Firstly using the ten year stats. we identify the race selection - the second step is how to establish the correct Dutch Frame in any particular race.
Race selection is 90% of the way there but there is a huge problem left before we can generate a profit long term.
Question: How many runners to dutch in any particular race?
There are many many Dutching betting methods which have appeared over the years, almost always with insufficient proof of results. You need to see a sample of 1000-1500 results to verify a possible long term edge even when you are covering several runners in the field at a very low odds equivalent. The Dutching methods out there nearly all fail because they use 'static' methods of race analysis; cover 3 runners, select races with max 'X' runners, two runners below price 'Y' etc.
And where does this leave tipsters providing a few months past results on single selections at long odds? A few months doesn't hack it you need large samples to verify a small edge and the numbers escalate v. significantly with increased odds selections.
Using the ten year trend stats for the race selections even the low frequency races such as Non H'cap Chases are off good sample sizes.
The principle I used to answer the How Many question?
Let the Shape of the Market tell you how many runners to include.
Once you place a number of Dutch bets it becomes clear that the 'shape' each market develops is complex. Some races with a large field would have say just two runners in contention with the remainder at long prices through every imaginable combination out to races where every runner is in contention and there is no clear low priced favorite? Finally at the opposite extreme races where the fav. is so short its not possible to place a Dutch bet (and a percentage of these v.short favs. still get beaten sufficient to generate a loss backing them)
So there are two strands which follow on to arrive at a market 'tell'
- Firstly establishing a Dutch Ratio. The return on the bet as a ratio of the total stake. This drives the basic Dutch Frame size for each race.
- Second to develop a series of filters which respond to the Shape of the market. These are not filters in the sense of eliminating race types, past form, going, number of runner, etc. They are probability limits for each frame size and race type combination.
Dutch Ratio:
The total stake figure I publish is 485 returning a win of 100 (when using the most efficient fav_hedg bet in larger frames)
The origin for this was checking through a series of large samples of races by trial and error to establish the largest Dutch frames that could sensibly be achieved. In a large number of the races there is a tipping point at which the addition of one extra runner into the dutch bet produces a huge hike in the stake size.
A spin off from this process is a zone for the Book Value represented by the dutch Bet, the largest Dutch Frame achievable usually falling with in a range.
This makes sense after the trial and error checking and now with the huge benefit of hind sight using Book Value to generate the dutch frame size would have been the more methodical approach.
The end product?
by establishing:-
A simple piece of arithmetic covering 4 runners at the average price less betfair deduction came out at 485/486.
Its hardly a formula just a starting point but I found no reason to vary it and it seems a 'fit' with all the varying markets which are thrown up.
Shape of the Market Filters:
A spin off from this process is a zone for the Book Value represented by the dutch Bet, the largest Dutch Frame achievable usually falling with in a range.
This makes sense after the trial and error checking and now with the huge benefit of hind sight using Book Value to generate the dutch frame size would have been the more methodical approach.
The end product?
by establishing:-
- The average Dutch frame size achievable
- The average Betfair price returned off the samples.
A simple piece of arithmetic covering 4 runners at the average price less betfair deduction came out at 485/486.
Its hardly a formula just a starting point but I found no reason to vary it and it seems a 'fit' with all the varying markets which are thrown up.
Shape of the Market Filters:
On the screen shots I put up usually they include 10 columns on the RH side which together with the Uplift value for the race type feed into a series of filters.
How to put a Shape of the Market concept together?
Most people would accept that a 3 runner Dutch bet with jt. favs. @ say 3.5 and then a bunch of other runners 13.0 -16.0 zone is entirely different to one where the fav. is 1.4 and the two others within the Dutch market are 9.0 & 10.0 followed by a sizable gap to the next in the market?
Any market is dominated by the favorite price.
So the full market I split into the following segments.
There are no absolutes in bringing the factors together. A market with a large field of runners totally dominated by a short price favorite obviously has plenty of opportunities to go wrong. Conversely a small Maiden field of say 6 runners with say a high priced >3.0 favorite again is highly probable to throw up a winner from the back of the market.
Surprisingly there is little direct cause and effect due to the size of the Price Gap between the Dutch Frame and the Next runner in the market. The Frame needs to be stable in the few minutes before the off, the Price Gap is to indicate this, not increasing advantage.
The idea then is to set limits for each race type and dutch frame combo outside of which the probabilities are falling away. These Limits interact and create the Shape of the Market. If the market does not crystallize within the Limits criteria i.e it is to far away from a mean or standard market profile - then the probabilities on which the Dutching method is based will not hold?
How to put a Shape of the Market concept together?
Most people would accept that a 3 runner Dutch bet with jt. favs. @ say 3.5 and then a bunch of other runners 13.0 -16.0 zone is entirely different to one where the fav. is 1.4 and the two others within the Dutch market are 9.0 & 10.0 followed by a sizable gap to the next in the market?
Any market is dominated by the favorite price.
So the full market I split into the following segments.
- Price of the favorite and its ratio to the number of runners included in the 'field envelope'.
- Number of runners below a 'Pivot' price of 12.5 A fixed balance point in the market taken for all field sizes.
- Number of live contenders within a 'field envelope' A varying top price related to the number of runners in the race. Starting at 40.0 for large fields & graded down to below 30.0 for 5 runners and less.
- Finally excluded runners outside the field envelope - supposedly (but unfortunately not always) outside contention to win the race.
There are no absolutes in bringing the factors together. A market with a large field of runners totally dominated by a short price favorite obviously has plenty of opportunities to go wrong. Conversely a small Maiden field of say 6 runners with say a high priced >3.0 favorite again is highly probable to throw up a winner from the back of the market.
Surprisingly there is little direct cause and effect due to the size of the Price Gap between the Dutch Frame and the Next runner in the market. The Frame needs to be stable in the few minutes before the off, the Price Gap is to indicate this, not increasing advantage.
The idea then is to set limits for each race type and dutch frame combo outside of which the probabilities are falling away. These Limits interact and create the Shape of the Market. If the market does not crystallize within the Limits criteria i.e it is to far away from a mean or standard market profile - then the probabilities on which the Dutching method is based will not hold?
No comments:
Post a Comment