Posts

Novice Hurdles: What’s the form worth?

As regular readers of these Punting Angles articles will know, most of the focus is on the staple diet of UK day to day racing, writes Jon Shenton. This is at least partially deliberate. Firstly, there is lots of it and therefore more data to crunch. Secondly, it makes at least some sense that higher class racing is  watched more, tracked more closely, better understood and that it is consequently harder to find an edge from data. After all, there is wall to wall coverage of the big days and events.

Better late than never, it’s time for us to get involved in the upper echelons of the sport. For one or two editions it’s going to be less about Plumpton, Sedgefield and Southwell (with all due respect) and more about Cheltenham, Leopardstown and the like.

This article, the first of two, is solely focusing on Graded novice hurdle races, exuding the mares' programme.  I will be evaluating most of the key dates in the calendar from Chepstow in October up to the festival at Cheltenham in March, with one eye on trying to find contenders for those mega spring festivals. This means events such as the Aintree and Punchestown festivals in April are not included.

To do this, I’ve pulled together data on Graded (Grade 1, 2 and 3) novice hurdles from both sides of the Irish Sea. In all honesty, the process has been quite a long one, and painstaking at times, manually checking race data and inputting it into a spreadsheet. However, it’s been a fantastic education and ultimately a rewarding exercise.  Whilst there are no usual point and click recommendations, I hope it’s of some use in your punting: the process has certainly opened my eyes to the world of novice hurdling.

Approach and method

Don’t worry, it’s not a science paper!   However, I do think it’s important to outline the process that I’ve used as a basis for much of this article.

To the best of my knowledge every Graded novice hurdle race since autumn 2015 (run in October to March of those years) has been evaluated to establish how runners performed over the subsequent 365-day period. That intel has then been pulled into a data table.  Based on key criteria a rating has then been generated to measure the quality of that race based on the future results of participants.

It only includes data up to 16th January this year so doesn’t contain any of the races from the most recent weekend, the Rossington Main, for example.

 

An example using the Tolworth

Sandown's Grade 1 takes place in early January and was won by Fidderontheroof earlier this month.   It is  run over a 2-mile trip and the question I’m trying to answer objectively is whether the race form is worth following or not based on the recent history of the race.  The below table shows a breakdown, by renewal, of the subsequent performance of all competing horses over a 90-day and 365-day period.   Horseracebase has been used to obtain the data.

Hopefully the column names make at least a modicum of sense. But, to explain further, the columns numbered 90 and 365 relate to form for that length of time, in days, after the Tolworth was run. So for example, the 6 in the 90run column for 2019 means that there were 6 runs from horses that ran within 90 days of the Tolworth in that year, the next column (90win) illustrates the number of winners from those 6 runners, 90pl the number of places and so on. Already this table gives a flavour as to whether this may be a race to follow in general terms.

The second part of the standing data shown in the table is evaluating the quality of the future form in terms of wins and places over 365 days, the column headers have “365” titles for clarity.

It is of intrigue that by backing every Tolworth runner blind for 90 days after the race at SP you’d walk away with a profit of £26.20 to a £1 level stake, a 72% return! (The P&L numbers are marked in yellow)

Volume of wins and places is interesting, but it’s helpful and important to understand the quality of those victories. So again, evaluating the Tolworth form in terms of the breakdown of those W’s and P’s in relation to the class they have been attained in, the below table gives the split.

The table shows the number of subsequent runs in Graded races at any level (GPrun) then working across from left to right:

  • G1W – number of G1 wins
  • G2W – number of G2 wins
  • G3W – you’ve guessed it, number of G3 wins
  • OthrW – wins in all other classes (inc. Listed)
  • G1PlOnly, G2PlOnly and G3PlOnly are the number of places attained in those grades, not inclusive of any winners

Summarising, the data paints a picture that, from 2015, there have been 49 runs from horses that competed in the Tolworth who then ran in Graded company during the following 365 days.  Of those, there have been four Grade 1 wins, three at Grade 2, three at Grade 3 and the column OthrW represents 19 wins in Listed or lower classes.

