mathare
9th August 2006, 22:39
Details of the systems described below may be found here... (http://www.win2win.co.uk/forum/showthread.php?p=358674)
Results
For each system I recorded the daily number of dutches, number of winning bets, strike rate and profit/loss. This data was recorded for the 44 days under test and then summed to give a summary result for each system. I have no plans to present this data in any numerical format. With 544 systems it would soon get boring having tables of strike rates and profit/loss figures. Instead I have opted to display the results graphically. The attached spreadsheet contains 9 separate sheets - 1 per category of system as described above. Each sheet shows the graphs of all the systems in that category on the left and the data used to create those systems towards the right (column X onwards). The graphs are total profit plotted against date unless otherwise stated.
I'm pretty sure you don't need me to talk you through the graphs in detail but I will offer my thoughts as necessary.
Generally speaking, each chart shows 4 systems. This is usually the base system and various combinations of the filters described above. The key for each chart should make it pretty clear which line is which and what filters
were applied.
I should also point out that SP was used throughout and that use of Betfair would improve these figures, significantly in some cases. But SP was used as an objective measure of system profitability and because that way bookmakers can be used and stakes of less than £20 per race and £2 per runner are possible.
Top Rated Charts
The top two charts show the effects of dutching different numbers of runners on the profit at the end of the test period. The chart on the left excludes NRs from the number included (Count NRs filter = N), the chart on the right includes them in the count. Both charts follow roughly the same sort of shape for each line; starting high with a single selection before the losses increase as we add more selections, a trend that reverses at around 6 runners and the lines climb again. There is also an interesting mini-peak at 4 selections.
One thing you should instantly notice about the top two charts is that the lines never make their way above zero i.e. all these systems are losing systems over this period. It should also be noted that the filters applied all serve to improve the basic system with the Exc Races offering the best improvement. But it still can't turn any of these systems into long-term winners.
Following the summary charts are the individual charts for 1 to 10 selections, including and excluding NRs. Each chart shows the basic system plus the effects of the Exc Runners and Exc Races filters as well as the two filters combined.
For 1 selection the filters have no effect with all 4 lines being plotted on top of one another. A pretty spiky system this one with both big winning days and big losing days. It only barely ever slips into profit though and ultimately ends in a loss regardless of the filters.
The filters start to show up for 2 selections with once again the Exc Races filter showing the greatest improvement. There's a bit of spiking at the start of May but there is a significant downward trend in these graphs. Losing systems all round here.
The story is pretty much the same as we continue to add more selections into the bet but note for 4 and 5 selections that the Exc Races filter becomes the worst of the three applied filters before jumping back to the top of the pile for 6 onwards. The spikiness that we saw before is ironed out as more runners are added to the bet but nothing can change the stark downward trend exhibited by each plot. These are all losing systems and the longer they are run the more they lose. All the filters can do is reduce the losses but they can't turn any of these systems into profitable ventures.
Bottom Rated Charts
The format of these charts is identical to that of the Top Rated charts - we start with two summary charts before displaying the individual charts for 1 to 10 selections in the dutch.
Both charts showing the summary of systems including and excluding NRs follow the same trend again. After an initial dip around 2 selections it's onwards and upwards before peaking at 9 selections. However, even the best filtering of this data results in a loss of just under £2000 over the 44 days (actually -£1698.20 for 9 runners, Exc Races, Exc Runners and Count NRs = N). Note again that the three filtered lines show an improvement on the basic system in all but a couple of cases.
When we look at the individual charts for 1 to 10 selections we see once more that the filters don't really show up for a single selection. The left hand and right hand charts show the same trends though so we can conclude that whether or not we include the NRs in our counts is pretty irrelevant on the face of it. All charts show a strong downward trend again, with a bit of a twiddle at the end of the period covered. This is hardly the start of a significant upturn in fortunes though.
