The last few months have proved a fascinating battle of man vs. the elements (hot air, turf, passion and rain).Armed only with a laptop computer and a mobile phone, we have set off in the never ending quest to turn coppers into guineas.
I intend to keep maintain a section that will try and keep you informed of the strengths, weaknesses, improvements, successes and failures of our approach.
Our basic approach is one of correlated form, if A beats B and B beats C then surely A would beat C. We study the effect of going, race distance, weight carried and time trends using tried and tested statistical methodology such as regression. (Trust me , I'm a mathematician!). The toughest part is ensuring a stable solution. There are far too many factors to consider for the amount of results available. A useful rule of thumb is that to make sense of a certain number of factors, say x, one needs (x+1)² results; e.g. for x=4 (going, weight, time and distance) one needs 25 results involving that horse. An attempt to get away with less tends to lead to some pretty unstable predictions.
The whole program is now awash with damping and safety valves to ensure that we do not get carried away by over exaggerated extrapolation. This safety first approach seems to give fine results for both well exposed horses and newcomers with just a couple of runs. The penalty we pay is a slight slowness in responding to horses that establish a steady base rating but then start to dramatically improve.
First of all Malc and I independently look at the cards. I tend to note down 8 or 10 races that seem to offer a possible consideration for a bet. I then make a more detailed analysis of these races and tick selections that I would be happy to bet on. Malc does something similar and then we get together on the phone for the grand debate. This whittles down the selections to those on which we both agree. I keep careful notes on all of these selections and this is what we have managed to achieve so far.
Initial selections = 703; return = 802.83; profit = 14.2%
(these selections date back way before we started publishing tips)
My suggestions for a bet = 283; return = 351.77; profit = 24.2% (again dating way back)
Final selections = 44; return = 70.29; profit = 59.8% (This figure needs to be treated
carefully as I only started keeping notes in April, a rather good period for us).
Now the thing that is interesting me most is that our selection process is not actually achieving much in terms of overall profit.
| No of selections | Selections per Day (approx) | 50 bet Pool turnover per year | Profit per pool turnover | Yearly profit (% of pool) |
||
| Initial selections | 703 | 7 | 51 | 14.2% | 724% | |
| Selections after analysis | 283 | 4 | 29 | 24.2% | 702% | |
| Agreed tips | 44 | 1.6 | 11.7 | 59.8% | 700% |
Now a reasonably small selection of tips has the merit of being practical in terms of the amount of time it takes to bet at the rather slow bookmaker's sites. However the initial selections have the merit of taking very little time to analyse. Almost to the point where I could write a little script to make the bets automatically without even having to get up in the mornings. It certainly makes a nice dream anyway.
Dave Taylor, April 25th 2000