For information, the G1 victories are;

  • 2018 - Summerville Boy in the Supreme Novices Hurdle at Cheltenham
  • 2017 – Finians Oscar in the Mersey Novices Hurdle at Aintree
  • 2016 – Yorkhill in the Neptune (Ballymore) Novices Hurdle at Cheltenham and the Mersey Novices Hurdle at Aintree

These three animals prevailed in the Tolworth, all progressing to festival success and rubber stamping it as a race to follow.  You’d be right in thinking that fancy data is not required to confirm that the Grade 1 Tolworth is a strong contest.  However, understanding how the race compares against other quality races in terms of future form is of potential interest.

To contemplate its relative strength against other events, finding a way to rate or score each race is required. As a result, a relatively straightforward race rating has been constructed to do the job. The race rating system is highly subjective and there is a strong suspicion that if a hundred people did it, no two individuals would do it in the same way!  The exact method isn’t too important though, as the objective is to evaluate without bias which novice races are best to follow. A relatively simple (even with flaws) rating system still should give enough detail to be a bridge to further analysis.

Below is a breakdown of the ratings for the Tolworth, 2015 to 2020.

Here is a quick run-down of the columns and what they represent;

365%Score – this relates to the general quality of the race.  It’s calculated in the following way.

(The Win % of the race form for 365 days) + (the Place % of the race form for 365 days divided by 3)

A real example, the 2019 has 3 winners and 8 places from 19 runs (data in the first table), resulting in a winning percentage of 15.8% and the place percentage of 42.1%. Therefore the 365%Sc is calculated as below

15.8 + (42.1/3) = 29.8. All scores are rounded to the nearest number. A minimum of 10 runners in each race is required to generate a rating. 

GPWScore – this represents a rating generated from the subsequent winners over the next 365 days from each race.  The scores are comprised of;

  • Grade 1 win 10
  • Grade 2 win 6
  • Grade 3 win 3
  • Other Win 1

GPPLScore – this is the score value generated from the placed horses (excluding winners) over the following 365 days

  • Grade 1 place 3
  • Grade 2 place 2
  • Grade 3 place 1

GPScore = GPWScore + GPPLScore (i.e. a combined score from the win and place data)

RTNG  = the overall rating for the race in question, adding 365%Score + GPSccore

RaceRNK – is the overall rank of the race in terms of quality from the 163 races evaluated.   The lower the number the better.

Therefore, in the case of the Tolworth, the 2016, 17 and 18 renewals were relatively strong, with the 2016 renewal having the 7th best race rating in the dataset. 2015 and 2019 were disappointing with rankings of 122 and 97 respectively.

And that’s the process, fully transparent and easy to follow, I hope.

That’s quite a long scene set and explanation, but necessary in my view!

Onto the results...

 

Novice Hurdle Race Ratings

The below table is a consolidated summary of all of the analysed races from the years 2015 to present day and, as explained previously, only contests that have 10 or more subsequent runners are included in the data (the number of qualifying races is shown in the column titled QualR).

The table is sorted by the highest average rating of the race over the 5-year period.

 

There is unsurprisingly a large variance in quality based on subsequent 365-day form, from the Prestige, averaging a rating of 32.8, to the Chanelle Pharma, previously known as the Deloitte, averaging 96+ at the top of the pile.

It is at least mildly reassuring that the Grade 1’s feature in the higher end of the table in general. The Tolworth ranks in joint 7th confirming the view from the opening section that it’s a solid race to follow.

As someone who struggles to keep on top of the racing calendar and track the key movers and shakers, these data focus the mind. The bad news is that from here onwards there are no easy answers or instant takeaways: the only truly effective way to progress to a deeper understanding is good old-fashioned hard work and metaphorical elbow grease.

Having said that, interestingly, the 365P&L column shows in yellow where backing every subsequent runner from the events in question for 365 days post-race has been profitable to a level stake of £1 at SP. The fact there are so many is a pleasant surprise and worthy of more focus; there may be something to consider for building profitable angles, but ideally more than five years of race data would be needed to have the necessary confidence to invest.

