12 (Or Is It 14?) Straight Tie-Breakers Lost by Haase

This development is a little old, but I only noticed it today. Jeff Sackmann wrote on April 12 at his "Heavy Topspin" tennis website that men's pro Robin Haase had lost 13 consecutive tie-breakers. According to Sackmann:

No other active player has a streak of more than seven, and no tour-level regular has lost more than his last six.  In fact, Haase is now one lost tiebreak away from tying the all-time ATP [Association of Tennis Professionals] record of 14, jointly held by Graham Stilwell and Colin Dibley, two players who accomplished their feats in the 1970s.

With a 1-6, 6-2, 6-7 loss to Pablo Carreno-Busta on April 22, Haase has, in fact, now technically tied the record of 14 straight lost tie-breakers. Why do I say "technically"? If one peruses the ATP World Tour website for Haase's results, starting with the most recent ones and working backwards in time, one counts "only" 12 lost tie-breaks before noticing a breaker that he won. This win by Haase occurred on May 13, 2012 at the ATP World Tour Masters in Rome, as he beat Sergiy Stakhovsky, 6-4, 7-6 (11-9). Prior to winning that tie-breaker, Haase had lost his two previous ones.

The reason Haase's streak is officially listed as being 14, Sackmann has confirmed, is that the tie-breaker Haase won over Stakhovsky occurred in a qualifying-round match (note the "Q1" on the ATP site), not the tournament draw proper. If the ATP's policy has always been that only main-draw matches count for record purposes, so be it. Perhaps the thinking is that, because the level of competition presumably is less demanding in the qualifying rounds than in the main draw, only the latter should be counted for record purposes. Let's be clear, though: Whether one considers Haase's tie-breaker losing streak to be 12 or 14, it's unusually long in either case!  

The reason I was looking up all of Haase's matches, in the first place, is that I was curious with regards to how many points he was garnering in the tie-breakers during his stretch of futility. Has he been getting blown out in them (e.g., 7-0, 7-1, 7-2)? That might suggest a major freeze-up. Or, on the other hand, has he been taking his opponents to the wire, losing the tie-breakers by the minimum two points (e.g., 7-5, 8-6, 9-7)? If the latter, Haase wouldn't be experiencing full implosion at the umpire's call of "Games are tied at 6-all," but one would still have to wonder why he consistently lost close tie-breakers. I created the following graph to show Haase's frequency of different point totals in tie-breakers during his losing streak.  

The picture is mixed, yielding what statisticians call a bimodal distribution (i.e., two values tied for being most frequent). Five times, Haase was blown out 7-2, whereas he lost another tie-break 7-3. However, he has also stayed within two points on five occasions, losing 7-5 (twice) or 8-6 (thrice). 

Sackmann has done extensive research on the likelihood of winning tie-breakers, which you can find by following the links in his various Heavy Topspin postings (e.g., here). Sackmann has developed formulas for how many tie-breaks a given player would be expected to win, given the player's success for entire matches at winning points on serve and receiving. Players with better-than-average skills should win more than 50% of tie-breakers, whereas those with less-stellar skills should win fewer. Sackmann finds that, "Aside from a small minority of players with extensive tiebreak experience (such as Roger Federer, John Isner, and Andy Roddick), ATP pros tend to win about as many breakers as 'expected.'"

Obviously, Haase has been an outlier on the unfavorable end. Interested readers can readily follow Haase's match-by-match results, to find out if/when his streak of lost tie-breakers ends. As noted above, the ATP website has Haase's results available. Sackmann also provides frequent Twitter updates on major tennis developments, including Haase's streak

Hot and Cold Starts to Current MLB Season

The 2013 Major League Baseball season is a couple of weeks old and we now see which teams have gotten off to the hottest starts: the Atlanta Braves (13-2), the Oakland A's (12-4), and the Boston Red Sox and Colorado Rockies (each 11-4). To some extent, one could say that there always seems to be a team or two starting off with a bang every year. Last year, for example, the Rangers started off 15-4, whereas the Dodgers began with a 12-3 burst.

