This is the third in my series of articles on draw and pace bias at UK courses, writes Dave Renham. Pontefract and York were the first two to come under the spotlight, and now it is the turn of arguably the most draw biased track in the country, Chester.
Chester is by all accounts a lovely place to go and watch racing and, as I write this piece entering the fourth week of lockdown due to the coronavirus, the Roodee is a course that I will make every effort to visit in the future. It lies close to the River Dee and is officially the oldest racecourse in the world, dating back to 1539.
Chester Racecourse is a very tight track only just over a mile in circumference, which means that it does not suit long striding horses because the runners are so frequently on a left-hand lean. The home straight is only 240 yards in length - barely more than a furlong - which tends to help horses close to or on the pace. Traditionally, lower drawn horses near to the inside rail have had a decent edge at most distances. This bias is well understood across the racing fraternity, though, so whether we can actually get a worthwhile betting edge constitutes the purpose of what follows.
As with the last two articles, I am using some of the tools available on the Geegeez website: Draw Analyser, Pace Analyser and the Query Tool. The main period of study is a long one, going back to 2009, but I will also examine a more recent (2015 to 2019) data set in detail where appropriate. I will be focusing once again on 8+ runner handicap races only and looking exclusively at the distances from 5f to 7f.
When looking at each race I will be dividing the draw into three equal sections (low, middle, high). This is how the Geegeez Draw Analyser does it and has been the way I have generally done it for 25 years. Draw positions are also adjusted when there are non runners to make the data as accurate as possible. For a strong draw bias I am looking for a figure of over 50% for one ‘third’ of the draw, and I am looking for this starting point to ideally correlate with other metrics such as A/E values.
Chester 5 furlongs (8+ runner handicaps)
Since 2009 there have been 86 races that have qualified. I have also included races over five and a half furlongs, of which there of which there were 29. Here are the overall draw splits:
As can be seen, very strong bias exists which has been consistent year in year out. Below I have split out the 5f data and 5½f data for interest:
The low draw figures mirror each other, but the middle and top thirds seem to ‘reverse’. Possible reasons could be that high draws really don’t have enough time to recover over 5f, whereas the extra half furlong gives them more of an opportunity to get back into it. It could be that middle draws get ‘squeezed' somehow over 5½f. Of course it could perfectly conceivably be down to random results as there were only 29 races in total for the 5½f stats.
Onto A/E values now (5 and 5½f races) and low draws have actually offered value over the longer term:
In terms of breakdown, the five-furlong races produced a low draw A/E value of 1.05, whereas races over 5½f had a slightly better value of 1.14.
It is time now to look at each draw position broken down by individual stall number (5f and 5½f):
As we can see, stalls 2, 3 and 4 have all made a blind profit to SP, and stalls 1 to 4 have provided 63 of the 86 winners which equates to 73.3% of all races. Horses drawn 11 or wider have provided just two winners from 74 runners with only three others getting placed. The huge advantage to low draws is clear.
We can consider that macro picture in context by looking at a more recent data set covering the past five seasons (2015-2019). Here are the draw splits for this shorter time frame (total races 42):
Low draws have been consistently winning at greater than the 60% mark so on the face of it the bias seems as strong as ever. For those who like exotic bets, doing a reverse forecast on the two lowest drawn horses would have netted you an amazing ten winning bets in 42 races. Now the dividends varied greatly as they are dependent on the prices of the horses that fill the first two places. Three of the dividends paid under £10, but two paid between £50 and £70. If you had placed a £1 reverse forecast on all 42 races going back to 2015 you would have been in profit by an impressive £182.41. For any tricast punters out there, if you had permed the three lowest draws in 6 x £1 full cover tricasts, you would have won 4 times and netted a profit of £189.85. I cannot guarantee such returns in the future but it is certainly food for thought.
Here are the 5 year stats for individual draw positions:
In this shorter snapshot, 71.4% of all races have been won by horses drawn 1 to 4, which correlates with the 11 year figure of 73.3%.
Despite everything looking very rosy still for low drawn horses and amazingly still producing some profitable angles I do have a word of caution: I think there are signs that the bias is getting less strong, more notably in the past two seasons. If you look very closely at the 2018-2019 results as a whole, the low draw bias seems less pronounced. Now there have only been 15 races during these two seasons so it could simply be an anomaly due to an extremely small sample.
However, there is a reason why I think this might be the case rather than simply hypothesising over a set of numbers. In the last two seasons 12 of the 15 qualifying races (80%) have been run with the inside rail having been moved. This is almost certainly an attempt by the course to try and nullify the bias as best they can – indeed rail movement has risen from 22% in 2016 to 50% in 2017 to this new figure of 80% during the past two seasons. The rail movement is also not consistent from meeting to meeting having moved different distances ranging from a minimum of 10 yards to a maximum of 33 yards.