For now, as a starter for ten, a quick dive into a couple of the prevalent races to follow seems a sensible path to follow.

Chanelle Pharma Novice Hurdle

This rating system shows that the Chanelle Pharma Novice hurdle (Deloitte until 2019) at Leopardstown is a clear and obvious winner with an average score of 96.2. That's higher than the second placed Cheltenham Classic Novices Hurdle by over 18 points! The Chanelle Pharma is now contested over a 2-mile trip since the newly-formed Dublin Racing Festival became reality in 2018 (it was previously run over 2m2f) and it is well known as a good pathfinder towards the Supreme and Ballymore in March.

This novice event has racked up 44 subsequent winners from 200 runs with a £1 level stake loss of £18 if you’d backed every one blind up to a year after the race.

Below is the view by renewal year, using the key columns described earlier.   Immediately the eye is drawn to the RaceRNK column, confirming that this contest had the 1st (joint), 3rd and 6th best individual races in the novice sphere since 2015.

Significantly, 19 of those 44 wins were delivered in elite Grade 1 company. That’s a whole ten more than any other race on the list and obviously worth delving into.

On closer inspection, those 19 triumphs are attributable to 11 individual horses. Nicholls Canyon with 4 of the victories (from the 2015 renewal), Sharjah with 4 (2018), Le Richebourg (2018) & Petit Mouchoir (2016) with 2 each. With sole G1 wins secured by Klassical Dream (2019), Samcro (2018), Barcardys (2017), Bellshill & Coney Island (2016), Windsor Park & Identity Thief (2015).

Perhaps surprisingly, there are only a trio of same season Cheltenham Festival winners after competing in the Chanelle Pharma for the analysed races. Klassical Dream won the Supreme last year with the two other two Prestbury Park winners coming in the Ballymore, Samcro in 2018 and Windsor Park in 2015.

It is noted that the 2019 renewal has had a relatively disappointing outturn. The law of averages perhaps would nod to a better 2020 vintage.

The Ballymore Novices Hurdle (Classic Novices Hurdle)

The second race on the list by some distance is the Ballymore Novices Hurdle (Classic) run at Cheltenham on Trials Day, which is very much on the radar for the upcoming weekend. Arguably, this race is a better one for the trackers than its Leopardstown counterpart as it’s delivered a £1 level stake profit of £57, through backing all runners each time they took to the track over the subsequent 365-day period.  That’s nearly a 36% return which seems utterly insane for 5 years-worth of renewals encompassing 159 total runners.   Perhaps it is the fact that it’s a Grade 2 which may drive some of that potential future value. Whatever the reason it’s a race about which to sit up and take notice.

I’ve added the 365P&L column to this table showing the value of backing all runners blind at SP for each renewal of the race. This event has a solid feel in terms of consistency, and whilst there have been 12 fewer G1 wins than the Leopardstown race previously discussed, the overall number of winners is only one fewer at 43, from a much smaller number of runs too: 159 compared to 200 in the Chanelle Pharma. Each Classic renewal has generated its share of future winners, with the 2016 version being the cream of the crop with a RaceRNK of 7.

Considering it’s an event which occurs on Cheltenham Trials Day, a good starting point would be to check how horses go on to perform at the big event.

It’s no silver bullet based on the last five years' data, that’s for sure. Not a single winner has been drawn from this race at the Festival in the same year, although it must be stated that five years is not a significant sample size. Also, in fairness, the crossbar has been rattled several times with Yanworth and Black Op coming close in the Ballymore, and Santini, Champers on Ice and Wholestone hitting the frame in the longer distance Albert Bartlett. Black Op and Santini did go on to enjoy Grade 1 victory at the Aintree Festival a month or so later in the Mersey and Sefton respectively. Several horses have developed into Festival winners in future years too.

On the point of future winners, whilst trawling through the results it was very easy to spot some eye-catching names finishing in eye-watering places in this contest historically. It’s best represented by this result card from the 2015 renewal.

Whatever happened to some of those also-rans failing to complete or trailing in 60 or so lengths behind the winner (whatever happened to the winner too?!)?

