Federer has lost a tiebreak in 12 Grand Slam semifinal/final matches.
Remarkably, he has won only 1 of those matches.
(2006 Wimbledon Final)
Well, he used to win nearly every tiebreaker in those circumstances, that's why.
Poor stat interpretation.
Agree with
@Gazelle there's nothing to see here. The real picture emerges if you look at these data:
When only losing tiebreaks: 0-8
When only winning tiebreaks: 22-4
When both winning AND losing tiebreaks: 1-3
When NOT playing tiebreaks: 23-8 (74%)
When playing ANY tiebreaks: 23-15 (61%)
Implications:
(1) his record when he ONLY WINS tiebreaks is statistically indistinguishable from his record when he ONLY LOSES tiebreaks. (22-4 vs 0-8 are essentially the same given the sample sizes). This is reasonable because it's REALLY hard to win a major final or semi if you spot your opponent 1 or more sets.
(2) When both winning and losing it's only a match away from 50-50. It's like getting 3 heads out of 4 coin tosses. Totally normal.
(3) Finally, when he doesn't play tiebreaks his record is better than when he plays ANY tiebreaks. This also makes sense because Federer is, on average, better than his opponents. And the matches with tiebreaks represent a reversion from 74% towards 50%. Which is exactly what you'd expect to see, because tie breaks are more random than winning by a break.