Digging deeper into these very recent results, in the three races when the rail was not moved in 2018/19 all three races were won by low drawn horses – in fact both 1st and 2nd were filled every time by horses drawn either 1, 2 or 3. In the 12 races where the rail was moved, less than half of the races (five) were won by the bottom third of the draw and generally the races looked far more even when looking at placed positions too. As mentioned above, the data set is far too small to be anywhere near confident, but it will be interesting to see if this emerging pattern continues in the near future.
Let us consider pace and running styles now. For the pace section I am going to study just the overall figures (2009-19):
A significant edge has been to front runners which win roughly three races in every ten while, in contrast, hold up horses are at a huge disadvantage here. That is almost solely due to the tight turning nature of the track, especially at this trip, coupled with the very short straight. Hold up horses just haven’t got enough time to pass the many runners that would be in front of them as they straighten up for home. The figures are very similar for both the 5f distance and the extended 5 furlongs (5½f). Front runners also produce an impressive A/E value of 1.84.
In terms of whether the going makes a difference, the figures are fairly even across the board although on good ground (34 races) the win percentage for front runners edges up to just over 34% with an IV of 3.79. I don’t personally believe the front running bias is stronger on good ground – the place data is no stronger for example - but I felt it was worth mentioning.
In terms of field size and the number of runners, the front running bias seems strong across the board; it is possibly slightly stronger in smaller fields but when you split the data up some of the sample sizes are a little small to confidently make that inference.
Now a look at draw / pace (running style) combinations for front runners in these 5f races. Remember this is looking at which third of the draw is responsible for the early leader of the race (in % terms):
As one would expect, the early leader comes from lower draws more than half of the time; it is clearly difficult to lead early from a wide draw which makes perfect sense given the appeal of such a position and the topology of the course. It is also easier to win having led early from a low draw rather than from a middle or high draw. Low drawn early leaders have gone onto win 37.2% of their races (23 wins from 61) – an impressive stat.
Having said that the figures are still solid from middle and higher draws: middle draws have won eight times from 35 (SR 22.9%) when leading early, and high draws have won four from 14 (SR 28.6%). Digging deeper, horses that take the early lead from stalls 1 and 2 do lead more often than any other draw and go on to win around 40% of the time.
Below is the draw/run style heat map, displaying place strike rate, for 8+ runner five- or five-and-a-half furlong Chester handicaps since 2009:
The summary is that the combination of a low draw coupled with good early pace, or at least the ability to lead, is extremely important at Chester over 5f.
Chester 6 furlongs (8+ runner handicaps)
The 6f distance is run less regularly than the 5f / 5½f one, with 66 qualifying races going back to 2009. Here are the overall draw splits:
The strong low bias seen at 5f does not occur in the same way here with middle draws being almost as successful as low draws. What is clear however, is that there continues to be a strong bias AGAINST high draws. The A/E values illustrate that middle draws have definitely been the value:
A look again at individual draw positions and how they have fared over this time frame:
Although the lowest third of the draw did not totally dominate the win percentages, draws 1 and 2 have been very successful in terms of wins: 25 wins out of the 66 races equates to just under 38% of all races won by the two lowest boxes. Both have made a blind profit – draw 1 having performed particularly well. It should be noted, not unexpectedly of course, that double figure draws have provided just one winner from 118 runners!
Looking at ground conditions it is possible that firmer ground accentuates the bias against higher draws a little, which seems logical also. On good ground or firmer the top third of the draw have won just 4 races from 38; on good to firm or firmer they have won 0 from 12 races with only 5 horses placing. Limited data, yes, but something to bear in mind I feel.
Onto the last five seasons now. Although there have been only 30 qualifying races since 2015, I believe it is still worth sharing the draw splits:
High draws have continued to struggle in the more recent past, while middle draws have performed slightly better, and the centre is where the value seems to be once again. Double figure draws have secured zero wins and just four places from 46 runners. For A/E values I am going to split the data by draws 1 to 4, then 5 to 8 and finally 9 or higher:
Again, it appears that draws 5 to 8 have been the value stalls in the past five seasons.
Let's break down the individual draw figures for the last five seasons, 2015 to 2019:
With only 30 races in the past five seasons the individual stall data is rather limited and I personally would not read too much into it. Stall 1 has performed well as one might expect in the context of the 11 year data shared earlier.
Below are the running style figures for 6f 8+ runner handicaps at Chester, going back to 2009:
These figures show that front runners have a decent edge while horses that track the pace also perform above the expected ‘norm’. Horses that race mid pack or at the back early are at a clear disadvantage for the same reason that they are over 5f: the short straight makes it very difficult to win when coming from off the pace. Horses held up in the back are worse off than horses that race mid-division.
This pace bias AGAINST hold up horses strengthens as the field size increases. The table below looks at the 11 year splits for the 32 six-furlong handicap races with 11 or more runners:
As can be seen, there were no wins at all for hold up horses, whereas 26 of the 32 races were won by horses that raced close to or up with the early pace. The majority of races (18 of 32) were won by prominent runners, though there were, naturally, more prominent races than race leaders and their peer group strike rates are very similar.