One thing worth noting here is the increasingly pronounced effect of the filters as increase the number of selections, and also that the best filter differs from chart to chart. Once more though all filters serve to decrease the losses suffered proving that they were indeed worth testing but that all of these systems can be cast to the scrap heap as not one of them shows any promise of profit.
Mid-Up Rated Charts
I'm sure by now you've worked out what you're getting chart-wise.
The summary charts show again that 2 selections is the worst case here but that is to be expected as these systems are very similar to the Bottom Rated systems only the starting point is the middle of the ratings list rather than the foot of it. Nothing is ultimately profitable around here though.
The single selection charts show early profit and had you been following this in early April you would no doubt have been delighted. After all, around Easter you have made a profit of 1000% of your starting bank. Not bad for a fortnight's work that. Then watch as all that profit, and some, goes down the pan and you're left staring a hefty loss in the face a few weeks later. And there's nothing the filters can do to help you.
A lot of newbies don't appreciate it but that graph there is basically a gambler's worst nightmare. These figures are computed to a £20 stake per race throughout but when your £200 bank has hit £2000 in about 10 days would you still be staking at £20 a race? Didn't think so. Which means that downward trend would be even steeper. Ouch!
Nothing much of interest to see in the charts for 2 to 10 runners. There is the familiar downward trend, sometimes with a bit of spikiness to keep you interested but ultimately what we have here is 80 losing systems. Yet again the filters all offer some improvement on the basic system though.
Mid-Down Rated Charts
The Mid-Up charts followed similar trends to the Bottom Rated charts so we could reasonably expect the Top Rated trends to be repeated here. And we wouldn't be too disappointed here as those trends are apparent.
The summary charts show a much flatter plot than the U-shaped curves we saw for the Top Rated charts but that is to be expected here as the highest rated selections for these systems are still only in the middle of the pack so are theoretically worse than the horses considered in the Top Rated systems.
Each of the individual curves for 1 to 10 selections continues the downward theme and once more we have a bunch of losing systems that can be consigned to the bin and forgotten about. Unless you have several grand you want to get rid off in the space of only a few weeks.
Min Ratings Charts
A similar approach has been employed here in as much as we have two set of plots showing a summary of final profit from the test period against minimum rating at the top and then below these we have the plots of running profit
plotted against date for all 72 systems. This time the left hand set of charts shows the data with singles allowed whereas the right hand set of plots is dutched bets only.
The summary charts show that the lowest of the minimum ratings under test are big losers and that the profit increases as we increase our minimum rating, peaking at 230. Notice also that at that point the figures with singles
allowed are best whereas generally speaking the summary lines without singles are better. It is also worth pointing out the effect of the Exc Races filter, a factor in both the pink and light blue lines. This filter alone makes a big difference to the profitability of the systems.
Moving down through the charts of running profits for each minimum rating now we see from the first few charts that the removal of singles from the betting reduces the losses incurred but that these are still losing systems. Some systems show a fair degree of fluctuation but not enough to spark any sort of profit rush. That is until we get down to the charts for 230.
For horses rated at least 230 the graph barely ever drops into the red and an overall profit is returned at the end of the test period, making this system stand out like a sore thumb. Yes, there are peaks and there are troughs and I may have stumbled across a purple patch for this system but a profit is a profit isn't it? This system is certainly worthy of further testing. Interstingly, the Exc Races filter doesn't have much of an effect at all here but the Exc
Runners filter does. So it seems runners requiring a stake of less than the Betfair minimum of £2 can be omitted and profits improved. Well, well. Oh yeah, and it helps to keep the singles in there. Dropping those from the system results in a long term loss.
The charts for horses rated 220 and above and 240 and above also show some profit potential and should not be ignored. Further testing may be required to find out where the optimal profit rating is but we have narrowed it down to somewhere in the range 220 to 240.
