Burghley Horse Trials 2020 simulation: one in 100,000 possibilities *H&H Plus*

  • In the absence of a real Burghley, it’s fun to speculate on how this year’s leaderboard might have looked – even though such “events” can never be used to give prize money or influence Olympic selection. Diarmuid Byrne from EquiRatings plays out a simulated competition...

    Another “lost” competition. We have had longer to digest the disappointment that Burghley wasn’t going ahead than we had for Badminton, but despite this and the fact we have some real sport starting up again now, we are still missing the big fixtures of this season.

    So while we wait, we at EquiRatings headquarters have dreamed up another “what if” scenario based on the past performance data of our top horses and the prediction technology provided to us by SAP.

    Burghley is a venue where speed always comes into play. To decide who should take part in this simulation, we took a look at our database to find horses who had made the time at five-star level at least once in the past two seasons, which gave us 52 pairs. We included the 2018 World Equestrian Games as a five-star to ensure a nice mix of nationalities.

    Now the fun starts – we place these 52 combinations into the same competition and we tell the machine (SAP Predictive Analytics) what variables to take into account.

    The first and most important will be horse and rider performance over a five-year period.

    Understanding this allows our analysts to develop a model for assessing win-place predictions – how impactful a good dressage score will be at different levels, how different horses suit different events – and the margin of error during simulation.

    One of the most important factors to remember is that the actual results of riders and horses contain what many would describe as “impossible” outcomes.

    Our sport is subject to huge degrees of chance and does not lend itself well to people who say “that will” or “that can’t” happen. Even the top horses in the world only have about an 80% chance of a clear cross-country round. The average at CCI5* is one in every two jump clear across country and one in every four in the showjumping. With that in mind, let’s press the button and see how it plays out.

    Phase one: dressage

    The major surprise from the dressage simulation was not the leader. We are well used to seeing SAP Hale Bob OLD deliver sub-20 dressage tests at this stage.

    If anything, possibly apart from Kazuma Tomoto (second at this stage), I would say most of the big names would have been expecting to deliver a lower score. Chris Burton broke the then-CCI5* dressage record here on Nobilis 18 in 2016 (20.5). Allstar B delivered the best-ever test of Ros Canter’s career at Burgham and we know from Badminton 2019 that Oliver Townend is among the best in the world at delivering a dressage test on the big day.

    When we look at the simulation after this phase, it looks like a competition which has seen higher scores across the board, with the exception of Ingrid. Maybe the cross-country will sort them out; the time will be tight.

    Phase two: cross-country

    Nobody made the time across country in wet conditions – what an exciting day it must have been. The fastest round was from Classic Moet, who added four penalties and climbed from 15th after dressage to second. We don’t see Ingrid at Burghley that often, but 10 seconds over the time, the second fastest round of the day, has left her with a commanding lead going into the final day.

    I’m looking at this leaderboard and imagining what it would feel like to be at Burghley on Saturday night.

    It is possible, if not that likely, that SAP Hale Bob OLD could have three poles to lose the title – the big story is whether Ingrid can hang on for first place. Andrew Nicholson and Oliver would both fancy their chances of a clear jumping round and Jonelle has delivered before to win a CCI5*, at Badminton 2018, when we doubted Classic Moet would jump clear but she did.

    So many stories could unfold from here…

    Phase three: showjumping

    Ingrid led from the start and her rivals needed more than her one pole down to overtake. Ingrid Klimke has won Burghley! What a shame for her that all the action took place on an overheating laptop in a bedroom/study in Dublin, rather than in front of the crowds in Lincolnshire…

    Time-penalties cost Jonelle second place and Mr Bass and Vanir Kamira made it into the top five. Looking through the field, it feels like the showjumping was easier than we have seen at the top level in recent years.

    The limits of simulations

    Simulation as a method of sports analytics is widely used by various professionals across all of the major sports. In this case, we have run the simulation once, but to get the best picture for an Olympic manager (or for TV content and planning cameras), the simulation is run 100,000 times and then ranked so that we can identify what would happen if we were to put this field together, in these conditions, at this event, 100,000 times.

    When EquiRatings talk about an event “favourite”, that is how we do it. Running a simulation once doesn’t do much other than allow us to have a little fun and imagine we were there to watch the drama unfold.

    The ‘one in 100,000’ outcome

    H&H eventing editor Pippa Roome shares her thoughts…

    This feels like a “real” Burghley leaderboard, with no one making the time and plenty with time-penalties in double figures making the top 20.

    The only surprise about Ingrid’s win would be seeing her at Burghley at all with SAP Hale Bob OLD – having been such a sought-after German team horse, attending every championship since 2015, he has never had the chance to run at the British autumn classic. Ballaghmor Class and Classic Moet, in second and third, are proven Burghley horses in real life.

    As Diarmuid says, the simulation can turn up the “one in 100,000” outcome, which is why we mustn’t take it too seriously. Vinci De La Vigne JRA has never had a cross-country stop in an FEI competition and yet he does here; Cooley Master Class and Nobilis 18 both have very good cross-country records but also lose their top 10 positions after dressage in this simulation with cross-country jumping penalties.

    Similarly, neither Swallow Springs nor Reve Du Rouet have had 12 jumping penalties in the final phase since 2017, so I wouldn’t expect this outcome. Indeed, if Reve Du Rouet’s strong performance at Burgham is anything to go by, it looks like the best is yet to come from this 16-year-old.

    Ref Horse & Hound; 3 September 2020

    You may also be interested in…