Posts

Cheltenham Festival 2021: Trainer Form

We're just a week away from the biggest gathering in jump racing, the Cheltenham Festival, and what a chaotic lead in it has been this year. As if a pandemic wasn't enough, we have subsequently had to endure reservations about ease of transit for Irish- (and French-) based runners post-Brexit, the shocking Gordon Elliott revelations, and rumours of a pan-European equine herpes outbreak potentially throwing a further spanner into the works. On top of all that, now we have the Meghan Markle story!

OK, so that last one, and probably/hopefully the last two, are of no consequence to Cheltenham, mercifully; but the others have each caused some degree of consternation in the weeks and months preceding the Festival. With six days until tapes rise on the Supreme, we can hope that all will hereafter be more serene, barring the perennial raft of late scratches and shock race switcheroos. So we can get down to business, the business of this post being to review current trainer form for the big guns heading into Cheltenham Festival 2021.

How to quantify trainer form?

This presents an immediate and obvious question: what actually is trainer form? When referenced in general - "xyz is in flying form" - it normally means xyz has had a couple of winners recently. Is that 'in form' or merely the happy end of the variance spectrum? How can we even things out and judge trainer performance more broadly? And should we even bother given that winners are winners and losers are losers, right? Plenty to chew on here.

Let's start with a pretty much unarguable contention: trainer form is how well or poorly a given trainer is faring at a point in time. So far so meh. The challenge is isolating an agreeable metric (or metrics) against which to vaguely scientifically measure form; and to then further layer on the wagering component of profitability (and, of course, how best to measure that).

Happily, geegeez.co.uk publishes a few metrics that cut through the thorny thicket of quantifying these data, namely Impact Value, Percentage of Rivals Beaten, and Actual vs Expected. We do also display win and place percentages but, in truth, these are the equivalent of answering the question, "What time is it?", with "Tuesday afternoon".

Let's (very) quickly recap what the numbers mean.

Impact Value (IV) is a measure of how frequently something happens for x in relation to how frequently it happens for all. For instance, how often the going is good to soft on the opening day of the Cheltenham Festival compared to how often the going is good to soft over all days of the Cheltenham Festival.

A figure of 1.00 means the 'thing' - Cheltenham day one good to soft, in this case - happens the same amount as in the wider set of data. A figure above 1.00 means it happens more often, below 1.00 signifies that it happens less often. The further away from 1.00 the IV the more or less something happens in relation to the round.

Still here? Topper, PRB next.

Percentage of Rivals Beaten (PRB) is a means of applying a sliding scale of merit to every finishing position, and doing it in relation to the field size in which that finish was achieved. For instance, 2nd of four has less merit on this metric than 2nd of 40 - and rightly so, of course.

In this case, 2nd of four beats two horses and loses to one horse, so has a PRB score of 67% (or 0.67) for beating two out of three of its rivals.

2nd of 40 beats 38 horses and loses to one, so has a PRB score of... gets calculator... 38 beaten divided by 39 total rivals = 97%, or 0.97.

When looking at a combination of events - say, all trainer's runners over a period of time - we can derive an overall PRB figure and use that for comparative purposes.

Actual vs Expected (A/E) is the betting number. It works as an index where, like IV, 1.00 is a par figure and better or worse than 1.00 is a degree of good or bad respectively. It's calculation requires a little unpacking and, rather than do that here, you're encouraged to look at this racing metrics article where I explain and exemplify each of IV, A/E and PRB in more detail. The key here is that north of 1.00 is good, south of 1.00 not so much.

How to quantify trainer form pre-Cheltenham?

So we'll use IV, PRB and A/E as way points to navigate to a conclusion; but against which period(s) should we measure performance? It probably makes sense to compare a longer period pre-Cheltenham with a shorter period pre-Cheltenham with performance at the Festival itself.

Willie Mullins

The following table has performance data for Willie Mullins-trained runners (WPM) for three different periods in each of the previous five seasons:

- The four days of Cheltenham
- The four weeks prior to Cheltenham
- The four months prior to the four weeks prior to Cheltenham

 

Looking for correlation is difficult in what is, granted, a crowded table. And it is still more confusing when noting that comfortably Mullins' poorest win strike rate (6.78% in 2019) produced his best ROI (+30.51%).

The message there is simple enough: the microcosm of win strike rate - and indeed Impact Value - in tiny sample sizes is misleading. At the same 2019 Festival, Mullins could boast a 25% each way strike rate, in line with placed efforts at the two preceding Festivals. In PRB terms, 2019 only ranked third of five.

Below is the same information but with the key metrics ranked, e.g. Mullins' 2020 Festival win percentage was his second best of the past five Festivals; it was his best of five Festivals on each of EW%, PRB, IV, and A/E.

 

Looking for correlation is tricky. It can be said that the 2018/19 season was not great for Mullins, and that was something which manifested almost across the board at the Festival - with the counter-intuitive exception of ROI. Closer scrutiny reveals that Willie backers were saved by the 50/1 success of Eglantine Du Seuil in the Mares' Novices' Hurdle as well as, to a lesser degree, Al Boum Photo's 12/1 maiden Gold Cup score. Take out the big priced winner and it's -29 and ROI rank 4.

