No Evidence of Lleyton Hewitt Fixing Matches

Match fixing has been the talk of the tennis world in the past couple of days, ever since the joint Buzzfeed and BBC reports were published late on Sunday night on the eve of the Australian Open. Buzzfeed also made their data available to the public, albeit with the player names replaced by 64 digit codes. Despite this, it took me less than 6 hours to decode the 15 players that they claim "regularly lost matches in which heavily lopsided betting appeared to substantially shift the odds – a red flag for possible match-fixing."

I had been debating whether to publish this list, not wanting to do so without doing further research. I did not want to put player names out there if there were valid explanations for their matches - to me, just publishing the names without context seems rather irresponsible. However, it was not surprising when the full list was revealed this morning by @show_legend. So, now that the list is out there, here are the names:

Now, the most striking name on the list, and the one that has gained the most attention, is undoubtedly #6. Lleyton Hewitt is a former world number one and a US Open and Wimbledon champion, so any suggestion that he had been fixing matches would have been a huge story. The Buzzfeed algorithm flagged up 15 matches involved Lleyton Hewitt where there had been a significant odds movement, 13 of which Lleyton Hewitt went on to lose.

The important thing to remember is that pre-match movements in odds can be caused by any number of factors, one of which is fixing, but others also include injury concerns, players returning after a long absence or even just poor initial pricing by the bookmakers themselves.

I have randomly picked out eight of the fifteen matches that were flagged up to look at in more detail to see whether there are alternative explanations. I also enlisted the help of a professional tennis analyst, who provided predicted probabilities for some of the matches involved using a tennis model that he has created.

1. Lleyton Hewitt v Stan Wawrinka (Davis Cup 2011) - Buzzfeed Bookmaker: "B"

Opening Probability of Wawrinka winning - 43.7%
Closing Probability of Wawrinka winning - 85.5%

Quite what is going on with the Bwin odds here is rather baffling. As we can see, their prices are completely different to any of the other major bookmakers, all of which actually saw the price on Hewitt fall, suggesting that he was more likely to win the match. Really, to suggest that this match is any evidence of Hewitt fixing is bizarre and looks more like either a mistake on Oddsportal's part or on Bwin's part.

Conclusion: No evidence of anything suspicious

2. Lleyton Hewitt v Blaz Kavcic (French Open 2012) - Buzzfeed Bookmaker: "A" & "C"

Opening Probability of Kavcic winning - 48.9%
Closing Probability of Kavcic winning - 59.5%

This match at Roland Garros in 2012 was the first time that Lleyton Hewitt had been seen since the Davis Cup tie in early February, having undergone surgery to have two screws and a metal plate put into his foot in an attempt to prolong his career. Hewitt had even admitted that he had only been back on the practice courts for two weeks ahead of the French Open, so it is a match that would have been incredibly difficult for bookmakers to price up. The reality is that matches involving players coming back from injury are almost impossible to price correctly, so the fact that there is a big price move is not surprising in the slightest.

Conclusion: No evidence of anything suspicious

3. Lleyton Hewitt v Stan Wawrinka (Indian Wells 2013) - Buzzfeed Bookmaker: "D"

Opening Probability of Wawrinka winning - 67.6%
Closing Probability of Wawrinka winning - 80.0%

Another match against Stan Wawrinka shows up in this sample. This time, it followed an upset win by Lleyton Hewitt in three sets over the 15th seed, John Isner. It was a second consecutive three set victory for Hewitt in the tournament and there are possibly suggestions that he might have been fatigued or overrated after the upset win.

However, the reality is that the opening price for this bookmaker was the highest of all the six bookmakers in the sample (one bookmaker opened with a 73.5% chance for Wawrinka), while the closing price was the lowest of all the bookmakers in the sample (one closed as 75.2% chance on Wawrinka). While all the bookmakers saw a move toward Wawrinka, the others were all comfortably below the 10% move that was used as a red-flag by Buzzfeed, so likely that the bookmaker in question opened too big on Wawrinka, saw plenty of money and were forced to go short on the Swiss player to balance their book.

The calculated probability by the professional tennis analyst for this match was actually 72.5%, so right in the middle of the range of bookmaker prices, so there was nothing all that suspicious about the actual price itself.