Whilst it’s a stretch to claim this picture is typical there are a whole raft of horses in this event who go on win on much bigger stages, often chasing ones too. In no particular order, Topofthegame, Elegant Escape, Slate House, Poetic Rhythm, Royal Vacation, William Henry and, going back further, Whisper, Coneygree and The New One have all cut their teeth in this race. That is an impressive roll call, which bodes well for Birchdale, Brewin’upastorm and Jarvey’s Plate from the mildly disappointing up to now 2019 crop.

Originally, I planned to go into more detail, but the powder will have to remain dry for a second part (this is already too long!) where I’ll cover the potentially profitable races to follow in more detail; including analysis of a horse's next run only after competing in one of these Graded novice hurdles.  I’ll also be evaluating the winners of the novice hurdles at Cheltenham to ascertain if there are any patterns linking back to the races included in this article.

- JS

p.s you can read PART TWO of this novice hurdle analysis here

Punting Angles: Kempton Park

Kempton Park is dripping in racing heritage, having staged its first event more than 140 years ago, writes Jon Shenton.  However, it is the polytrack racing that has been the most prominent fixture from 2006, and that will form the content for today’s piece. There are plenty of data to get stuck in to, hardly surprising considering the number of fixtures at the venue.

The course map reminds us that Kempton is the only right-handed all-weather track in the UK, and it also highlights the existence of two racing loops. Only the five-furlong and 1m 2f trips use the inner ring, the other distances all charting the outer course.

As a supplementary starter, if you want a real expert opinion on the track, David Probert’s blog was published on geegeez a few months ago and contains some very useful first-hand snippets from a rider’s perspective.  It certainly sets the scene nicely for this article if you have time.

https://www.geegeez.co.uk/catching-up-with-david-probert/

Kempton AW Trainers: Richard Fahey

As usual, let us first delve into the performance of trainers at the track. Before getting into the positive angles it’s worth noting a high-profile and generally prolific yard that appears to a have a few challenges at the Sunbury circuit.

The above data represent the powerhouse Richard Fahey team at Kempton from 2012 onwards. A strike rate of less than 4% is not fantastic by any measure and such runners should perhaps be given second thoughts prior to investment. That said, earlier in 2019 George Bowen was a Class 2 winner from just three runners this year.

Kempton AW Trainers: General

Moving into positive territory, below are the best performing trainers (still active) at the track since the same 2012 date.

To qualify for the table, 75 runners are required with minimum at SP’s of 20/1 or less and a bar of an A/E of over 1.10 needs to be overcome.

Frankly, the list is quite underwhelming in terms of potential angle development. All are probably worthy of further analysis, but nothing really jumps off the page.

Kempton AW Trainers: Rae Guest

However, for some reason it feels impolite to move on without at least a cursory glance at the top of the list. So, with that in mind, an evaluation of Rae Guest’s numbers is in order.

I find that a key factor to always consider when analysing all-weather data is the time of year. I’m now into my fourth annual wagering cycle and am getting a better feel for performance variation and seasonality impact within my portfolio. As I’ve mentioned in previous articles all-weather punting is my staple diet and where most of my effort is centred.

However, being brutally honest, my all-weather angles generally under-perform over the summer months. It may be usual variance but each summer I watch my bank (from AW) glide downwards to then power up over the the winter. It makes sense, the majority of AW racing occurs through the colder months with many yards gearing around the season, or potentially focussing their efforts elsewhere during the summer months.

The Rae Guest info does show some of the hallmarks of that fallow summer performance. The below table illustrates the yard results at Kempton for May to August (inclusive)

Granted, not a huge number of runners, but not the best record either. It seems logical to check this record by opening the data to the yard’s performance across all AW tracks over the same period to see if there is a general downturn or if it’s course specific.

It’s a slightly better record, but still somewhat underwhelming as a collective.  The companion data (from the other months) across the artificial tracks may be of interest and is as follows:

That’s a pretty impressive record relating to over 300 runners and indicates the Guest yard is generally one to track on the artificial surfaces.

Delving deeper, here is a view of performance by race class.

