Toolbox: A breakaway in a cycling race is a hotbed of intrigue, with ‘friends’ working together for the common good while also plotting their betrayal of the group as a whole. What are the factors that actually enhance the odds that the early breakaway takes it all the way to the line?
Van Vleuten tore up not just the World’s, but also all economic models.
Efforts like Annemiek van Vleuten at Yorkshire World’s is an extreme outlier, with a solo long-range breakaway from 105km out succeeding. Outside of that, most professional men’s WorldTour racing follows a relatively predictable pattern, with an early breakaway establishing itself. There then follows a prolonged period leading up to the finish, where the break may or may not become caught.
Sport is often said to be a reflection of society, and parents often say sports teaches kids valuable life skills like cooperation, perseverance, etc. But for economists, sport is also a valuable model for social dilemmas and real-life questions like cooperation, because it offers a fairly self-contained and transparent situation.
To that end, a paper in the Journal of Economic Psychology modelled the economic strategies buried within the cycling break away to help answer larger questions about private versus public good and cooperation (Brouwer and Potters 2018).
The ‘Break of the Day’
Within the break, there are private incentives at play versus the common good. Riders need each other’s strength in order to keep the peloton at bay. Thus they are friends for a while. However, even during this period of cooperation, games are afoot. While riders need to expend energy to keep the pack at bay, spending too much energy individually decreases your own chance for eventually winning against your break partners.
One unique aspect of bike racing that makes it not necessarily represent society at large is that cycling is largely a winner-takes-all sport, where the winner gets all the spoils and others are largely forgotten and considered to get zero benefits. This was the assumption made by the scientists, even though stage races can have all sorts of secondary benefits for the GC, race categorizations, etc.
The simplest thing to do to remove this confound would be to ignore stage racing and study only one-day races. However, that would remove a vast dataset and leave relatively little to study. Also, the odds for a breakaway succeeding in one-day races is almost nil, with the only exception in the studied 2011-2017 dataset being Mathew Hayman’s 2016 Roubaix win from the early break.
Indeed, Brouwer and Potters went the other way and ONLY analyzed stage races.
Solo win in Liège-Bastogne-Liège from Remco Evenepoel
Brouwer and Potters 2018
The two economists from Tillburg University in the Netherlands (trust the bike-mad Dutch to throw bike racing under the economics microscope!) did a ton of complex mathematical modelling that I won’t reproduce. However, some key features and assumptions in the model include mathematically adjusting for peloton size and relative quality, stage type and difficulty, weather, and a whole host of other variables that may affect race dynamics.
The bulk of World Tour racing from 2011-2017 was analyzed, with about 780 actual races in the dataset.
Them’s the Breaks
Somewhat surprisingly, the authors’ empirical analysis of actual race results showed a higher percentage of successful breaks than what most of us would’ve guessed, with 23% of stages being won by a rider from the early break. This definitely goes against the common perception of the “doomed early break.”
As most seasoned race fans know, a host of factors help shape the possibility for a break’s ultimate success. These largely match up against the theoretical model and also the empirical data, and include:
The size of the break. Too few and the work becomes too hard. Too many and there’s bound to be a bunch of free-loaders destroying cooperation. The model showed that increasing break size with a rider of average ability raised the odds of success by about 3.1% up until about 8 riders, after which each additional break member increases success odds relatively less and less, and then ultimately the odds become negative with additional members beyond 26 riders.
The strength of riders in the break. Often, weaker teams send their riders into breaks not expecting any chance of race success, but rather for publicity. However, when serial break magnets like Thomas de Gendt or Thomas Voeckler get into a break, magically the odds for break success seems to increase exponentially. Increasing the relative strength of the break by one ‘unit’ (I’ll avoid the math) increased the odds of success by 5.8%. However, regardless of individual rider strength, the same pattern holds where increasing break size eventually outstrips the benefits of stronger riders.
Riders like De Gendt joining the party is like adding rocket fuel to the breakaway.
Stage profile. Flat stages almost never see break success, as hungry sprinter teams all work to lock down and reel in the break. The classic break stage seems to be the medium mountain or transition stages in the middle or later parts of a stage race. This was found in the models, with hilly or mountainous stages more likely to see a successful break.
As with any mathematical model, the assumptions used to build it can be argued ad infinitum. However, I felt that this study brings an interesting econometric approach to understanding what makes for a successful break. Hopefully it helps you to gain insight into your 2020 bike race viewing!
Jens Reynders’ failed solo in Roubaix, but due to a puncture
Ride fast and have fun!
Brouwer T, Potters J (2018) Friends for (almost) a day: Studying breakaways in cycling races. J Econ Psychol S0167487017307596. https://doi.org/10.1016/j.joep.2018.08.001