Stage 9 of the Tour de France and the end of the one of the most unpredictable days of racing in recent times. Perhaps, especially with hindsight, Dan Martin winning a Pyrenean stage is predictable but Chris Froome isolated whilst his team goes into meltdown with Richie Porte losing minutes and Vasil Kiriyenka missing the time cut? Who predicted that?
Race previews have a predictive element. Some aspects are known, the route of a stage of the Tour de France has been written down for months but other elements are less certain, like the weather. The greatest uncertainty comes with the race itself and picking the winner.
During the Tour de France I tipped some riders to win before the race and each day’s preview contained some likely stage winners. It turns out some readers were using the info to place bets and I got messages of thanks but also a few angry emails about lost money. Given real money is being spent, not to mention the anxious credit of fantasy cycling, I wanted to explore the idea and merit of forecasting. Forecasts are for fools.
“Prediction is very difficult, especially if it’s about the future.”
-Nils Bohr, winner Nobel Prize for Physics in 1922
Nothing is certain. You can analyse the route of a race but this can change. See Milan-Sanremo with the weather or the Orica-Greenedge bus incident on Stage 1 of the Tour de France where the finish line was switched and switched back. In other words even the things we take for granted are not certain. But we have to make assumptions and these elements are the “known knowns” to borrow from Donald Rumsfeld and help us explore the more unknown elements, for example the weather, a road might be fixed but the way it is ridden will vary according to the weather, a headwind on the long, wide section of the new Tour of Flanders finishing circuit discourages breakaways and therefore changes the scenario.
But predicting the weather is not easy. There’s a whole field of science dedicated to this – meteorology – and the experts get it wrong. During the Giro ilmeteo.it offered a dedicated forecast for the following day’s stage. It seemed to be wrong every day, when it said a dry start would be followed by rain for the finish the reverse happened and vice versa. As the race went on I used several websites to make an aggregate forecast. The same was true for the Tour de France where the official bulletin seemed to promise atmospheric apocalypse on the Alpe d’Huez stage but there was just light drizzle for a few minutes.
There are two lessons from this. First we struggle to predict the weather for the next day, all the supercomputing power deployed does not guarantee accurate forecasts. Second aggregating weather forecasts can help and we can deploy this technique for other forecasts. Put simply using different sources allows a wider view, a point essential if your after race predictions rather than the weather. For example see the forecasts on here but go and check Mikkel Condé’s race previews at C-Cycling and other sources.
Follow the Money
Another way is to see what the market is saying. Rather than relying on a few bloggers, see what the expert bookmakers say and where the money is going. Websites like oddschecker.com are very useful for comparing prices across the betting market. Bookmakers try to price the likelihood of an event occurring so this is a useful numerical source.
But remember price is not always equal to probability, especially in cycling. It’s not a sport that attracts floods of money and therefore the betting market can be illiquid. Prices can also reflect the amount of money going in one direction. For example if Chris Froome was the obvious choice to win the Tour de France then this can attract more money which then drives down the odds, creating a self-fulfilling event.
The imprecision of the bookmakers means there can opportunities for betting, to spot mispriced odds. For example Rui Costa was 100-1 to win on the day of his first stage win in July, very generous offer for a rider well down on GC but with form from winning the Tour of Switzerland. But I suspect you’d have to have some kind of fund to punt on a range of underpriced events and hope that if you lost pennies on most bets you’d land a big win once in a while.
How accurate were the picks? For the Tour my podium picks were Chris Froome, Richie Porte and Nairo Quintana, after discounting Alberto Contador because he’s still rated by many for his past wins rather than his actual performance. This worked out ok but for the Giro I saw Bradley Wiggins and Vincenzo Nibali on the podium with Cadel Evans and Ryder Hesjedal fighting for the third spot. Wiggins and Hesjedal both flopped.
It’s one reason why previews on here list many of the contenders for the day. Rather than pick a single winner I prefer to list the chances of several, even many, riders in a preview. It’s more cautious but also helps ID more riders, for example Damien Gaudin as the prologue pick for Paris-Nice and personally it’s more fun to weigh things up rather than cast out a name.
Some sports have binary outcomes, for example tennis is either win or lose. Football can see teams draw. But cycling can have 200 riders each with a chance of winning, of course some have better odds than others. But the point is that this creates a lot of uncertainty. This is to be embraced, it’s what makes the racing fun to watch, an early breakaway on a mountain stage might allow non-climbers to build up a lead, creating a wider contest.
Even during a race events are wonderfully unpredictable. Watch Stage 18 to Alpe d’Huez and Christophe Riblon’s odds must have fluctuated wildly during the day from total outsider to breakaway member to crash victim and then looking a lesser rider than Tejay van Garderen until clawing his way back and passing the American with the confidence of a stage victor.
In fact the more random the outcome the better the race. We can only give thanks that Eddy Merckx was at his peak 40 years ago because if he was racing today the fun could be sucked out of a lot of racing. It’s probably also why some want race radios and power meters banned, moves to make the racing more unpredictable.
I can’t help noticing the way we long for certainty and hang on predictions for a range of things. There were complaints when I didn’t include names in race previews in the past, even if the forecasts prove wrong it’s still fun to weigh up the contenders and discuss their chances.
It can be hard to get next weekend’s weather forecast right but this doesn’t stop people making long range predictions for things like stockmarket prices, economic data or population numbers even if these things often, if not always prove wrong.
“Wall Street indices predicted nine out of the last five recessions”
-Paul A. Samuelson, Newsweek, 1966
We’re suckers for a glimpse into the future and the more certain the forecast, the more appealing even if rational thought suggests such certainty is really dogma and therefore more likely to be wrong.
Sometimes even the weather forecast goes wrong for a stage so picking the winner of a race is a much harder task. Even the bookmakers, backed by the crowd and the market forces often get it wrong.
But if forecasts are for fools, they’re still fun and whether it’s sport, finance or anything else, we’re always trying to look ahead. In race previews, it’s a form of story-telling, a way to establish the characters who will feature in a race. This is what makes watching a race on TV for hours so compelling as you can see the story unfold, each turn in the road is a new page, every climb a new chapter and with time the narrative of how the race was won develops. Just don’t bet your shirt using the predictions on here.