The data above show a 1-from-18 record in Class 1 to 3. That’s most likely a representation of the materials available to the yard in terms of equine talent rather than any training limitation. It might be argued that Class 4 races are marginal from a betting perspective, too, with a strike rate of 11.6% and an A/E of 0.72 but for now, at least, they remain included.

There is also something very interesting when splitting out Guest runners by gender as the numbers below illustrate:

Taking the not specified gender (I assume missing data) out of the equation over 80% of the horses competing for Guest are female. This is quite unusual and even more interesting is that these female animals are outperforming their male counterparts, at least in market terms (A/E 1.30 vs. 0.98).  It must be noted that strike rates and IV are broadly similar.

In general terms, fillies and mares underperform on the artificial surfaces compared to colts and geldings. Strike rates for females are approx. 12.5% vs 14.2% for the male runners with A/E measuring 0.85 vs 0.88 since 2012, that’s an evaluation of 145,000 runners. Therefore, the Rae Guest yard seems to buck the trend and consequently there could be value in backing his fillies as a result. Perusing their website for horses currently in training, the majority are fillies so perhaps it is as simple as specialising in the development and training of the fairer sex. Nevertheless, it is worth noting all the same.

Suggestion: Back Female Rae Guest All Weather runners from September through to April in Class 4-7 races at an SP of 20/1 or less

 

Draw at Kempton

To search for clues in terms of which race distances to drill down into, the table below contains a summary of all distances up to a mile and a half using the Draw Analyser tool from the Gold toolkit.

Essentially the numbers demonstrate by race distance the average IV3 number (Impact Value of a stall and its nearest adjacent stalls) for each draw. It’s not perfect, but it does offer solid indications regarding where to look more closely, as well as giving a good reference table for general study. A summary of the key findings are:

  • The low draw bias looks most acute on the inner-course 5-furlong trip
  • Inside/low draws also appear to be beneficial for other distances up to 7-furlongs
  • Races at a mile and above show a slight accent to favouring more mid-range draws, with perhaps the most pronounced being for the mile and a quarter (10f) trip around the inner loop.

On the back of that it seems prudent that a detailed analysis of the two inner-course trips would be the most sensible use of word count.

Kempton 5 Furlong Draw and Pace

Firstly, a point of order: with all races at Kempton a low draw is closest to the inside rail and all data from here on relates to Standard and Standard/Slow going using actual stall position (not card number), that is taking out non-runners.

Over the minimum, at least half of the burn-up takes place around the inner course bend, so a low draw can mean travelling a shorter distance than the competition because claiming a spot close to the rail should be a simpler task.

The above table shows the numbers in more detail by specific field sizes (the column RN means number of runners). It’s in the usual format for regular readers. If you’re new to it then the left-hand section shows the IV3 number for each stall position by number of runners; the right-hand table shows performance in relation to early track position, i.e. pace, for the same field sizes.

Firstly, draw. The green colours are largely concentrated in the lower stall numbers, confirming the reasonable bias towards these positions. Interestingly, the greater the number of runners the more pronounced the bias appears to be. Incidentally, the maximum number of entrants over the five-furlong distance is twelve; however, the volume of races with a full field is very small so I’ve ignored them within this analysis.

The pace data is very interesting. In very basic terms, the horse that gets to the front early has at least twice the chance of emerging victorious: early speed is a huge advantage.

Given what we know about the five-furlong course topology, we’d expect to see that. If an animal can get to the front around the tight inner course loop it’s going to be in pole position, given the almost constant turning nature of the trip.

Early pace is undoubtedly a great asset, a low draw is also a great asset. So, combining both, surely must be a licence to print money? Well, yes and no, it’s not quite as simple as that. Why? Because it’s widely understood that a low draw is advantageous on the Kempton polytrack, so it’s probable that stall position is factored into available prices.

To establish the effect of the draw on value, the below table contains the equivalent A/E information for the race set ups covered in the IV3 table. As a quick reminder, A/E is an index of market value where 1 is neither good nor poor value, and a number above or below is good or poor respectively. The further away from 1, the better or worse things are.