Max Ratings Charts
The same sort of charts here and instantly were back to the downward trends. The summary charts show that the higher our maximum rating the greater the loss, except for a weird upward turn at around 150. Still doesn't put us back anywhere near a profit though. OK, it might if you extrapolated the lines but think about it - the higher the maximum rating the more horses you can include in the dutch and the more you include the greater your % book included until you eventually dutch every runner and guarantee yourself a loss. Nope, doesn't look to be anything promising here. 104 systems and all of 'em losers.
The individual charts show nothing we haven't seen before in previous charts really.
Enough Runners Charts
Ah, my bizarrely monikered Just Enough systems. Only 16 of them so only four separate charts each with four systems plotted on them. And each and every line shows the same downward trend we have seen hundreds of times previously.
It's vaguely worth noting that you're better off starting at the top of the list with this system, insisting on at least £2 potential profit per race and omitting any runner below the Betfair minimum stake. But only vaguely as none of these are going to make you rich.
Ave Rating Charts
Another set of four charts with four plots per chart and what we see here is just spiky versions of the all-too-familiar downward trend as we splurge more and more money away as time passes. The best version of this system selects only runners above (and not equal to) the rolling average winner rating, excludes singles and incorporates the Exc Races and Exc Runners filters.
Percentage Charts
There doesn't seem to be a lot of difference between the charts on the left and those on the right so we can conclude that it doesn't make a lot of difference whether you using the average winner's position expressed as a percentage of the actual starters or declared runners. I'll bet you're glad to know that. The filters have no major impact here and we can expect to lose around 100 times our race stake in 6 weeks following any of these 16 systems.
Conclusions
There we have it - 544 systems and how many of them show any signs of promise? Yep, one. One measly system out of all those tested. And even that's only around £80 profit. Big wow! But at least I have found over 500 new ways of losing several grand.
Most of that is true but hang on. It's four systems really isn't it? They may all have the same basis but they are four different sets of selection criteria so four different systems. And assuming a starting bank of £200 our £80 profit represents a 40% profit in around 6 weeks. That's not so bad is it? The £80 only looks so small because it is orders of magnitude smaller than the absolute value of the losses. And if you apply the Exc Races and Exc Runners filters you get a profit of just over £150. That represents a 75% profit in those 6 weeks and I don't know of many high street banks that offer that sort of return.
But just because we have found a seemingly profitable way of dutching BigC's ratings should we get all excited? Perhaps not all excited but maybe quite excited. This study should be extended before getting carried away and plunging cash into this dutching system. More races are required with more ratings and more SPs to check that a minimum rating of 230 is still profitable. It looked from my first study (of 1 week's data from the start of May) that the peak level was somewhere around 190 or 200 so it just goes to show that extending the study can lead to different conclusions being drawn. And extending this study is a job for another day (albeit one in the not too distant future) as generating the data in the required format for this testing is a pig of a job!
And finally a bright note to finish on - this study was done using SP. That basically represents the worst case scenario. Dutch these runners with Betfair and you'll get better prices and greater profits. Woohoo! Money! :)
Disclaimer
I usually include a disclaimer in posts like this and no exception here.
The results discussed in this thread are produced by extensive use of macros and spreadsheets. At each stage of testing I spot-checked the figures and the macros were thoroughly debugged during development but bugs can and do occur so the above shouldn't be taken as gospel. I am confident the figures are correct but there remains a chance they are wrong in places.
The conclusions drawn from this study are mine and mine alone. Please do not go lumping on any system I have identified as being profitable without thinking through the consequences and making sure you know what you are
doing. If in doubt, ask. And if it goes wrong and you bust any bank you may set aside for this dutching then I am no more to blame than the man in the street, OK? You're gambling at your own risk, I just provide the analysis but it's you who places the bets.
Each and every system was tested over the same set of data encompassing UK racing from 1st April to 21st May. This is by no means an incredibly extensive study (although it does take in 14320 horses in 1206 races) and further testing (such as paper-trading) should be done before committing your hard-earned.
