I'm looking forward to checking out that documentary -- it's not available outside of the UK (as far as I know), but I'll get my hands on it somehow.

I don't think i've looked specifically at the first point of the tiebreak. The important distinction to make is that winning the first point definitely *does* increase your odds of winning the tiebreak, for the very boring reason that winning a point gets you closer to 7 than the other guy. What commentators (or whoever makes this claim) usually mean, though, is something else. Say two players are evenly matched, each one has a 50% chance of winning the tiebreak. If the server wins the first point, say his chance goes up to 55%, just based on math, because 1-0 is better than 0-1. When someone says that the first point really matters, they mean that the player's *actual* probability is greater than 55%, because of momentum, or confidence, or whatever.

So, in essence, it makes no difference really - winning the first point of a tie breaker other than the obvious point that winning the first point is advantageous because youve won the first point and therefore are ahead.

It suggests that, if Jeff had run the stats on this, the person winning the first point of a tie breaker would have statistically better chances of winning the tie break

But his last comment in the article is interesting: maybe momentum is part of the answer here and how you go into the tiebreak is more important than which points you specifically win in it

Jeff wrote : "Theres nothing special about the first change of ends, and there probably isnt any other point in a tiebreak that is more crucial than the model suggests. Instead, weve discovered that underdogs have a slightly better chance of coming back than their paper probabilities indicate. I suspect were seeing the effect of front-runners getting tight and underdogs swinging more freelyan aspect of tenniss conventional wisdom that has much more to recommend itself than the idea of a magic score after the first six points of a tiebreak."

Thank you to Strongbow, JonH and Jeff for replies. Enjoyed reading Jeffs article - love a nerdy fact or two. So it seems more down to the mental side of tennis than tb points at certain points in the game.

Thanks Strongbow, yes very different to the first snapshot. Reflects the need for big data?
Have been following recent matches with interest (so far n of 1), George L won first point in the tb, lost the set and then the match which ties in with Jeffs comments.

Thanks Strongbow, yes very different to the first snapshot. Reflects the need for big data? Have been following recent matches with interest (so far n of 1), George L won first point in the tb, lost the set and then the match which ties in with Jeffs comments.

Yes, interesting as a subject. But these ( and still the combined one ) are far too small samples to conclude much about anything.

Proper 'big data' would be interesting.

The momentum question Jeff raises is fascinating, and whether the particular nature of the normally pretty. crucial TB would bring different dynamics to what he has seen in his normal game 15-0 studies ( already different anyway of course in that the TB has alternating aerves ).

Does winning that first point give such momentum so as to be more likely to win the TB than just the fact that the player is 1-0 up? Or maybe the other player becomes more determined and it is actually less likely than might be expected?

The more I think about it all though, the more yes the need for really big data, such are the margins, to separate out the different factors.

Then could add in ( or maybe rather more interestingly, sepetate out ) - what if it's a 1-0 lead as the result of a mini break?

Jeff Sackmann has several times tried to analyse differences in surface speed, in the past he has used aces as a proxy for surface speed - ie grass will have more aces for any given player than , say, clay. More aces overall, but Jeff looks at it based on a given players expected ace count and how that varies on different surfaces.

Anyway, here is using rally length to see if surfaces have converged.

The evidence suggests that yes, they have, possibly but that there is probably more to it than that ie technology of racquets etc has caused a change in tennis playing style (more baseline less serve and volley) that has possibly also led to a convergence in rally length on different surfaces. SO surfaces are playing in a more homogenised fashion, suggesting surface speeds may have converged, but that this might not be the whole story to explain it - tactics due to technology is also part of it

Interesting read and Jeff has loads more general tennis stats for those interested in such stuff