One thing that is reasonably clear in relation to Mullins is that in the past two years he's found things more competitive, not just at the Festival but generally, a number of his top rankings being largely accumulated between the 2015/16 and 2017/18 seasons.

 

Gordon Elliott

It is fair to say that nobody really knows what to expect of the Cullentra House yard, currently fronted by Denise Foster while Gordon Elliott serves out his suspension. What we do know is that flagbearers like Envoi Allen have been moved to other yards and that has to have a negative bearing on overall figures this time around. To frame this year's expectation, we need to look backwards.

 

Elliott's Festival record, in terms of scale and punter-friendliness, has been unrivalled. Apart from a big blip in 2019 - same as Mullins, interesting? - his performance has been off the chart by almost any measure.

Using the ranking approach gives us the following.

 

Here there appears to be quite strong correlation between Elliott's four-month form and his Festival form.

 

Nicky Henderson

The Irish haven't (quite) had it all their own way in the past five years at Cheltenham, and Britain's top man - sometimes persisting in the wind - has been Nicky Henderson.

 

Four-win hauls in the last two Festivals help to explain the mini-lull, relatively, in the fortunes of Messrs. Elliott and Mullins, and represent a welcome return for Seven Barrows, in the shadow of the Irish challengers for the prior few years.

 

Paul Nicholls

It has been slim pickings for the former multiple Champion Trainer who has failed to record more than a pair of victories in the last four Festivals. However, in the 2019 and 2020 renewals, Nicholls' three wins were all at Grade 1 level: quality over quantity he might say.

 

Looking at the rankings shows a loose, perhaps tenuous, link between four-month form and Festival form in recent years.

 

2021 Pre-Festival Form and Predictions

At this point, you'd be forgiven for thinking "so what?". So let's try to review recent form approaching this year's Festival in the context of previous years.

In the table below, I've included four-week form up to 7th March (not quite up to Festival Eve, obvs, as I'm writing this a week earlier); and performance in the four months prior to that.

Specifically, from 8th February to 7th March, and from 8th October 2020 to 7th February 2021.

Willie Mullins has the same 75% PRB figure as he did in 2018: that year he won seven races from 62 runners. His four-month form is also the best it's been since 2018. And yet, Cheltenham Festival 2020 was arguably Mullins' best in recent seasons in spite of coming into it off the back of his second-worst recent form of the past five.

Elliott's team, meanwhile, has been in top form despite the challenging circumstances. Who knows what impact the loss of key horses and the absence of the hitherto licence holder (and the new named holder) will have? Likely some, but probably not a huge amount is my best guess. Elliott has had three phenomenal CheltFests in the past four years, 2019 being a sharp reminder of the perils of blind backing a yard; and he's had at least three winners in each of those years - 27 in all during that time.

Paul Nicholls has enjoyed enjoyed a relative resurgence in the last two renewals courtesy of that hat-trick of Grade 1 scores. He comes to Cheltenham Festival 2021 in similar form to 2019.

Most interesting is probably Nicky Henderson, whose form this season is notably lower than in each of the previous four seasons (current season four-month PRB of 0.57 vs ultra-consistent 0.65 in 2016/17, 2017/18, 2018/19 and 2019/20). In the past four weeks he's been pitching at 58% PRB which again compares unfavourably with far higher PRB figures for the run-up to the past couple of Festivals. It will be fascinating to see how Team Henderson fares next week, with Shishkin an early barometer.

 

Cheltenham Festival Trainer Form Conclusions

Sometimes you can spend a lot of time looking for something which, in the end, only tells you that there is probably nothing to be found. This may be one such occasion. Indeed, the writing was on the wall when I mulled the research overnight, came back to my computer this morning and discovered Windows had decided to undertake a forced update and, further, had corrupted the open - and, naturally, unsaved - spreadsheet document containing all of the data. Ugh.

Ignoring my computer woes, what can be seen from the above is that there is little to no strong correlation between various preceding periodicities and the meeting itself; and sometimes it is useful simply to know that. Of course, that won't stop a swathe of "he's in form" or "she's out of form" observations casually lobbed into the chat next week, the kind of positive/negative reinforcement, generally delivered after the fact, that adds now't but sounds knowledgeable.

It may ultimately be the case that the best gauge of Festival trainer form is from previous Festivals. In that regard, we should expect each of Mullins, Elliott/Foster, Henderson and Nicholls to hit their mark, and at least three of them to do so multiple times.

At the last five Festivals, they have collectively bagged 82 of the 140 races. Four trainers responsible for 59% of the winners. Throw in Henry de Bromhead - whose team is bolstered by the high profile addition of the Cheveley Park bluebloods - and Dan Skelton and you have six handlers responsible for two-thirds of the Festival winners in the last five years. Between them, they'll be long odds-on to take at least half of the 28 prizes on offer next week.