Credits
Thanks must go to BigC for the ratings in the first place and for letting me pester him in the early stages of this testing. Thanks mate, and keep up the good work :)
And thank you for reading all the way through this :wink
Results
For each system I recorded the daily number of dutches, number of winning bets, strike rate and profit/loss. This data was recorded for the 44 days under test and then summed to give a summary result for each system. I have no plans to present this data in any numerical format. With 544 systems it would soon get boring having tables of strike rates and profit/loss figures. Instead I have opted to display the results graphically. The attached spreadsheet contains 9 separate sheets - 1 per category of system as described above. Each sheet shows the graphs of all the systems in that category on the left and the data used to create those systems towards the right (column X onwards). The graphs are total profit plotted against date unless otherwise stated.
I'm pretty sure you don't need me to talk you through the graphs in detail but I will offer my thoughts as necessary.
Generally speaking, each chart shows 4 systems. This is usually the base system and various combinations of the filters described above. The key for each chart should make it pretty clear which line is which and what filters
were applied.
I should also point out that SP was used throughout and that use of Betfair would improve these figures, significantly in some cases. But SP was used as an objective measure of system profitability and because that way bookmakers can be used and stakes of less than £20 per race and £2 per runner are possible.
Top Rated Charts
The top two charts show the effects of dutching different numbers of runners on the profit at the end of the test period. The chart on the left excludes NRs from the number included (Count NRs filter = N), the chart on the right includes them in the count. Both charts follow roughly the same sort of shape for each line; starting high with a single selection before the losses increase as we add more selections, a trend that reverses at around 6 runners and the lines climb again. There is also an interesting mini-peak at 4 selections.
One thing you should instantly notice about the top two charts is that the lines never make their way above zero i.e. all these systems are losing systems over this period. It should also be noted that the filters applied all serve to improve the basic system with the Exc Races offering the best improvement. But it still can't turn any of these systems into long-term winners.
Following the summary charts are the individual charts for 1 to 10 selections, including and excluding NRs. Each chart shows the basic system plus the effects of the Exc Runners and Exc Races filters as well as the two filters combined.
For 1 selection the filters have no effect with all 4 lines being plotted on top of one another. A pretty spiky system this one with both big winning days and big losing days. It only barely ever slips into profit though and ultimately ends in a loss regardless of the filters.
The filters start to show up for 2 selections with once again the Exc Races filter showing the greatest improvement. There's a bit of spiking at the start of May but there is a significant downward trend in these graphs. Losing systems all round here.
The story is pretty much the same as we continue to add more selections into the bet but note for 4 and 5 selections that the Exc Races filter becomes the worst of the three applied filters before jumping back to the top of the pile for 6 onwards. The spikiness that we saw before is ironed out as more runners are added to the bet but nothing can change the stark downward trend exhibited by each plot. These are all losing systems and the longer they are run the more they lose. All the filters can do is reduce the losses but they can't turn any of these systems into profitable ventures.
Bottom Rated Charts
The format of these charts is identical to that of the Top Rated charts - we start with two summary charts before displaying the individual charts for 1 to 10 selections in the dutch.
Both charts showing the summary of systems including and excluding NRs follow the same trend again. After an initial dip around 2 selections it's onwards and upwards before peaking at 9 selections. However, even the best filtering of this data results in a loss of just under £2000 over the 44 days (actually -£1698.20 for 9 runners, Exc Races, Exc Runners and Count NRs = N). Note again that the three filtered lines show an improvement on the basic system in all but a couple of cases.
When we look at the individual charts for 1 to 10 selections we see once more that the filters don't really show up for a single selection. The left hand and right hand charts show the same trends though so we can conclude that whether or not we include the NRs in our counts is pretty irrelevant on the face of it. All charts show a strong downward trend again, with a bit of a twiddle at the end of the period covered. This is hardly the start of a significant upturn in fortunes though.
One thing worth noting here is the increasingly pronounced effect of the filters as increase the number of selections, and also that the best filter differs from chart to chart. Once more though all filters serve to decrease the losses suffered proving that they were indeed worth testing but that all of these systems can be cast to the scrap heap as not one of them shows any promise of profit.
