Faint Yet Pursuing -
Signals from the Market (Part 2)
|
|
Faint Yet Pursuing - Signals from the Market (Part 2) |
|
|
Article 12_005_1
20th January, 2012 |
This article is the second Part of a pair that give an overview of a brief study done to see how horses performed compared to how they moved in the Market before the race. Those that were noted as shortening from the opening priced are called 'Supported' horses or 'Support' for short in the articles. Those horses that lengthened in price from the opening price are said to be 'Drifting' or 'Drifts' for short. The first article gave full details of what data the study is based on and how it was anlaysed. It also covered in detail the results found from looking at the sample of Supported horses including considering what could be learnt from the outcome but also highlighting any shortcomings in the study approach. It is recommended that the first article should be read before moving onto this Part 2.
The Market prices used are those reported by the media in the lead-up to each race, usually available from around 10-15 minutes before the event. These are the ones that will be reported in most publications post-race. They have some clear limitations in that they are based on betting patterns reported from the on-course betting ring plus a small number of other sources. The on-course Markets can be very small in turnover terms, for example. These reported prices also do not reflect fully the Exchange prices available although there are clearly links between for a variety of reasons.
The reason for doing the study was to do something active to investigate how informative Market moves before a race actually are. Look, in retrospect, at the final results that supported and drifting horses achieved. This is interesting in itself but also needs to be done to gather evidence to counteract the 'Fantasy' view on the Market that is promoted by most strands of the Media. The core of this view is that every race is carved up in advance by a group of insiders who have decided which horse is going to win. Starting with this view then the Media personnel can then get away with their stance of reporting every, even minor, changes in Market price as indications of knowledgeable people fixing things in the appropriate places. This is rather than accepting that prices move in various ways in Markets in response to many inputs and assigning causes is a difficult business.
The detrimental affects of this Media conceit in relation to Betting Markets are widespread and destructive. Generation after generation of horse racing followers taught that racing is 'fixed' to some high degree. The Media will then set up Bookmakers as a Bogey Figure so that they can celebrate any outcomes which fit their warped view because it can be painted as beating up on the bad guys (i.e. the bookmakers). When it is you, the ordinary punter, who is being most affected if this view were really true. Look at any Racing Channel or terrestrial TV coverage and trying to build up this fantasy and using it as the core of the coverage will be apparent.
Read any 'Man on the Spot' type summary of a 2yo maiden in the written media before it is run and you will lose count of the number of uses of the "... worth a check in the Market..." format for individual horses. If you think about what this can mean you come down to the same Fasntasy view that a sharp shortening in price will tell you it is here to compete to win. A drift will mean the insiders know it is not going to win. B2yoR's advice would be to put a line through every such reference if it is on paper and translate it as "... I know nothing about this horse but have to fill the space to summarise it so will use a meaningless phrase that invokes a belief in some external agency who knows more than the 'Nothing' I know....".
This second article looks in more detail at the results produced by looking at how horses noted as Drifts performed in races. If we take the Fantasy view of the Media as the hypothesis to be tested then we can suggest in advance what we should see. Low Strike Rates and large losses presumably. The direct link between the insiders who arrange all the race results and Market moves will dictate this. The next section gives a summary of the results achieved.
The following table summarises the results attained by all of the runs designated 'Drift' in the study. The data comes from the years 2010 & 2011 and the first line in the table covers every run for the the numbers of Wins, Runs and Strike Rate. The final column then shows what the 'Return on Investment' (ROI) would have been for someone having a level stake £1.00 bet on every run. Negatives figures are losses and are colour coded in Red. For example an ROI of -0.03 means a loss of 3 pence on every pound bet, or 3%. Conversely a positive figure shows an overall profit and +0.46 would mean a profit of 46 pence on every £1.00 bet (invested), or 46%. Unlike the Supports data there is no need to report a separate 'Opening Price' ROI because the Final SP is the higher figure.
