So many doubts remain around the form of the top players and the effect of a lopsided draw, writes Jack Houghton, that a waiting game is wisest
"But then Nadal is a special case in the French Open. After all, he's aiming to win his 13th title. How do we account for that in his odds?"
The struggle for profitability
Whenever I have thought about this year's French Open I get wrenched back to a time of self-doubt, a period when I wrestled to become profitable.
Most punters - like my younger self - lose money in the long run. Some deny this, often subconsciously constructing a version of their punting history that has them remembering winning bets, whilst handily ignoring the losing ones. Others accept their loss making, happy that the costs are sustainable, and seeing them as the price of a hobby they enjoy.
Among these losing punters is an interesting subspecies: those that recognise their habit of deficit and make a concerted effort to change it. To make this shift successfully requires the punter to tackle a range of practical and psychological challenges.
This is not the time for an examination of all those challenges, but two of them are illustrative of why this year's French Open is so difficult to assess as a betting proposition.
Challenge 1: comparing competitors
In technical terms, the first challenge for most would-be profiteers is finding a valid and reliable method for comparing the ability and form of participants. For most tennis bettors, this means creating a rating system that assigns a value to each player's performance which sees their rating move up and down relative to other players as they win and lose matches.
There are several approaches to ratings - the most common being Elo and Glicko - with varying degrees of complexity.
Building a ratings system from scratch takes time and is usually iterative, requiring several rebuilds as practical and statistical limitations present themselves. When I first created my tennis Elo rating system, for example, weeks of loving work was undone when a test of the ratings on historical data found that the numbers were pure fantasy. Not only that, but the way I'd constructed my spreadsheets meant there was no effortless way to change the assumptions that underpinned the ratings.
More disappointment followed when, having addressed these issues, it became apparent that the ratings themselves were useless if I couldn't convert them into betting odds.
Challenge 2: converting ratings to odds
At the time, there was little information available publicly about how to address this issue. With lots of work and false starts, though, I figured out a way to convert my Elo ratings for match betting.
However, it took several more weeks and a fortuitous stumble across an article about modelling in physics to recognise that Monte Carlo simulations would also allow me to work out the odds for an entire tennis tournament.
At the time, running these simulations on a home computer meant buying a bespoke programme. Now, there are workarounds on most spreadsheet packages that are just as efficient.
What's all this got to do with the French Open?
The problem with this year's French Open is that all that work on ratings and simulations seems somehow useless in working out how to bet on it.
The first issue is that my Elo ratings are hard to believe.
Any robust statistical analysis should include an assessment of confidence. In other words, after the raw number-crunching, you must ask yourself whether the results are worth believing.
A mistake made by many punters embarking on a road to profitability is that they become so invested in the work they have done to create the ratings, that they then value them far more highly than they should. Psychologists call this the Ikea effect. A related psychological phenomenon - the McNamara fallacy - is also an issue: failing to question whether the data you have is the right data.
It would be easy to roll into this year's French Open with the same blind faith that I would usually have in my Elo ratings but, if I'm honest, I'm not sure how useful they are at the moment.
The issue is how many other factors - that my ratings fail to consider - are likely to influence its outcome.
Rafa Nadal, for example, currently the market favourite at 5/42.28 should, according to my ratings, be the second favourite at around 2/13.00. In some respects, here, the ratings are doing their job. Nadal has played only three matches since February, looking ordinary when losing in Rome on clay, and yet again struggling with injury. Can we really make a player favourite who hasn't shown anywhere near his best form for over 14 months?
But then Nadal is a special case in the French Open. After all, he's aiming to win his 13th title. How do we account for that in his odds?
There are other challenges, too. Roland Garros now has a retractable roof and will use a new type of ball this year. How will these factors affect Nadal's tendency to dominate?
My Elo ratings may suggest Nadal has around a 34% chance of winning in Paris, but if I'm honest, I'd struggle to argue against anyone who felt his chance was more accurately as low as 15% or as high as 60%.
And whilst Nadal's rating is an extreme example, a similar doubt underpins the ratings of all players, affected as they have been by the uncertainties of the cessation of competitive tennis.
Monte Carlo is bust by draw
It's not just Elo uncertainties that make this French Open a difficult puzzle, though. My Monte Carlo tournament simulation hasn't been helped by a somewhat perverse draw either.
Take Dominic Thiem, for example. After his heroic US Open win, his Elo rating is now at a near career-high, seeing him closer to Djokovic, Nadal, and Federer than any player has been since Andy Murray's peak.
Unfortunately for Thiem, he's drawn Marin Cilic in the first round. Thiem is the likely winner of that match - I make him a 1/41.25-shot - but it's not the usual 1/201.05 favouritism that a player of Thiem's standing might usually expect in the first round.
This means that the Monte Carlo simulation I'm using is unusually skewed. I currently make Thiem a 13/114.00-shot to take the overall title, whereas the market views his chance much more favourably (at around 11/26.40).
It feels as if the market is valuing Thiem as if he has already successfully made the second round. If he does, my own tournament odds will certainly readjust dramatically. For now, though, the anomaly is difficult to judge, and given the doubts I already have about the reliability of my ratings, I wouldn't argue against the market view with any confidence.
The draw has also thrown up difficulties for Stan Wawrinka and Daniil Medvedev that are hard to compute. Wawrinka's opponent, Andy Murray, is another extreme case of Elo-unreliability, and Medvedev finds his section packed with beatable but tricky opponents in every round.
Bookmakers must play, punters don't have to
All this angst and self-doubt about the value of my ratings system and method of odds conversion recalls a lesson that I learnt in those days of transitioning to profitability. An ex-colleague at Betfair and one of the savviest punters I've met, Tony Calvin, would tell me repeatedly that whilst fixed-odds bookies had to price up every event, punters didn't have to bet in them, and that this was a punter's big advantage.
That advice is as sage now as it was then. As the early rounds in Paris resolve, perhaps my confidence will increase, but for now, I have so many doubts about my numbers and what they are predicting that the recommendation is not to bet.
New on Betting.Betfair - Betslip
You can now bet without leaving Betting.Betfair with our brand new on-site betslip for Exchange markets. You'll see the Exchange back and lay prices at the end of articles - simply login and place your bets as you would do on the main Exchange site