The home of intelligent horse racing discussion
The home of intelligent horse racing discussion

CUMIN

Viewing 15 posts - 46 through 60 (of 85 total)
  • Author
    Posts
  • #74686
    Artemis
    Participant
    • Total Posts 1736

    EC

    I think you may have misunderstood what I was trying to say.

    Once you have arrived at a standard time for a particular distance on good ground(and I don’t know exactly how this is done), you should be able to make predictions or deductions about the times you might expect from horses of different class and age with minor adjustments for weight carried, given a fairly run race.

    You can also work in the other direction and deduce  an estimate of the standard time given any race time(fairly run) and the age, class and weight carried by the winner.  

    You also need a table for comparing time, weight and distance, and WFA tables.

    For example, at 5f(1sec =22lbs) in August(WFA 2yo 17lbs, 3yo 2lbs)

    If a standard of 57.80(RPR 100 at 9st, good) has been established, you can estimate the time for a race with a form value(at 9st) of RPR 60, no wfa. It should be 40lbs slower than standard, or 1.81secs slower than 57.80 i.e 59.61 secs.

    You can take every race time over 5f on good ground over the last 5 years and work back to a standardised time, as Wallace has done, and you should get a fair estimate of what the standard should be. The fly in the ointment is that if you take every time, you inevitably include races that are falsely run and these will make your standard higher than it should be.

    Granted, there shouldn’t be too many 5f/6f races that are not run at a fair pace, but at 7f+, only the most competitive races are almost guaranteed to be truly run.

    The relationship between class, age, weight and distances used by the RP, Timeform and others is long established. It was used by Phil Bull, who founded Timeform, to great effect about 40 years ago, in conjunction with Dick Whitford, the pioneer of ratings -based modern handicapping methods, and has been used successfully by many successful professionals over the years. The RP follows the same methods as Bull, Whitford and many others who have followed in their footsteps. Hardly cowboys, EC.

    It would be interesting to see Wallace’s standardised times using only the best times(best 10,say).

    #74687
    Prufrock
    Participant
    • Total Posts 2081

    Good post Artemis.

    After all the times have been adjusted to take into account weight carried compared to wfa, ability of horses, putative ground conditions etc, you certainly don’t want to be taking averages as certain circumstances/distances lead to more truly run races than others. I was taught to use SD sets.

    Timeform has actually been around for about 60 years, BTW.

    #74688
    Prufrock
    Participant
    • Total Posts 2081

    Some of you will know that I am, cough, no great fan of Mordin. But he did write something a year or two back in which he said that a valid way to deal with standard times is to get some reasonably accurate ones in the first place and then just keep fine-tuning them and tweaking them by applying them to real events until you are happy.

    My own experience is that there is something to be said for this approach (in the right hands), not least because it cuts through a mountain of maths.

    #74689
    Andrew Hughes
    Member
    • Total Posts 1904

    I was taught to use SD sets

    Excuse my ignorance but what are SD sets and how do they work?

    #74690
    Prufrock
    Participant
    • Total Posts 2081

    Standard deviation sets.

    It is a way of addressing the fact that the distribution of times is not necessarily consistent. We keep being told that most races on the Polytrack at Lingfield are not truly run. If that is so, why would we use simple averages or medians? Similarly, the proportion of truly run races at, say, 5f is likely to be significantly different to the proportion of truly run races at 16f. An average or median is likely to represent something different in one than the other.

    I don’t work at Timeform, by the way.

    #74691
    Andrew Hughes
    Member
    • Total Posts 1904

    Thanks for that, Prufock.

    I have had a quick look on Wikipedia, and the entries under standard deviation look rather complicated. I may sit down for an hour or two over the weekend and try to fathom them out.

    #74692
    Prufrock
    Participant
    • Total Posts 2081

    2yo races..3yo races…none of those has any place in making standards..if you do use them..your standards are pap.

    I can see the point of your suggestion, EC, but the above comment seems to presume that either younger horse races cannot produce truly run races in the way that older horse races do, or that you cannot or will not adjust younger horse times for weight carried (it would not be necessary to factor in weight-for-age if you did not want to) and ability shown. That I would not agree with.

    #74693
    empty wallet
    Member
    • Total Posts 1631

    Interesting discussion peeps

    <br>Just Dust is overpriced imo if able to repeat Goodwood run

    #74694
    Artemis
    Participant
    • Total Posts 1736

    As I’ve said, I don’t know exactly how RP get their standard. I imagine their method has been published in the Weekender over the past year, but I haven’t read it for ages.

    Here’s how I would do it.

    I would look back over the past five seasons at races with ‘good’ in the going description. For each distance, I would identify the best twenty times and standardise these times using the method explained in my earlier post. I would then look through those races that produced the lowest computed standard times and see if I thought they were reliable races – using my own judgement. I would expect good quality competitive handicaps to account for most of the better times.

    I would then try to fit the times together, given the topography of the course, to see if there were any anomalies or things that didn’t look right. If I was satisfied, I would be happy to go with this time as my standard for the course and distance.

    #74695
    empty wallet
    Member
    • Total Posts 1631

    EC

    i’m just a little worried about the stiff 7f, hence my not saying it’ll win  and i’m on at 5.0 ;)

    I’ve got Gosden horse  as a place lay btw

    (Edited by empty wallet at 6:32 pm on Aug. 11, 2006)

    #74696
    empty wallet
    Member
    • Total Posts 1631

    Quote: from EC on 6:49 pm on Aug. 11, 2006[br] Tobosa stayed,,Just Dust didn’t

    <br>Yep

    (Edited by empty wallet at 6:58 pm on Aug. 11, 2006)

    #74697
    Prufrock
    Participant
    • Total Posts 2081

    EC, do you adjust for the specific apparent ability the horses have shown in these historical races, or are you guided by class pars?

    #74698
    empty wallet
    Member
    • Total Posts 1631

    Sorry EC  didn’t see yer post in time, Chinese  first, racing 2nd :biggrin:

    #74699
    Wallace
    Participant
    • Total Posts 862

    I have been away racing today and missed must of the follow up to my post.  I can produce any set of times from the data I have and will do some more on this in the morning.  For the Goodwood 5,6 & 7 furlong tracks, I ‘ll show the figures based on top 10 or 20 times, do same with handicaps for mature horses only and also do a split by class.

    I’m not sure of the value of using SD’s.  Can this be done using the SD function in Excel with the full range of times for a distance?

    #74700
    Prufrock
    Participant
    • Total Posts 2081

    I don’t know whether you can do SD sets on Excel. I’d appreciate it if you could point me to where I can find out for myself.

    The thing about SD sets is that if you get one set of adjusted times of, say:

    60.00, 60.10, 60.20, 61.00, 62.00 etc (skewed)

    And another of:

    60.00, 60.50, 61.00, 61.50, 62.00 etc (normal)

    I suspect they should be dealt with by means other than simple medians or means and SD sets do this.

    But I learnt this second hand from a renowned mathematician and punter and did not devise it myself.

    As mentioned earlier, I think using ball-park figures and then fine-tuning them until they satisfy you is a valid approach.

Viewing 15 posts - 46 through 60 (of 85 total)
  • You must be logged in to reply to this topic.