| Group | Wins | Runs | Strike Rate |
ROI (Final SP) |
| 2010-11 | 320 | 1,791 | 17.9% | +0.46 |
| 2010 | 167 | 903 | 18.5% | +0.52 |
| 2011 | 153 | 888 | 17.2% | +0.41 |
| FTOs | 77 | 458 | 16.8% | +0.75 |
| STOs | 62 | 339 | 18.3% | +0.21 |
| SPL4 | 142 | 512 | 27.7% | +0.03 |
| SPG4 | 178 | 1,279 | 13.9% | +0.64 |
| SCL4 | 173 | 1,082 | 16.0% | +0.77 |
| SCG4 | 147 | 728 | 20.2% | -0.03 |
| SC Top 10 | 1 | 10 | 10.0% | -0.74 |
| SC >= 10% | 12 | 41 | 28.6% | -0.04 |
The lines in the table below the first line then show how the results vary if you split the overall data in two by some factor. Or, if you take a particular sample from within the overall group. The first two lines show the overall data for 2010-11 split down by individual years. This gives an indication of how consistent the data was in the two years. In this case the answer is more variation than with Supports but still pretty closely aligned..
The rows labelled FTOs & STOs show the results for the just the debut runs in the survey (i.e. First Time Out or FTO), or the second runs (STO). These are sub-sets or samples from the complete data and a third figure for all runs later than STO would complete the set.
The rows labelled SPL4 & SPG4 split the overall data in two depending upon the Final SP. All SPs up to, and including 4/1, make up the 'SPL4' group. All returned SPs above 4/1 are in the SPG4 group.
The rows labelled SCL4 & SCG4 split the overall data in two depending upon the percentage change of the SP from opening price to final SP. Smaller changes up to, and including 4% of relative probability, are in the SCL4 group. All price changes of 4% and above are in the SCG4 group. Note, that these 'change' groups can include SPs of any length since they are about relative changes.
The final two lines are really for reference and to get a feel for whether there might be any difference in results if you only assessed the very largest changes in price. The SC Top 10 line shows the returns for the ten largest lengthenings of prices. Note that the 11th & 12th on the list were both wins. Of the 10 runs as many as 7 were Final SPs below 2/1 and none higher than 6/1.
The lines labelled Supp >=10% shows the returns for every 'Drifting' run in the survery which showed a 10% change in price, or higher. Note that only 3 of the 41 noted had Opening Prices above 3/1 and 14 were Evens or shorter.
The 'Conclusions' section below examines the results shown in the table to try to assess what the data might be telling us and how the survey could be improved.
The data was then split down by individual trainers to see whether any patterns showed up. The reasoning behind this approach is covered In Part 1. Rather than publish the results as a table that first article picked out three examples in two areas which showed the large variation in performance that trainers produced. In that case some trainers had very good records and Profits with Supported horses and and others large losses.
A similar approach is taken here with the Trainer records with horses noted as 'Drifts'. The outcomes we are expecting are reversed because we should expect trainers to show low Strike Rates and losses with horses drifting in the Market. That is If the Market has quite such a direct link to the background Enforcers as we are supposed to believe. This section starts with three examples where Drifts appear to display some useful indicators.
William Haggas was used as a positive example with Supports but he is instructive to use here for two reasons. Firstly his record with drifting horses looks to be solid and has produced just 1 win in 2010-11, on the runs studied, and an overall loss of 83%. The second reason is that it highlights a point about the record with Market Movers being two-sided. If you have a trainer whose supported horses mostly win and the drifting horses mostly lose the double effect makes the possible correlation seem more plausible. What would you think of a trainer whereby his supported horses won at above his normal Strike Rate but the drifting horses won at his normal Strike Rate and produced a profit because they had drifted to long-term value Final SPs? How much does the Market know in that example?
