- This topic has 65 replies, 17 voices, and was last updated 21 years, 3 months ago by
.
- Posts
I agree with ACR1, inasmuch as the word ‘experience’ can be sustituted for ‘trends’. I have never made much use of the trends as presented in various publications, such as age of the winner, weight carried etc. However, I’m very keen on facts such as how many winners go on to win again next time, and how well horses respond to a particular jockey, or how true is the old ‘horses for courses’ adage. I try to assign some numerical value to what I consider to be positive attributes and then add them to the form and speed values to obtain a composite value that represents a horse’s chance i.e. a rating. The relative worth of these positive attributes as tools for winner finding are unique to each individual depending on their own experiences, which takes us back to the start of the thread.
My own preferences, leaving aside the all important going factor are:
1. Recent Speed Ratings<br>2. Form Ratings <br>3. Jockeyship<br>4. Winning form over the course/distance<br>5. Recent winning form
The Draw, according to its positive or negative effect can be fitted in just below jockeyship.
I don’t think there is much of an edge in this type of approach as it stands, but it can lead you down by-ways where interesting treasures may lie hidden. I rather like that shaded avenue of sectional times where Pru is beavering away.
Quote: from aston on 1:17 pm on Jan. 12, 2005[br]Trends are the very last thing you take into considderation imo. If you went by trends, you would have diregarded Essex chances on Sunday immediately.<br>
Aston – which trend or trends would have discarded Essex on Sunday?
Quote: from Prufrock on 10:06 am on Jan. 12, 2005[br]
<br>Ratings—sophisticated ones, or so I hope—provide the foundations of my analysis, but there is much more to consider besides.<br>
are these rating’s based on time and weight carried ?
you have got me frothing at the mouth:biggrin:
An interesting thread and one which makes it a pleasure to spend 10 mins or so in front of the pc.<br>Similar to a thread some months ago contending that, as far as betting was concerned, history was largely bunkum; this equates to "trends" in this thread.<br> To me, very few historical facts are pertinent: the draw yes, but things like SP of winners and age of winners, largely no.<br> History though is complete and utter b*******cks when it comes to soccer and I’m always amazed when otherwise intelligent pundits state that team B are up against it because they’ve never won away at team A’s ground in so many years.<br> Some years ago, when I bet more than I do now on the horses, I spent an absolute age analysing statistics gleaned from the RP 10-year flat results CD on the following themes, trying to get an extra "edge".<br> a) Top jockeys riding at their minimum weight for stables to which they weren’t retained.<br> b) Record of top jockeys when travelling to night meetings where their retainer or main sources of rides didn’t have runners.<br> c) Level of improvement shown by a stables horses when wearing blinkers for the first time.<br> It must have cost me a small fortune on the cost of the printer ink & paper used!<br> As far as the blinkers were concerned, two stables seemed to bring about a larger average improvement than most, but even then you’d have made a level stake loss in supporting them.<br> Only one jockey got near to a level stake profit for the night rides analysis (oddly enough, this was the same jockey who came out top for the minimum weight analysis). But, even then, there was a level stake loss.<br> In the end I concluded that there was nothing to be gained that merited such time-consuming research and anyway, had I analysed a different set of 10 year results, the names would have been different.<br> Anyone else tried a similar slightly off-centre way of tracking things?
Let’s look at trends using a real-life example.
I’ve done an analysis of the last ten years Lanzarote Hurdle results (this year’s renewal being run on Saturday at Kempton) and have established a set of criteria that would have eliminated all but 10 of the runners. Of these 10, 4 won showing a profit against turnover of +280%. The average theoretical chance of these 10 runners, as evaluated by the market before the race, was 17/2 (or 10.52%). Their actual winning chance, as can be established in hindsight by the result, was 6/4 (40%).<br>In other words they were almost 4 times more likely to win than the market had predicted beforehand.<br>Now, come Saturday, there is no guarantee whatsoever that if any horse or horses meet my criteria, they will have a chance 4 times greater than their starting price. However the evidence from past results would tend to suggest that horses who came into the race with similar credentials in the past have performed markedly better than it was generally thought beforehand that they would.<br>Now, a sample of 10 horses is very small and there is every likelihood that the performance of the horses who meet the criteria in the future will not do as well as those from the period our sample has been drawn from. However, the difference between the actual performance and the market assessment of the likely performance of this type of horse is so great that even if those meeting the criteria in the future do half as well as those from the sample (i.e if there are only 2 winners from the next 10) then we would still make a profit, betting to level stakes of £100 per selection, of £900 or 90% profit against turnover. Given that successful pro gamblers often target between 10 and 15% profit against turnover you can see that even with the future selections who meet the criteria doing only half as well as in the past we still have a large margin on our side to accommodate the likelihood that future trends in the race will differ from those prevalent in the past.<br>The proof of the pudding in all this is, of course, in the eating. This type of analysis has proven both successful and profitable for me in the past. Using this approach I put up selections on this site during last year’s Cheltenham Festival (which is ideal for this type of approach, as has been stated by someone else earlier) and they were very profitable. I also use this approach to make selections for the Pricewise Challenge competition where, after only a few races, I am in profit and, from five selections, have had an 11/2 winner, two seconds at 8/1 and 9/1 and two out of the frame. I also use it with some success to help select my horses in the Bob Rolf 4PP comp and in the recent match play comp on the forum.<br>I know from experience that trend analysis can work in the long term. The knowledge that is needed to apply it comes not only from a grasp of statistical analysis (and I’d suggest to anyone backing horses that the more study they do on the laws of probability the better) but also from a knowledge of what kind of trends throw up the most repeatable results and how much the past trend needs to be in your favour before you can approach future races with a degree of confidence that even if your future trend performance falls short of the historical data you are still likely to end up in front. As a rule of thumb, and it is only that, I would not normally consider heavily staking on any trend horse unless historical results indicated the trend had thrown up horses who had at least twice as much chance of winning as the market had predicted beforehand. This rules of thumb does, in my experience, give you enough leeway for the statistical variation to go against you while still leaving a handy profit.<br>I’ll post up the Lanzarote selections on this thread for fun on Saturday but would point out that even if they lose (which the laws of probability state is the likeliest scenario) the relevance of what I’ve said above is not affected!<br>
A lot of wisdom in your last (and earlier) posts. You seem to have identified an approach that is logical, relatively easy to apply with practise, and most importantly, produces excellent results.
