When Ratings Meet Systems
 

About 8 years ago I came across a publication which, inter alia, described a method of punting which the author claimed produced a POT of +/- 19%. Being somewhat greedy and intrigued by the return I attempted to reproduce the results over a couple of years but never managed to get near the target.

I put the method aside for a few years until I had accumulated a sufficiently accurate and complete database of results (+/- 12 years) and the ability to run complex queries. The method was then re-examined.

It is NOT my intention to pooh-pooh the claims of the author of the method as he/she has somewhat of a reputation in the industry and is apparently held in respect by his/her clients. My aim is to demonstrate what happens when one looks at the long term rather than rely on short term results as the basis of published claims.

The Method :-
(a) Going MUST be Slow or better.
(b) Distance MUST be greater than 1200 metres.
(c) Favourite TAB Price MUST be <$4.00 (Fixed)
(d) Fav. MUST be 1st or 2nd Rated in My Fine Ratings (MFR's)

If the above conditions are satisfied then Dutch the Top Two MFR's.

Run over a "couple of months' on weekdays a POT of >17% is claimed.

I DO NOT hold the MFR's for 12 years. However, what I can demonstrate is that regardless of what the MFR's are (always including one of the Top Two as Favourite) the Method is a LOSER (and will always be a LOSER) over more than a 'couple of months'.

Check Method used :-
(a) 7 days per week
(b) Going Slow or better
(c) Race Dist. >1200 m.
(d) Races 4 - 8 incl. (to eliminate weak early races)
(e) NO Maiden races (reason as above)
(f) Fav. TAB Price <$4.00
(g) 8 to 12 starters ONLY (to improve results)

Based on the above criteria the 12 year results of the Favourite are :-

32.05% Win Strike Rate @ Av. TAB of $2.73

So we now know that the first 'leg' of our Dutch will have a WSR of +/- 32.05% which requires a TAB Price of $3.12 JUST TO BREAK EVEN. When our MFR is Fav. it's going to have an SP of +/- $2.73 so we're already in trouble.

The proponent will then comment that it's the 2nd leg of the Dutch that will make up the difference and the Fav. is just there to help break-even.

Let's examine that proposition.

If the Fav. (#1 in the Odds queue) doesn't win then in a 12 horse race (to take the
worst example) I can expect the 6th Odds Ranked horse to win 5.2% of the time and
I'm covering 95% of the possibilities.

So, by doing a cross run of the Fav. versus ALL other contenders from 2nd to 6th Favourite and calculating the Dutches we should be able to detect where this claimed profitable area originates.

FAV 2 FAV 3 FAV 4 FAV 5 FAV 6
Strike Rate 18.88% 13.55% 10.72% 8.38% 6.37%
TAB Price $4.56 $6.06 $7.75 $9.99 $12.90
Dutch SR 50.93% 45.60% 42.77% 40.42% 38.42%
Dutch Need $1.96 $2.19 $2.34 $2.47 $2.60
Dutch Actual $1.71 $1.86 $1.99 $2.18 $2.25

What the last two lines of the table reveal is that the Fav. combined with ANY of Faves 2 to 6 is going to produce a Dutch LOSER at TAB Prices.

Think about what this actually means in terms of the 'method' run over a 'couple of months' ..........

Hi Demo - sounds like his"My Fine Ratings' could be the problem.
I use something similar by betting the top two but the second of the top two has to be at least six points clear of the third selection in my ratings.
Of course the approach I use would also depend on how the points are allocated and over what range. But what I do tries to ensure that the selections are standouts and not close to the other ratings.
Based on MFR I could be in trouble today as I have eight races weeded out and all have the fav in my top two.

Hi ,
I gather your findings are based on every race on the programme.

Interesting and valuable findings.

Cheers
darky

I think it's reasonable for a system to work for only a few months. The market is always changing. When I am being disciplined I re-examine my selection criteria every 2 to 4 weeks, it's almost as if I'm running a new system every time.

Why do people publish systems anyway? It's a conflict of interest.

Darky could be on the right track because if you blindly bet every race you are bound to lose out.
My eight races worked for me today because they were eight from approximately thirty races that I could have bet keeping in mind I only bet Ratings 70 and above. Five of those races were winners BUT Flem R7 got home by a pimple so that turned a good day into a much better day. But I think selection methods and the race selection are two areas that we should be looking at.