Conclusion: No evidence of anything suspicious

4. Lleyton Hewitt v Sam Querrey (Houston 2014) - Buzzfeed Bookmaker: "A"

Opening Probability of Querrey winning - 45.5%
Closing Probability of Querrey winning - 58.1%

This match on the clay in Houston followed on from Hewitt's first ATP clay court win in almost four years in the previous round, having defeated #135 ranked Peter Polansky. Similarly to the second Wawrinka match, we find that the opening price is relatively big compared to most of the other bookmakers in the sample, while their closing price is the lowest in the sample.

The calculated probability from our tennis analyst for this match was 56.5% for Sam Querrey, so this simply looks like a match that was incorrectly priced initially and was simply correcting to a more appropriate price.

Conclusion: No evidence of anything suspicious

5. Lleyton Hewitt v Jerzy Janowicz (Wimbledon 2014) - Buzzfeed Bookmaker: "B" & "G"

Opening Probability of Janowicz winning - 46.5%
Closing Probability of Janowicz winning - 68.0%

This match had me concerned for a while. It appears as though the majority of bookmakers suspended betting on the match, while a selection of others saw a huge price move, all the way from 2.15 on Janowicz into 1.47. There appeared to be no evidence of any form of injury or anything to explain the move.

However, after digging into the Betfair history for this match, the solution revealed itself. The match was actually suspended due to rain with Janowicz leading the match at 7-5, 4-4 up and the match was not completed until the following day. As a result, a number of bookmakers re-opened their pre-match markets with revised prices on this match, which is what Oddsportal and the Buzzfeed data has picked up on. So, the price that is being compared with the opening price here is actually the price on Janowicz at a set up and on serve in the second.

Conclusion: No evidence of anything suspicious

6. Lleyton Hewitt v Benjamin Becker (Australian Open 2015) - Buzzfeed Bookmaker: "A"

Opening Probability of Becker winning - 45.5%
Closing Probability of Becker winning - 58.1%

Another example of a match where a bookmaker has initially gone up with a price that is highest in the sample and followed it up by cutting the price to one that is lowest in the sample, presumably after seeing money on Becker.

The predicted probability from the tennis analyst for this match was 53.2%, which is precisely the closing price of one of the bookmakers in the sample and very close to three others, which simply suggests that it was a market correcting itself from an incorrect initial price.

Conclusion: No evidence of anything suspicious

7. Lleyton Hewitt v Thomaz Bellucci (Miami 2015) - Buzzfeed Bookmaker: "C"

Opening Probability of Bellucci winning - 42.2%
Closing Probability of Bellucci winning - 55.9%

This was the first match that Lleyton Hewitt had played since the Australian Open, so it was naturally difficult to price. Hewitt had dominated Becker in the opening two sets before falling away physically, so over the best-of-three format, there was suggestions that he might be able to be competitive against the horribly out of form Brazilian. As it happened, the match was a horrible match to watch with the two players combining for over 120 unforced errors, but with so few matches in the run-up to this event, it is almost impossible to price it correctly to begin with.

Conclusion: No evidence of anything suspicious

8. Lleyton Hewitt v Kevin Anderson (Queens 2015) - Buzzfeed Bookmaker: "C"

Opening Probability of Anderson winning - 69.9%
Closing Probability of Anderson winning - 80.6%

This was only Lleyton Hewitt's second match in over two months, so bookmakers had little apart from previous reputation and old matches to really go on in terms of pricing this match. With Hewitt playing so sporadically in 2015, it was incredibly difficult to really grasp what level his game was at, so on his favourite surface, it could be seen as natural to lean toward giving him a chance.

The predicted probability from the tennis analyst, albeit with a large margin for error due to small sample sizes for Hewitt, was around 75.2%, so right in the range of the actual price move.

Conclusion: No evidence of anything suspicious

Final Thoughts

As we can see, of the eight matches that I have looked at, four can easily be explained by long absences before the match resulting in matches that were incredibly difficult to price with little information on Hewitt's level or condition. Two can be explained away by incorrect or questionable data, while the rest simply appear to be the price correcting itself after poor initial pricing by one individual bookmaker.

While the analysis that Buzzfeed's algorithm carried out is a good way to generate a list of players and matches to investigate in further detail, the suggestion that a player appearing on the list is a major red-flag against his name is drawing conclusions that simply cannot be drawn.

From the matches that I have looked at, there is absolutely no suggestion whatsoever that Lleyton Hewitt has been involved in fixing matches and to simply post his name among a list of players suspected of fixing without performing any further analysis or providing context is simply irresponsible. I could continue to look at more of the matches, but my suspicion is that there would be similar circumstances for the rest.

This is not to say that other names on the list may not be fixing matches, but as I have shown, without performing further analysis, that conclusion can simply not be made.