I particularly like the fact that you are alive to the limitations of any approach as far as profitability is concerned, hence only limited disappointment when things don’t always pan out as expected . Also, you stress the importance of taking a long term view and making use of statistical analysis wherever possible. All top-drawer stuff in my book, cormack. I think I’ll have to take another look at those trends…….
Quote: from cormack15 on 9:35 pm on Jan. 12, 2005[br]The knowledge that is needed to apply it comes not only from a grasp of statistical analysis (and I’d suggest to anyone backing horses that the more study they do on the laws of probability the better) <br>
Surely anyone with a grasp of statistics trying to draw meaningful inferences from data where the number of cases is less than the number of independent variables is oxymoronic.
Sorry, but you’re just firing the arrow and painting the target around it.
Not if you concentrate only on specific variables which my analysis has proven have a disproportionate effect on the outcome. <br>As discussed earlier the limitations in the sample size is recognised but can be overcome by focusing on trends which are so strong that they allow for a large margin of error.<br>I accept that, from a pure statistics viewpoint, trend analysis has it’s flaws but my point is simply that it does, in practise, provide a useful tool to help predict the likely outcome of certain races.<br>The target may be being painted but so long as the arrow does fall into it with the required frequency then that’s all I need to happen.
good postings Cormack;)
I have been putting the a/w selections using solely trainer stats on the workshop section. <br>2 sels wed (1 sec beaten 1 lenght)<br>2 sels today (another 2nd beat a neck)<br>but since I have started doing it l.s.p. of 19.25 points at s.p win only. All the trainers I follow on the a/w tracks are listed except to add j. boyle in 4 yo handicaps ling a.w<br>rgds
Just a little experiment  using trainer course form
wolves 5f 216 yds 3yo 0-70 hcp
               W-R   %  3yo hcp’s SF<br>R A HARRIS       0-2     0    0-0 100<br>K R BURKE        4-31  12.9   0-2 98<br>M R CHANNON     7-29  24.1   1-4  non runner<br>M C CHAPMAN     0-12    0   0-0 96 <br>P S McENTEE      1-36   2.8   0-2  non runner<br>B A McMAHON      1-8   12.5   0-1 93<br>T.J HETHERINGTON  1-15   6.7   0-1 83<br>N.TINKLER         0-7     0   0-1 92
According to the small sample of stat’s M Channon has the best course form with K Burke following,it will be interesting to see if the stat’s hold up in the race today ÂÂÂ
(Edited by empty wallet at 12:25 pm on Jan. 14, 2005)<br>
(Edited by empty wallet at 12:41 pm on Jan. 14, 2005)
I compile my own ratings so that’s where I voted, but ratings are a numerical expression of the value of form arrived at via an analysis of collateral form. At least mine are.
I have serious difficulties with speed figures. I compile my own time ratings (speed figures by another name, I suppose) but I am extremely selective with them. I am very wary of any speed figures which are arrived at via consideration of averaging of times on any given day.
I am very wary of any speed figures which are arrived at via consideration of averaging of times on any given day.
You are quite right to be, Maurice. The distribution of actual times in comparison to expected times is highly unlikely to be even—on account of the fact that it is more difficult to run fast than it is to run slow and that running fast in terms of absolute times isn’t usually the purpose of the exercise in any case—and an average is therefore seldom an appropriate way of expressing the data.<br>
And i would be wary of collateral form lines
A beat’s B,C beat’s B and because A beat B by x giving 1lb more than C
A will then beat C :o <br> ÂÂÂ
(Edited by empty wallet at 7:42 pm on Jan. 14, 2005)
from Timeform.com
[font=Courier]There is a common misconception that handicapping involves little more than guessing which horse has “run to formâ€ÂÂ
Collateral form should not be ignored, IMO. It’s just that some people leap on it at the first opportunity in order that they don’t have to think about the full complexity of the riddle.
Should it not be under what condition’s the result of each race took place
strength of race ect, and not underestimating a horse that has come from of the pace to win rather than one that has achieved a faster time
- You must be logged in to reply to this topic.