However, if the Braves, A's, Red Sox, or Rockies can continue to mow down opponents, they may be able to join the pantheon of teams with all-time great starts. This 2011 Yahoo! Sports article lists "Five of the best starts in baseball history." Three of them might be considered fabulous long-term starts: 30-5 by the 1902 Pirates, 33-7 by the 1939 Yankees, and 35-5 by the 1984 Tigers.* The other two great starts were over a shorter term: 13-0 by the 1982 Braves and 17-1 by the 1987 Brewers.

Blogger Jim Mancari identifies the pillars of the Braves' strong start this year:

Justin Upton has already clubbed a league-leading eight home runs, and Evan Gattis — who is supposed to just be a placeholder for the injured Brian McCann — has shown impressive pop at the plate.

[Chipper] Jones’ replacement Chris Johnson is hitting over .400 to start the season — Chipper who? Freddie Freeman is also over the .400 mark to start the year.

It’s a small sample size, but the Braves have the capability to give the [Washington] Nationals fits all season... 

The starting pitching staff has been great so far. Through his first three starts, Paul Maholm has not given up a run — earned or unearned — in 20 1/3 innings. Tim Hudson, Kris Medlen and Mike Minor have also looked good early.
 
Where the Braves definitely have the Nationals beat is in the bullpen. Closer Craig Kimbrel is perfect in save opportunities and has only allowed two hits and two walks (no runs) in 6 2/3 innings of work.

For Oakland, a fast start seems almost out of character, as recent A's squads have tended to be at their best in the second half of the season. The A's had a hot July last year and also did very well in August during several seasons in the 2000s.

We also, of course, have seen some terrible starts to the 2013 season. Not surprising is that of the Miami Marlins (3-13), who undertook an off-season fire sale. More surprising is the start of the superstar-laden Angels (4-10).

---
*In fact, I have a chapter in a new book about the '84 Detroit squad.

Louisville's Hancock and Michigan's Albrecht Exhibit Hot Hands in First Half of Men's College Hoops Title Game

With 3:02 remaining in the first half of Monday night's NCAA championship game and Michigan leading 35-23, Cardinal coach Rick Pitino called a time-out. Hot shooting from behind the three-point arc -- including four straight shots without a miss by reserve guard Spike Albrecht -- had given the Wolverines their sizable lead at this point. However, in a two-minute stretch that will give Michigan fans nightmares for years to come, Louisville's Luke Hancock hit four straight treys of his own and, next thing you knew, the teams were heading to the locker room with the Wolverines ahead only 38-37 at intermission. So compelling was the first-half three-point duel, it inspired the New York Times to write a human interest story on Hancock and Albrecht.

Based on the play-by-play sheet of the Louisville-Michigan game, I have graphed all three-point attempts of the first half. The chart indicates which player (by last initial) took each shot (UM players depicted in blue and UL players in red), with how much time remaining in the half, and whether the shot was good (bright colors in top row) or a miss (lighter colors in bottom row). You may click on the graphic to enlarge it.

Though Albrecht and Hancock each hit four first-half threes, as can be seen, Hancock's occurred in much quicker succession. Also, until Hancock's shooting outburst, Michigan had made six treys to only one for Louisville.

Louisville dominated the second half except for some brief stretches, with Hancock hitting a fifth three-pointer to put the Cardinals up 76-66 with 3:27 left in the game. The final score was 82-76 for UL.

Hancock went 5-for-5 from downtown in the title game, as well as 3-for-5 on treys vs. Wichita State in the national semifinal, giving him an impressive 8-for-10 three-point shooting percentage in the Final Four (see Hancock's game-by-game log). Such a shooting performance by Hancock in the biggest games of the season would have seemed very unlikely, however, based on his start to the 2012-13 campaign. A transfer from George Mason University whose Louisville career began last fall, Hancock compiled an ugly 9-of-41 (.220) record on three-point attempts in his first eight games.

Though he later had some good three-point shooting games (e.g., 4-of-5 at Syracuse on March 2), Hancock entered the Final Four (including the regular season, Big East tournament, and first four games of the NCAA tournament) hitting only 55-for-148 (.372) from downtown.

