[Wunfluova]

SeeDee, I can't select (left click) the prices on that page to copy and paste, but the right click and 'export to excel' option still works fine. It downloads all the price data showing on the page - price fluctuations etc plus all the results and exotic info on the page above the field and price info.

Wunfluova

p.s. If you right click on the page do you get a menu which includes 'view source'? If so this will get the figures for you in amongst all the html tags etc so the info is there but it is probably going to be a nightmare trying to isolate the 'good stuff' from the 'bad stuff'. Easier I think to upgrade your Excel.

[SeeDee]

Easier I think to upgrade your Excel.Agreed. Many thanks Wunfluova.

[KennyVictor]

Using the formula quinella probability equals :

1/(X(Y+1))+1/(Y(X+1))

where X and Y are the odds of the runners concerned. (i.e. decimal 'odds' -1)

(formula courtesy of Roger Dedman in Commonsense Punting)

Love that formula Wun, does Roger have any more gems you might like to share, like a similar one for trifectas or boxed trifectas?

KV

[Heavyweight]

The attached isn't pretty, but it is what I made up to predict correct quinella and trifecta dividends.

I entered the numbers and expected dividends as per KennyVictor's example in the shaded area, and the expected dividends are shown to the right for each combination.

I've given up on trifectas (just can't nail enough), but my experience with my winning quinella bets is that they pay better than suggested about 90% of the time. So they should, given the reduced TAB take for the pair (compared to win dividends)

[KennyVictor]

Typically this "fiddle-factor" (Quin odds/Win1 odds/ Win2 odds) is between 0.4 and 0.8. Would any of you guys have the means to knock out any useful stats on this? Sadly the details of all but the place-getters are no longer available post-race. But is there any value in looking at, say, the average fiddle-factor of past races by race type, or field size, etc.?


Hi SeeDee,
I've fiddled with the fiddle factor on races since feb 04 with the following results:
0 to 8 Horses ----- 0.4464
9 - 12 Horses ----- 0.4592
13 + Horses ------ 0.4704.
Not much difference there.

I also found the average fiddle factor for different quinella prices and they are as follows.

Quinella 0 to $6, Avg fiddle factor 0.4207
6 - 12-------0.5001
12 - 25--------0.5026
25 - 50--------0.5079
50 - 100-------0.4914
100-200-------0.4591
200+ ---------0.4041
There is a regular sort of curve formed by the results with quinellas in the midrange having a higher fiddle factor. To me that means the short priced horses and the long priced horses are overbet in quinellas compared to the mid-priced ones.
Interesting even if it's not useful. :-)

Heavyweight,
Belated thanks for those trifecta formulae.

KV