Mid-Up Rated Charts
I'm sure by now you've worked out what you're getting chart-wise.
The summary charts show again that 2 selections is the worst case here but that is to be expected as these systems are very similar to the Bottom Rated systems only the starting point is the middle of the ratings list rather than the foot of it. Nothing is ultimately profitable around here though.
The single selection charts show early profit and had you been following this in early April you would no doubt have been delighted. After all, around Easter you have made a profit of 1000% of your starting bank. Not bad for a fortnight's work that. Then watch as all that profit, and some, goes down the pan and you're left staring a hefty loss in the face a few weeks later. And there's nothing the filters can do to help you.
A lot of newbies don't appreciate it but that graph there is basically a gambler's worst nightmare. These figures are computed to a £20 stake per race throughout but when your £200 bank has hit £2000 in about 10 days would you still be staking at £20 a race? Didn't think so. Which means that downward trend would be even steeper. Ouch!
Nothing much of interest to see in the charts for 2 to 10 runners. There is the familiar downward trend, sometimes with a bit of spikiness to keep you interested but ultimately what we have here is 80 losing systems. Yet again the filters all offer some improvement on the basic system though.
Mid-Down Rated Charts
The Mid-Up charts followed similar trends to the Bottom Rated charts so we could reasonably expect the Top Rated trends to be repeated here. And we wouldn't be too disappointed here as those trends are apparent.
The summary charts show a much flatter plot than the U-shaped curves we saw for the Top Rated charts but that is to be expected here as the highest rated selections for these systems are still only in the middle of the pack so are theoretically worse than the horses considered in the Top Rated systems.
Each of the individual curves for 1 to 10 selections continues the downward theme and once more we have a bunch of losing systems that can be consigned to the bin and forgotten about. Unless you have several grand you want to get rid off in the space of only a few weeks.
Min Ratings Charts
A similar approach has been employed here in as much as we have two set of plots showing a summary of final profit from the test period against minimum rating at the top and then below these we have the plots of running profit
plotted against date for all 72 systems. This time the left hand set of charts shows the data with singles allowed whereas the right hand set of plots is dutched bets only.
The summary charts show that the lowest of the minimum ratings under test are big losers and that the profit increases as we increase our minimum rating, peaking at 230. Notice also that at that point the figures with singles
allowed are best whereas generally speaking the summary lines without singles are better. It is also worth pointing out the effect of the Exc Races filter, a factor in both the pink and light blue lines. This filter alone makes a big difference to the profitability of the systems.
Moving down through the charts of running profits for each minimum rating now we see from the first few charts that the removal of singles from the betting reduces the losses incurred but that these are still losing systems. Some systems show a fair degree of fluctuation but not enough to spark any sort of profit rush. That is until we get down to the charts for 230.
For horses rated at least 230 the graph barely ever drops into the red and an overall profit is returned at the end of the test period, making this system stand out like a sore thumb. Yes, there are peaks and there are troughs and I may have stumbled across a purple patch for this system but a profit is a profit isn't it? This system is certainly worthy of further testing. Interstingly, the Exc Races filter doesn't have much of an effect at all here but the Exc
Runners filter does. So it seems runners requiring a stake of less than the Betfair minimum of £2 can be omitted and profits improved. Well, well. Oh yeah, and it helps to keep the singles in there. Dropping those from the system results in a long term loss.
The charts for horses rated 220 and above and 240 and above also show some profit potential and should not be ignored. Further testing may be required to find out where the optimal profit rating is but we have narrowed it down to somewhere in the range 220 to 240.
Max Ratings Charts
The same sort of charts here and instantly were back to the downward trends. The summary charts show that the higher our maximum rating the greater the loss, except for a weird upward turn at around 150. Still doesn't put us back anywhere near a profit though. OK, it might if you extrapolated the lines but think about it - the higher the maximum rating the more horses you can include in the dutch and the more you include the greater your % book included until you eventually dutch every runner and guarantee yourself a loss. Nope, doesn't look to be anything promising here. 104 systems and all of 'em losers.