One can calculate the probability of a prior .372 shooter hitting 8 (or more) of his next 10 three-point attempts, using what is known as a binomial probability calculator. This probability turns out to be .007, a little less than 1-in-100. Unlikely, but not astronomical.

What about Michigan's Albrecht? As his game-by-game log shows, Albrecht attempted far fewer three-pointers during the season than did Hancock. In fact, until Monday night's championship game, the frosh Albrecht had never attempted more than 2 three-point shots in a single game. For the regular season and Big 10 tournament combined, Albrecht made 9-of-23 (.391) behind the arc.

In the Wolverines' NCAA tournament opener vs. South Dakota State, Albrecht didn't attempt any treys. In each of the next three tournament games -- vs. Virginia Commonwealth, Kansas, and Florida -- Albrecht went 1-for-1 on three-pointers. In the national semifinal vs. Syracuse, Albrecht upped his three-point output to 2-for-2. He then, of course, hit his first 4 shots from beyond the arc in the title game against Louisville, making him 9-of-9 for the entire NCAA tournament at that point. With 11:28 left in the game, Albrecht missed a three-point shot that would have put UM, then trailing 52-54, back in the lead. The miss was the only mar on Albrecht's tournament three-point record, which ended up 9-for-10.

Returning to the binomial probability calculator, we ask in Albrecht's case what the probability is of a prior .391 shooter making 9 (or more) of his next 10 three-point shots. That probability is .001 or 1-in-1,000.

Clearly, two rare performances were on display Monday night!

Aftermath of Heat Streak

Bill Simmons offers a lengthy postmortem on the Miami Heat's 27-game winning streak, which ended last Wednesday in Chicago.

UPDATE (4/4/2013): Arturo Galletti offers a statistical comparison of the three longest NBA winning streaks in history (1971-72 Lakers' 33 games, 2012-13 Heat's 27, and 2007-08 Rockets' 22). Conditions faced by these teams, such as difficulty of opposition during the respective streaks, were not necessarily equal. Therefore, Galletti looks at power-rankings, among other stats, to try to quantify the impressiveness of the three streaks.

NHL: Penguins Win 14th Straight

The Pittsburgh Penguins won their 14th straight game last night, shutting out the Winnipeg Jets, 4-0. Three of the 14 games were tied after regulation play, with the Pens winning two of them during the five-minute sudden-death overtime period and one in the post-overtime shoot-out (game-by-game log).

The longest winning streak in league history is also held by the Penguins franchise, as the 1992-93 squad won 17 straight. Looking at the Penguins' 1992-93 game-by-game log, we see that the winning streak occurred from March 9-April 10, 1993, and that two of the wins occurred in overtime.

Five-minute overtime was introduced in 1983-84, which meant games could still end as ties. Shoot-outs to determine a winner, if the overtime period had not, began in 2005-06 (click here for a history of hockey overtime rule changes).

In the pre-overtime era, in which games could end as ties after 60 minutes, the NHL record winning streak was 15 games by the 1981-82 New York Islanders (January 21-February 20, 1982). Arguably, the Islanders' streak is more impressive than the '92-'93 Penguins' streak, due to the overtime rule changes. Had the rules in '92-'93 been such that deadlocked games after regulation been entered into the books as ties, the Penguins would have won only 12 straight.

UPDATE:  The Penguins extended their winning streak to 15 games, but then lost 4-1 to Buffalo on April 2.

Miami Heat Winning Streak Ended at 27 Games

With the Chicago Bulls' 101-97 victory over Miami last night, the Heat's winning streak has fallen six short of the NBA record 33-game mark set by the 1971-72 Los Angeles Lakers.

My primary impression from the Heat-Bulls game is that Miami never got the late barrage of three-point shots that had carried it to victories in other games during the streak. The Heat shot only 7-of-20 for the game from behind the arc. Four of these made shots occurred in the second half.