Now a look at draw / pace (running style) combinations for front runners in 6f handicaps (2009 – 2019):
Over 60% of the races have seen the early leader come from the bottom third of the draw (low), a percentage that is even higher than the 5f data. This is a very high figure and worth noting.
The 6f trip does not give low draws the edge, however, as it does over 5f. Rather, only the two lowest stalls as noted earlier have managed to win consistently more than might be expected, taking roughly 38% of all races.
Horses drawn 10 or higher have had little or no chance of winning throughout the eleven-year sample period.
Pace wise, early leaders and prominent runners are clearly most favoured.
The graph (sorted by IV3, the average Impace Value of a stall and its immediate neighbours) and heat map (displaying A/E) below both illustrate this:
Chester 7 furlongs (8+ runner handicaps)
There are two 7f distances at Chester, one over 7f and the other over 7½ furlongs. In this article I am going to look solely at the shorter distance of the two. I plan to look at the extended 7f trip in a subsequent 'Part 2' article. Since 2009, over 7f there have 75 qualifying races.
Low draws have had a strong edge having been responsible for just over half the winners. The A/E values look like this:
The market definitely factors in the advantage that lower draws generally have here, an A/E figure of 0.94 implying that there is a slight negative expectation from backing such runners. Having said that there may still be some value in backing selected lower drawn runners.
Let us now look at each individual draw and its associated stats since 2009:
36 of the 75 races were won by one of the three lowest stalls, but only draw 1 has shown a blind profit during the long period of study.
When checking the data for specific ground conditions the stats suggest that on softer ground the low draw bias increases. On good to soft or softer there have been 37 races of which 22 (SR 59.5%) have been won from the bottom third of the draw (low). The A/E value nudges up to 1.09, and backing all low drawn horses would have secured a small profit to £6 to £1 level stakes at SP.
Time now to switch attention to more recent results and the past five seasons. Here are the draw splits for 2015 to 2019 (total races 33):
The bias has been similar in this shorter time frame although in medium to bigger size fields the results have been quite even. Low draws have dominated in 8 and 9 runner races, but as I have said before it is generally not smart be too dogmatic about results when looking at very small samples.
Let us look at the individual draw figures for 2015 to 2019 (33 races):
Horses drawn 3 have done well, but although very low draws have had an edge, when looking closely at this table I am starting to think that the bias has not been as strong in recent years. My main reservation is that the place percentages for draws 1 to 7 are relatively similar and also the A/E values for draws 7 to 10 are higher as a group than draws 1 to 4, suggesting that any value that might exist is in that counter-intuitive area of the stalls.
The great thing about statistical research is that different individuals will interpret the data in different ways. This is simply my view: I could be wrong and once again it needs to be said that 33 races is still a very small data sample in which to have any real confidence.
Onto to 7f handicap running styles now. Here are the overall stats going back to 2009:
Front runners have had a fair edge with hold up horses again at a disadvantage. However, the bias is not as strong over 5 and 6f, with later runners having greater opportunity to get into a challenging position.
We actually have a reverse pattern to 6f in terms of field size. Over 7f front runners have had a huge edge in smaller fields: in races of 8 to 10 runners, (43 races in total) they have secured an impressive strike rate of over 28% with a huge A/E value of 2.24. I cannot explain why - perhaps it is because they are better able to stack the field up in behind and control the race - but the strike rates and A/E values do correlate strongly.
Finally let us examine the draw / pace (running style) combinations for front runners in 7f handicaps:
Low draws are once again more likely to lead as they are closest to the inside rail and thus have the least distance to travel. High drawn runners manage to get to the early lead in just one race in six on average. In terms of early leaders that have gone onto win the race, lower drawn front runners win more often (24.4%) than middle drawn ones (17.2%) who in turn go on to win more than high drawn front runners (13.3%).
It is also worth noting that hold up horses drawn low have won just once in 58 runs.
Combining the 5 to 7f data
Before I finish I wanted to combine all the data for the 5 to 7f distances at Chester (2009-2019). Firstly the draw splits:
Over half all of races in the full eleven-year study period were won by the bottom third of the draw; middle drawn runners have been roughly twice as successful when comparing them to higher draws.
Now the individual draw by draw data:
This table neatly demonstrates that despite the very well known low draw bias the very lowest draws continue to offer some value. Both draws 1 and 2 have made a blind profit to SP and both have A/E values in excess of 1.00, from a strike rate of almost 19%. The table also illustrates neatly that in general the higher the draw the lower the chance of winning: stalls 11+ have a collective win record of just six from 215 (2.8%).
The table below shows the combined pace data for 8+ runner 5 to 7f handicaps:
Looking at these figures, why on earth would you not try and get to the front at Chester over distances of 7f or less? Jockeys who hold up their mounts here by choice are either not trying to win or they don't know how to!
I hope this article has given you plenty of positive angles from which hopefully we can profit sooner rather than later.