The odd win was his only STO representative so that grouping actually produced a profit. But, that winner is useful to consider because it makes a point about Drifts and how often those are price lengthenings of well fancied horses. In this case the winner was still the favourite in a 16 runner Newmarket maiden even after the drift. The longer final SP presumably just indicating that the opening SP was too short given the unknowns in the opposition. Things happen in Markets which are nothing to do with insiders connected with a horse or privy to hidden knowledge. There are not GAMBLES and plots going on in every race.
As with the Supports if you order Mr Haggas' drifting horses by change in price from opening to Final SP then the largest Drifts contain more of the unplaced runs than the lower ones. Despite the lower level of price change overall with Drifts than Supports (Bookmaker response driven?) the Haggas figures suggest a cut-off figure between 4-5% change for some real knowledge input over just the Market prices changing naturally. The one win was a change of 4.4% and somewhere in the borderline region.
Many people are not going to be surprised that Michael Stoute gets chosen as an example. A good trainer who knows how how much ability his horses have and how much of that ability they are ready to show on the track. On top of that there does appear to be a route for information about this area to get fed into the Market. Inspect his record with Drifters and just 2 winners from 26 recorded examples. The only drifter to win above the 4% price change level was in a Group race by a filly on STO after a break. The majority of the Drifts are typically in Maiden events.
Particularly interesting is his record with Drifters in maidens on debut. This group brought no wins from 14 runs with 11 opening at prices below 5/1 and all below 10/1. You would think that a lot of the opening price levels owed a lot to the Market over-reacting to a well bred, or expensive one (the only types he has), trained by him and ridden by Moore or Fallon. But Stoute is not a 'FTO Trainer' and his overall record with FTO runners in 2010-11 was 7 wins from 102 goes and a ROI loss over 60%.
Think about his FTO record in that light and why has he had all these runners opening at short SPs in the Market? So, not an absolutely clear story but a low strike rate overall. But, perhaps the low Strike Rate with FTOs just reflects his overall record? Look at his Supports record and his two of his best 2yos in the period were both noted as Supports and managed to be two of his debut wins. Another of his FTO successes was a Group class winner but opened as short as 7/4f and did not move in the Market so perhaps a case there of the opening level being an indicator. The only other horse he had returned at a shorter SP on debut in 2010-11 was Sea Moon who managed to get beaten but proved a 3yo Group 2 winner. But you would wonder what the Star Of Dance story was in that case who opened at 6/4f on debut, drifted to 2/1, and got well beaten and only ran once afterwards.
If you look at his overall record with Supports then it show consistent small profits in most categories including FTO although again on a below average Strike Rate by his standards. But, further evidence to confirm the view of a trainer where the Market has some knowledge. However, you would add a note of caution that the Market seems to over-react to the elevated regard he is held in to have a range of his horses at lower prices than they should be. Picking out the real Supports needs a bit of finesse in this set-up.
The final example is a slight surprise to B2yoR in Stan Moore. Not in the same class of surprise as Clive Brittain but Moore is another trainer whereby doing the survey forces you to update and out-of-date view of the trainer you carry as a quick check during the season. Mr Moore used to work maily with cheap ones and had a good record of getting wins out of them but it took a long time. Often in late season or on the AW over the 2-3yo winter. The last point usually an indicator of a trainer who is struggling and finally manages wins whent he competition becomes much easier in the weak races on the AW at that time of year. Often signs of 'punts' on his horses but usually unsuccessful so he got filed away under the 'take not much notice' label. He was good at getting STO wins in sellers very early in the season but that was it and he does not seem to do much of that any more.
Then you look at his record with Drifters and just 1 success from 18 recorded. The sole win by a horse who still started as favourite despite the 4.7% Drift and was coming back off a break and probably only won by getting a soft lead in a slow race. Unlike Stoute, who barely runs a 2yo more than twice, a lot of the Drifts are on horses on their fourth run or later so a different set-up. Here Moore is perhaps demonstrating he understands the real quality of the 'form' achieved in early races by his horses when the Market formers are still wrestling with the problem and over-valuing places in weak races.
