Forum Replies Created
- Posts
Sorry to disappoint, but I doubt I have either the time or the skills to unravel all of what is in that paper! It’s a couple of years old, and I did read it near the time of publication, but I had forgotten about it until now.
I will say that I have come across these authors in a different context and was far from impressed. The paper about horses being slowed down by declines seemed fundamentally flawed: some of the speeds quoted were implausible and suggested problems with the data; it seemed to make little/no adjustment for the fact that declines occurred at stages of a race at which horses might be decelerating anyway; it made no reference or explanation to the opposite phenomenon which is manifested at e.g. Epsom; and so on. In this study they have used average speeds, seemingly without context, and that may be suspect.
On drafting, the principle is incontrovertible, but things would appear to play out rather differently in horseracing. Else, we would be seeing the vast majority of races being won by horses that stalked and pounced. It would overwhelm many other factors, and it doesn’t. Drafting must save energy, but it is probably far more difficult to achieve effectively in horseracing than in the named alternatives. Can you stick a horse’s nose up the leader’s bum and then switch and accelerate on command?
That races are NOT being won overwhelmingly by stalk-and-pounce runners can be seen by the record of Early Position Figure (EPF) categories (1 is a front-runner, 2 is prominent/stalking, 3 is mid-field, 4 is towards rear, 5 is in rear).
Across British and Irish Flat racing in 2014: EPF1 beat 57.3% of their rivals; 2 beat 54.2%; 3 beat 50.8%; 4 beat 45.8%; and 5 beat 37.5%.
The performance of 1s and 5s varied by distance (unsurprisingly), with 1s doing better at shorter and 5s doing better at longer, but the 2 to 4 range (where we would expect to observe the advantages of drafting) was pretty consistent across all distance categories.
Interestingly, it was a similar story for field size, though “large” (16 plus runners) saw near-parity between 1s and 2s (54.1% and 54.3%) compared to “small/medium” (57.5% and 54.0%). That’s parity, where “Drafting Theory” might imagine 2s and 3s would take off: they don’t.
In addition to the likely difficulty in drafting effectively on horseback, other factors include: 1s get to dictate the pace and unless going too fast are usually nearer to efficiency than drafting horses; horses in the pack are more likely to encounter trouble in running; there is a degree of selection bias in field position, in that out-of-sorts/unsuited-by-conditions horses very seldom occupy a front-running role but occupy an increasing proportion of each EPF category as you go further back in the field.
But, as I said, drafting should be a factor, even if it is seemingly not an overwhelming factor in horseracing (in most cases). I am unaware of anyone who allows for it, and that might be an opportunity. It would be great to do a detailed survey using a bit more horseracing “nous” to see if any rules can be formed.
SDR
Kingman ran 10.22s in the penultimate furlong of last year’s Sussex Stakes and 20.75s for the last two furlongs. Whenever it came down to a test of speed he was mustard.
https://www.timeform.com/racing/articles/sussex_stakes_sectional_debrief
There was some discussion about Frankel and Black Caviar here:
Though that was well before they both ran at Royal Ascot.
Some further context regarding fast (and not-so-fast) sectionals can be found in Timeform’s Guide To Sectional Timing
The fastest last 2f at Lingfield in 2014 was Double Up’s 21.39s, but that was smashed this AW Finals Day by Tryster running 20.97s. Jellicle Ball ran the fastest last 3f by a 2yo at Kempton in 2014 (33.71s).
Sub-11.0s furlongs are rare, sub-10.5s furlongs are very rare, and I don’t think I have come across a sub-10.0s furlong, though it would be just about possible in extreme circumstances.
Aye. The more honest answer would have been “Baby steps? You have to be kidding”
Thought something there (probably not Markov chains and MCsim, admittedly: I didn’t re-read the piece) might spark something.
There must be some theoretical race in which horses are rated 100, 99, 98, 97 etc and you “should” price them 5/1, 6/1, 7/1 etc, but it’s nowhere near that simple in the vast majority of cases.
As has been pointed out, the method put forward falls down once you get much of a skew in distribution.
Joe: judging the difference between 35 mph and 36 mph is essentially impossible to do repeatedly and precisely on horseback when the speed of the surface, qualities of horses and course/distances themselves vary so much (though Steve Cauthen got remarkably close to perfection, IMO), and yet that is more than enough to make the difference between winning and finishing out the back.
So jockeys tend to find the general circumstances of a race dictated to them and some adjust better or worse than others. I actually think jockeyship on AW may be at an all-time high: there’s more of it (so jockeys get more practice) and it’s more important (so better jockeys give it a go and lesser jockeys have more incentive to improve). Adam Kirby and George Baker are excellent jockeys, IMO, but not superhuman enough to judge sectionals with the precision required to render sectionals themselves obsolete.