Miami did experience a little boomlet of three-point success late in the third quarter, when it was either tying or taking a one-point lead in the game. According to the play-by-play sheet, Mario Chalmers hit a three with 5:31 remaining in the third, followed by a pair of Shane Battier treys at the 4:30 and 2:25 marks of the third. What followed were seven straight Miami misses from downtown, until Chris Bosh made a three with just five seconds left in the game.

Contrast the Heat's poor three-point shooting at Chicago with how Miami had done while winning some of its most recent games during the streak:

• Against Orlando (win number 27), the Heat went 7-of-10 from long distance in the final 15 minutes of the game.
• Against Cleveland (win 24), Miami hit 6-of-7 treys during a fourth-quarter stretch in scoring an amazing comeback win.
• Four three-pointers in roughly a two-minute span helped Miami put away Toronto (win 22).

Comparing This Year's Miami Heat (27 Straight Wins) to the 1971-72 L.A. Lakers (33 Straight Wins)

Having increased its winning streak to 27 games Monday night, the Miami Heat is now within reasonable striking distance of the NBA record 33-game winning streak of the 1971-72 Los Angeles Lakers.

The Heat has had to come back from some pretty big second-half deficits, such as 27 points to Cleveland midway through the third quarter and 13 points to Boston with 8:26 left in the fourth. Nothing has come easy for Miami. Even the lowly Charlotte Bobcats, owners of the NBA's worst record (currently 16-54, .229) jumped out to a 19-8 lead over the Heat last Sunday, before Miami came back and took control.

In a previous posting, I noted that, in contrast to this year's Heat, the 1971-72 Lakers rarely seemed to have trouble during their 33-game streak. However, that conclusion was based only on final margins of victory, without taking into account the possibility that the Lakers struggled early in games and only pulled away toward the end.

I have now gone back and looked at the Lakers' 1971-72 game-by-game log, to get more detail on each of the team's games during the streak. Specifically, I checked the quarter-by-quarter line score, which is part of the box score for each game. To take an example, in the Lakers' December 3, 1971 game at Philadelphia, which was L.A.'s 16th win during the streak, the line score of points per quarter read as follows (I added the cumulative score at the end of each period in parentheses).

	1st	2nd	3rd	4th	Final Score
Lakers	19	39 (58)	36 (94)	37 (131)	131
76ers	30	39 (69)	24 (93)	23 (116)	116

HTML Tables

I then plotted the Lakers' leads or deficits at the end of each quarter, for each game during the streak. Also, to get an idea of how L.A. did against opponents of varying quality, I made separate plots for games against top-tier (.537 winning percentage and above), middle-tier (.439-.512), and bottom-tier (the expansion Buffalo, Cleveland, and Portland clubs, all of which finished last in their respective divisions) opponents, based on teams' records in the preceding, 1970-71 season. (The Lakers did not play against the Cincinnati Royals, who were .402 in 1970-71, during the streak.)

As the first graph shows, in games during the streak against top opponents, the Lakers tended to hold modest leads after the first quarter, which they gradually expanded over the course of the game. The bottom curve represents the above-described Philly game, in which L.A. trailed by 11 points after the first and second quarters, led by 1 after three quarters, and won by 15. The one game touching the break-even point (red line) after four quarters was a December 10, 1971 overtime win over Phoenix, the Lakers' 20th win during the streak. In that game, the Lakers failed to hold a 12-point lead over the Suns after three quarters. The box score shows Wilt Chamberlain going 0-for-6 from the free-throw line overall; if some of those attempts took place in the fourth quarter, that could explain Phoenix's comeback.

 

Against mid-tier opponents, the Lakers did have some first-half difficulties, but usually led after three quarters.

Finally, against bottom-tier/recent-expansion opponents, the Lakers generally did not win by blow-out. The Buffalo Braves (now the L.A. Clippers), in particular, gave the Lakers trouble in the first and second quarters. The 1971-72 Lakers thus were similar to this year's Miami Heat in sometimes facing early deficits against weak competition.

Why is margin of victory important? In addition to the coronary health of Heat fans, margin of victory affects the team's ability to give LeBron James and other top players some rest time. As I showed in a previous posting, it appears that "James gets three and one-half extra minutes of rest when the Heat is winning handily."