You are right about race selection, i think that is 50% of the challenge, over the years i have worked through the hard stuff, pricing was a big hurdle for me but i am very happy with the formula i have now, race selection is my next challenge, i used to think to stick to higher class races but am finding i can find the hidden gems in the lower class races more easily, not sure, still a work in progress.

G'day folks,

Please remember that this refers to Dutching Systems ONLY.



This morning I started to build a box cross diagram of all TAB ranks from 1 - 6 run as a Dutch .... ie you could cross Rank 6 with Rank 2 and see what happens. I did this to demonstrate to those of you who 'select' races that, over time, it'll still go tits-up. After spending an hour or so I simply gave it away. It was an increasing loser dependent only on Odds rank. The bigger the Odds cross rank additor the worse the result.

This was done using some 2 million results.

Mathematically, even the best Dutch Race Selectors can't dodge the bullet for ever.

The REAL skill is knowing when to quit Dutching and try something else , and that time is when we're in profit .... because over the longer term we lose.

Why ????? Think about the +/- 18% TAB take-out.

Meanwhile, back at the ranch ........................

I still say the problem with what you are trying to do here is the method for selections is flawed or Race selection is wrong - or both. If anyone thinks they can take the favourite for every race and couple it up with any of the first six in the market they are doomed from the start. There has to be some logic used to select what is considered to be a contender for the winning spoils not just because it happens to be the race favourite.

Nicely put.

The sad thing is that WHATEVER the pairing is (Fav1 to Fav 99) as a Dutch with Fav 2 to Fav 99 using TAB odds we're going to lose. Just do the numbers.

The Race / horse choices are irrelevant over the LONG term. Over the SHORT term (say 100 selections) then there's a possibility of success by spending several Friday evenings studying form :-)

Yep, been down that path ages ago.

There is a strategy though to tweak the Odds before the Dutch Calc entry.
May be used as a filter, or I termed it personally, "Value_Leverage".

You use a Variable increase/decrease percent of your 3 prospective dutch selections, so for example if you have a runner @ 5.00, then Decrease the 5.00 by 10,15, 20 %, it's up to each one's choice, then enter that price into the Dutch Calculator.
Dutch 3 runners, lowering their given odds by a Variable_Percent.
Then your filter (or one of them ) is the Cost.
Hence the leverage if it's a bet or no bet, cost is the determining factor.

It works for the Target_Dutch method,
ie: Make 25.00 units Profit, not the other way round, spread 25.00 on the Dutch.

OK, so assume the Make_25_Units_Profit is your base, Bet 1.
Then you use another percent, the Cost_Factor that is after the Dutch Calc. That to is a Variable_Percent filter.

So the cost of the dutch after the prices have been lowered by 20% before the dutch calc entry, if that cost is let's say more than 20 units to make 25 units profit, then discard the bet, there is no "Value" in that particular race or bet for a Dutch_Bet/Target_Profit strategy.
The variable percentages is just playing around with the leverages, when to bet, when not to bet, and when to bet or not before the start of the race in fluctuating market.

Of course you can play with the Increase/Decrease percentages till the cows come back from Pluto, at the end of the day it comes down to Method_of_Selection, the Favorite_Ascending anywhere on the scale as you described box cross method, yeah, does not seem to work long term, BUT, technology has advanced since back then when it was being tried,
perhaps you will discover an edge.

From experience and some loose statistics;
6 strikes/losses and ya already maybe looking at 2000 Units required to chase the loss before a win.
Unless the odds are very good, 5's and up, losing on the 7 consecutive dutch_loss_chase_bet, you need a long shot selection method by then.

I got stats and stats of this stuff on a broken HD, there's at least 2 forum members that know of this experiment that was done ages ago.
At first it seemed successful with a, Stop_at_this_profit, or Stop_at_3_losses.
WE tried every angle till the machine/s broke down,
( Teletext days for data_input was the only thing that stopped this search for the hint of a grail, the phasing out of that old data_source technology to feed the number cruncher also stopped the search for a strategy.
The programmer at the time gave up went and got a job.


Good Luck !