The individual charts show nothing we haven't seen before in previous charts really.
Enough Runners Charts
Ah, my bizarrely monikered Just Enough systems. Only 16 of them so only four separate charts each with four systems plotted on them. And each and every line shows the same downward trend we have seen hundreds of times previously.
It's vaguely worth noting that you're better off starting at the top of the list with this system, insisting on at least £2 potential profit per race and omitting any runner below the Betfair minimum stake. But only vaguely as none of these are going to make you rich.
Ave Rating Charts
Another set of four charts with four plots per chart and what we see here is just spiky versions of the all-too-familiar downward trend as we splurge more and more money away as time passes. The best version of this system selects only runners above (and not equal to) the rolling average winner rating, excludes singles and incorporates the Exc Races and Exc Runners filters.
Percentage Charts
There doesn't seem to be a lot of difference between the charts on the left and those on the right so we can conclude that it doesn't make a lot of difference whether you using the average winner's position expressed as a percentage of the actual starters or declared runners. I'll bet you're glad to know that. The filters have no major impact here and we can expect to lose around 100 times our race stake in 6 weeks following any of these 16 systems.
Conclusions
There we have it - 544 systems and how many of them show any signs of promise? Yep, one. One measly system out of all those tested. And even that's only around £80 profit. Big wow! But at least I have found over 500 new ways of losing several grand.
Most of that is true but hang on. It's four systems really isn't it? They may all have the same basis but they are four different sets of selection criteria so four different systems. And assuming a starting bank of £200 our £80 profit represents a 40% profit in around 6 weeks. That's not so bad is it? The £80 only looks so small because it is orders of magnitude smaller than the absolute value of the losses. And if you apply the Exc Races and Exc Runners filters you get a profit of just over £150. That represents a 75% profit in those 6 weeks and I don't know of many high street banks that offer that sort of return.
But just because we have found a seemingly profitable way of dutching BigC's ratings should we get all excited? Perhaps not all excited but maybe quite excited. This study should be extended before getting carried away and plunging cash into this dutching system. More races are required with more ratings and more SPs to check that a minimum rating of 230 is still profitable. It looked from my first study (of 1 week's data from the start of May) that the peak level was somewhere around 190 or 200 so it just goes to show that extending the study can lead to different conclusions being drawn. And extending this study is a job for another day (albeit one in the not too distant future) as generating the data in the required format for this testing is a pig of a job!
And finally a bright note to finish on - this study was done using SP. That basically represents the worst case scenario. Dutch these runners with Betfair and you'll get better prices and greater profits. Woohoo! Money! :)
Disclaimer
I usually include a disclaimer in posts like this and no exception here.
The results discussed in this thread are produced by extensive use of macros and spreadsheets. At each stage of testing I spot-checked the figures and the macros were thoroughly debugged during development but bugs can and do occur so the above shouldn't be taken as gospel. I am confident the figures are correct but there remains a chance they are wrong in places.
The conclusions drawn from this study are mine and mine alone. Please do not go lumping on any system I have identified as being profitable without thinking through the consequences and making sure you know what you are
doing. If in doubt, ask. And if it goes wrong and you bust any bank you may set aside for this dutching then I am no more to blame than the man in the street, OK? You're gambling at your own risk, I just provide the analysis but it's you who places the bets.
Each and every system was tested over the same set of data encompassing UK racing from 1st April to 21st May. This is by no means an incredibly extensive study (although it does take in 14320 horses in 1206 races) and further testing (such as paper-trading) should be done before committing your hard-earned.
Credits
Thanks must go to BigC for the ratings in the first place and for letting me pester him in the early stages of this testing. Thanks mate, and keep up the good work :)
And thank you for reading all the way through this :wink