Looking at his record with Supports and this wwas even more of a surprise. A lot of examples, the majority of the changes above 4%, and a good Strike Rate. Solid profits in all groupings except when the SPs get below 50%. Another trainer with such different records with Drifts and Supports that they go together to increase the belief in a real correlation.
The previous section might have lulled you into a feeling that the Market knows an awful lot and the Pundits have got it right. Time to knock that view with a startling example in Richard Fahey. Big stable, lots of runners, shrewd backroom staff and so on. Surely another Stoute type record? Absolutely not and remarkable profits overall with Drifters and average profits, at worst, in the least good category when you you split the examples down. Add to that the fact that the record comes on a good sized sample of runs and unlikely to be a quirk you might get with a small stable on a handful of representatives.
Look in a bit of detail at the Drifter runs and you could propose a couple of related 'systems' to follow using a Market Drift as the selection tool for bets. On runs after debut look at for any drifters that open at 5/1 or under and ignore the Market. The majority of his drift are over 4% and for horses on later runs you should actively support them because. As a group they are drifting out to value Final SPs. With debut runs he produces his best profits for a grouping and you can ignore the opening SP qualification here because the horses that drift from longer opening SPs are profitable to follow.
This sort of finding is one that turns up with many good trainers with strong records overall including Hannon and the Godolphin yards. Start splitting their records up by some factor, not necessarily price changes, and you find that they produce profitable results. This reflects the strong performance overall of the trainer. For other factors the lack of knowledge this indicates in the Market's assessments of the stable's runners is indirect. In this case it shows that a busy Market with lots going on cannot properly assign the correct level for a set of Fahey runners, even if it has the necessary information to work on.
The next example is Ed MdMahon and partly chosen because of his close links with owner John Fretwell who would be a prime example of the sort of 'Agent' the Pundits believe is pre-ordaining the results of races. McMahon's record shows 17 noted drifters with 12 of those relating to Fretwell horses. His overall record is 5 wins from the 17 and profits it most groupings although the FTO record is only at break even level and a 0-2 record with second outings. 3 of the 5 wins were Fretwell horses and the 12 total runs for that owner also produced 5 places. The overall record with Fretwell horses, on a small sample, is therefore a solid profit.
Look at the results sorted by SP change and the 10 largest drifts include all 5 wins and only one horse that was unplaced. You might suggest looking at the results that at times Fretwell refrains from betting and perhaps because of the low opening prices. Once that decision is made then they do not step in after the drift. But, some of the wins, both with Fretwell horses and other owners, would appear to be either misjudging how much a horse in improving on later runs or, and equally likely, a stable that is not as active in the Market as the reputation woud suggest.
Unlike the case with some trainers the record with Supports does not match the inconclusive record with Drifts. A strong record overall with Supports although the wins are spread throughout the SP change percentage race. The two large changes, of 19% & 13%, both won although only one was Fretwell owned. In summary an interesting example becuase with Supported horses there seems a clear case of a message linking through to the Market but this is not matched by the record with Drifting horses. The Drifts are as competitive as the Supports overall and produce solid profits on a smaller number of examples.
A completely different story to McMahon with the last example and the resurgent fortunes of Henry Cecil. A top class trainer and one who knows how good his horses are at home before they run int he same manner as Stoute. But, no sign of the same indicators that the Market has any good link into how the juveniles are going to perform. To set the story with Drifts up we can start by considering his horses that were noted as shortening in the Market. Only a small number of examples but a below average Stike Rate for him and a loss in all groupings. Just a hint that if you restricted yourself to those examples shoing the largest changes you could improve results. But, the low Strike Rate along with the lower opening SPs mean you would struggle to make a profit however you split things.