I just ran the numbers for jockey efficiency for Nov 2014 to present. Martin Harley was still (just) top. Ryan Moore was some way down. Ryan Moore is a better jockey than Martin Harley, but anyone taking sectionals on AW won’t be amazed by that conclusion. Judging pace at tracks he rides at relatively infrequently is not always Moore’s strongest point. He makes up for it, and some, in many other ways and contexts.
Simon
I’ve blogged two or three times about jockey efficiency as measured through sectional times. Jim Crowley came out well in the early days; Think Martin Harley was top in 2013/2014 (sufficiently detailed sectionals exist only for AW).
However, the overwhelming impression was that there is a great deal of noise in the data. Jockeys don’t know how to ride efficiently with any great precision or repeatability. That’s kinda why sectionals are important.
This
https://betting.betfair.com/horse-racing/bloggers/simon-rowlands/post-222-040810.html
works only when ratings are quite closely grouped, such as in handicaps, but may give you some ideas.
I am intrigued by what you think you have discovered, Joe! I did not deliberately “hold anything back”, other than revealing sectional pars for only one of the courses, rather than all/most of them. But it is an “introduction” rather than an advanced work, let alone the final word. Some of the more advanced things – measuring jockey efficiency, modelling in play in conjunction with prices, projecting pace scenarios, identifying winning “systems” from sectionals etc – will need to wait for another day, if they get written about at all. It seemed necessary to set the ball rolling, and the BHA/DCMS were interested in funding it (but changed tack late in the day). You can always drop me a line if you like
It didn’t look good, but in terms of average sec/furlong (16.57 s/f) it wasn’t even the slowest jumps meeting yesterday (Southwell was) and is outside the "top" 25 since October. Lingfield on 16 Feb is well clear with 18.38 s/f.
Aware that such things should be adjusted for ability, distance (real, not imagined) etc, but I think it provides some sort of illustration that the runners were (to borrow a cliche) "going through it"
This has some details of attrition rates
https://www.timeform.com/Racing/Article … ,_part_one
But attrition rates are not the same as difficulty of fences, due to feedback and likelihood of better jumpers gravitating towards courses with fences that are difficult to jump.
Thanks.
You won’t be buying the booklet: for it will be free to download.
Now, that’s selfless/stupid (or is it…..?!)
As someone who has just written a book – Sectional Timing: A Guide by Timeform (out in March) – and appeared briefly on RacingUK’s Betting Lab (for no remuneration), I would like to offer an alternative view.
Some of us are motivated by things other than making as much money as we possibly can from betting, jealously guarding any "secrets" we may feel we have while basking in smug satisfaction about our imagined superiority over our peers.
I did the whole punting for a living thing for a couple of years: it was fun and liberating, at least to begin with, but it contributes next to nothing to anything other than one’s self.
I got into racing analysis because of the fun of puzzle-solving. The most obvious "puzzles" in racing are the races themselves, but there are also the fundamentals of the sport, of athletic performance and of market dynamics. The sort of stuff Betting Lab (and maybe a book) is attempting to tackle, in other words.
There is a hideous leakage of knowledge in our sport due to the absence of written material, so that you get situations like the one whereby the Head of Handicapping at the BHA does not begin to understand aspects of the discipline of which he is a figurehead. That is not entirely his fault, for little of consequence exists in written terms about the subject.
Fortunately, not all areas of human enquiry operate in this manner. "Science" and "philosophy" and so on share ideas, and feedback on those ideas, so that the entire subject matter can advance. Clever chap that Einstein, but did he become a billionaire? No. Instead, he wrote about his ideas, and spoke about his ideas, and those ideas improved as a result. Others cashed in more as a result of those initial ideas, of course.
Jealous guarding of knowledge in an environment where knowledge is power is understandable, but ultimately it corrodes that environment itself. Racing needs more people to engage with it as a set of intriguing puzzles, rather than as a crap shoot, and that is not happening in part because it appears to the outsider at times to be an intellectual wasteland.
Whether it is prompted by vanity or altruism, I, for one, would like to see that change.
Simon Rowlands
Are you familiar with "R", Cav?
I dabbled with it briefly, didn’t really take to it, but people I know who also use Excel swear by it. It is undoubtedly MUCH quicker with big data.
Good stuff. Not sure the comparisons with other sports are entirely valid. Whatever power the EPL or the RFU have over their fixtures, it is basically moving an agreed set of fixtures around. What the BHA/REL need to be able to do is the equivalent of telling Chelsea they can’t play Everton, or that Northampton Saints should take three months off in the middle of the rugby season.
But I agree that the status quo in this area should be challenged. It may have been many times behind the scenes for all I know.
Do you anticipate Rust making a positive difference?