The numbers do arguably ratify that the market has stall position covered in its starting prices.  The average (AVG) data confirms that A/E performance, whilst marginally better in the lower draws isn’t market busting by any means with averages for stalls 1-3 around the 1.00 mark: eking out a profit from picking low drawn runners may be a long-term challenge despite the clear higher propensity for providing winners, at least at industry SP.

If draw doesn’t necessarily give the edge that is craved, perhaps pace can. To try and get under the skin of the impact of pace by stall position, Gold’s Query Tool can assist.

The next table is using the tool data purely with the purpose of analysing only front runners by field size and starting gate. The reason for doing this is to try to understand if there is any commercial advantage in identifying these leaders by stall position.

The filters used in QT are:

Distance:            5-furlongs

Course:               Kempton

Race date:          1/1/2012 or later

Pace score:        4 (which is used to designate the early speed/lead horse)

The data is split by number of runners and again shows the A/E (performance against market expectation).

Initially, it appears that it’s a stiff ask to win from the widest draws even if the horse is an early speed merchant.  There is the most sizeable of sizeable caveats here though: the data samples are miniscule in places (so, for example, horses in stalls 9 and 10 in field sizes of 10-11 have only led in six races at this distance, with no leaders from stall 11).

These numbers confirm that front runners beat the market under all conditions apart from the aforementioned widest of the wide (the zero in stall 4, field sizes 6-7 is simply a quirk of a small data set). The numbers do, however, indicate greater value in the mid to wide gates, particularly in bigger fields. Small samples notwithstanding, this is worth due consideration.

To illustrate this point as a final check, here is the raw data from the Draw Analyser tool for races of 9-11 runners. The data contained within the blue dotted line illustrate the fate of the early pace (led) horse by draw position, split into thirds.  Win% across low/med/high is consistent at 22-25%, IV is marginally better in the lower drawn animals, emphasising they are more likely winners. But A/E is comfortably at its strongest in the higher drawn leaders at 1.81.

Looking for speed first, draw second and not self-talking myself out of a value play because of a wide stall is the main lesson I’ve taken from this info. Very similar to the last article on Chelmsford in that respect.

Suggestion: Try to identify the early leader in five-furlong races at Kempton

 

Kempton 1m2f Draw and Pace

Before wrapping up, a quick overview of the Kempton mile-and-a-quarter landscape is in order. A reminder that, if anything, there was a mid-to-high draw bias indicated in the initial numbers which piqued interest levels, and also keep in mind that this range also uses the tighter inner loop with the shorter finishing straight.

Below is the now standard format for assessing the pace and draw data.

The data seem to illustrate a reasonably fair and flat draw profile, apart from perhaps the outer stalls in large fields where it seems there may be too much to do.

The lowest gate numbers become increasingly difficult when the number of runners increases to 11 or greater. That is probably when horses are starved of room in the larger herd when forced/taken back during the early stages.

There is no doubt that a mid to “quite” high draw is no bad thing over this course and distance which is a mild surprise given the tight nature of the inner loop. However, in relative terms there is ample time from the starting position to the first bend, and up the back straight, for most horses / jockeys to find a position and avoid a wide trip.

These mid-range draws seem to offer greater flexibility in the run, giving lead animals the chance to get out in front, while hold up horses have less propensity for being trapped at the business end of the race.

Again, early pace is advantageous, as it is in most circumstances. However, the benefit isn’t quite as marked as some of the other trips or courses analysed in this series. In fact, the Hold-Up and Mid Div numbers hold up (!) relatively well considering there will likely be plenty of also-rans contained therein.

Using the draw analyser summary for the 11-14 field sizes (where low draws seem to underperform), the blue dotted box shows the challenge faced by a held-up low drawn horse.  Ridden for luck appears to be generally unlucky in this case. Any horse that is generally slowly away or repeatedly held back at the start should be treated with the utmost caution over this trip if its stall number is low.

Yet again, though, there appears to be some value to be gained from high-drawn leaders if they can be discovered (red dotted line). The prominent high-drawn animals don’t perform too badly either in market terms.

Hopefully the above ruminations will assist during the upcoming winter nights when poring over the Kempton form.

 - JS