The record with Drifters is very different and with more examples overall. The results are a little unbalanced with a break even in 2011 boosted by a terrific return in 2010. Add the two together and well above average proits in all other categories. Look at the 6 horses which drifted most in the Market and it produces three wins and two place. One of the wins by Frankel.
Although perhaps for subtly different reasons you get the same feel as with the Fahey returns. The Market respects the stable and many horses open at too short prices. Because there is little in the way of input from the stable into the prices and, in the Cecil case, apparently little general direct communication from any related sources the prices wander around without much direction. Ignore the drifts, as with Frankel, and tell yourself you are backing horses trained by an excellent trainer who has the horse in a race it can compete well in. That would seem to be a worthwhile summary.
To start this section it is worth repeating the caveats noted in Part 1 about the selection process for the study.
The first point to note with the analysis is that, on reflection, the selection of Supports and Drifts that have been recorded are too 'positive'. At the headline level Drifters, the sample records make up close to 9% of all of the 2yo runs over the two seasons. A large number of the other runs are not qualified as Drifts because they are Supports or the price did not change, or only a very small amount. Even so, the 9% figure is still a sampling, to some degree, within all Drifting runs. But, the total number of winning runs recorded in the Drifts section is 15% of all wins across the two years.
The selection process was more thorough in catching winning runs and those cases which produced the largest changes in price from the opening figure to the final SP. The following conclusions are likely :-
However, the results are still a useful introduction to the area and the results surprising. The outline table in Section shows above average to very good profits in many groupings and above the levels recorded with Supported horses. Even if the selection process is too positive there is a lot of profit to shift away to ever get to losses when considering drifting horses. If you want to believe in a simplistic view of drifting prices correlating strongly with horses that will perform poorly then the evidence is just not there. To make the point fully, if that were true then it would not be possible to make a positive selection, even of a small sample size, from the drifters. Let alone one which shows large profits overall across more than 1,791 individual outings.
The results for different groupings show more variability than with the Supports. The Drifts which ended up at a Final SP of 4/1 or less were around a third of the total and made a very small profit. This would almost certainly turn into a loss with a full selection of Drifts. However, a number of the individual trainers show a profit in this area. Those may well be successful trainers who are well respected by the Market so that competitive horses open at the wrong price on reputation. The large profit for horses which have final SPs above 4/1 do so despite a much lower Strike Rate and not that far above the 10.3% rate that would be Average across all runners in a 2yo season. This suggests that in many cases horses are lengthening to value prices and the Market is either reacting to incorrect insider information or just plain lack of any input.
In general the SP changes are at a lower level than with the Supported horses. This may be because of Supports reflecting positive input more often but also presumably reflects the fact that bookmakers are quicker to protect themselves by cutting prices than lengthening those of lightly supported runners. The results show a small loss with Drifts above 4% and the 4-5% cut-off point for fuller significance suggested for Supports would perhaps be a little lower with Drifts. But, splitting the levels between 4% to around 10% SP change would be interesting to see if the losses came on any particular range.
The losses recorded by the largest Drifts of over 10%, on a small sample, included half of the total runs that opened at 6/4 or under and only 3 at longer than 3/1. A good lesson to perhaps be careful with short price horses that drift and a useful reminder that the biggest Drifts are not those that appear superfically large. Always worth reminding youself that a Drift from 8/1 to 66/1 is less than a 10% change, for example although it sounds dramatic in raw numbers. A 4/5 favourite going out to 5/4 is a bigger change.
The most important point to note however the final ROI figures would be shifted by a full selection is the variability in trainers' record. As the examples above hopefully emphasise there is a full distribution amongst them of how well the Market can assess them and how well correlated the Market moves are to what the stable expects. A Drift in the Market for one trainer may be truly significant and a similar move be meaningless for another. Even with trainers where there is a Market correlation then further study into items such as owners involved, which run the horse is on, time of the season will show up useful variations in returns. As ever, a far richer and more complicated picture than the simplistic view expected by the media.