I've been working under the assumption that if an analytical question seemed pressing enough for me to take a quick and dirty look at it in medias res, while the season was going on, it's probably worth considering after the dust has settled.
Because the whole thing got so big and unwieldy, I'm doing it in two parts. The companion piece, "Those 47 Games," looks at Toronto's performance in much more detail. I'll post it tomorrow. What we have here is more of an overview of the whole subject.
Immediately after the season ended, we asked What Would Pythagoras Say? and examined the large gap between the win-loss record we would have expected from a team that scored and allowed as many runs as the Blue Jays. As you may recollect, a team that scored 775 runs and allowed 705 would normally be expected to post a record of about 89-73. While numerous suggestions were made to account for this unusually large discrepancy, the big elephant in the room, of course, was Toronto's abysmal record in games decided by one run. The Jays played .340 (16-31) ball in one-run games.
Toronto's record in one-run games was the worst mark in the American League this past season: yes, even worse than Kansas City. (It was actually the worst mark in the major leagues, but I'm just going to run the AL numbers. If you're curious, the best NL team was Arizona, followed by San Diego and the Cubs; the worst was Pittsburgh, which figures, but St. Louis was surprisingly bad as well.) Anyway, here are the numbers:
One-Run Games Other Games
W L PCT W L PCT Diff
Chicago Sox 35 19 .648 64 44 .593 .055
Boston 27 15 .643 68 52 .567 .076
NY Yankees 27 16 .628 68 51 .571 .057
LA Angels 33 26 .559 62 41 .602 -.043
Tampa Bay 29 25 .537 38 70 .352 .185
Seattle 26 23 .531 43 70 .381 .150
Oakland 26 24 .520 62 50 .554 -.034
Minnesota 27 30 .474 56 49 .533 -.059
Detroit 22 26 .458 49 65 .430 .028
Texas 24 29 .453 55 54 .505 -.052
Cleveland 22 36 .379 71 33 .683 -.304
Kansas City 18 30 .375 38 76 .333 .042
Baltimore 14 25 .359 60 63 .488 -.129
Toronto 16 31 .340 64 51 .557 -.217
Toronto played .557 ball in all the other games, a difference of -.217. This was not, as it turned out, the largest differential in the majors - hello Cleveland! - but it was still pretty awful. Let's just reshuffle the same chart to show you that ranking (and for you NL fans, St. Louis had the largest falling-off, and Arizona had the biggest improvement in the major leagues):
One-Run Games Other Games
W L PCT W L PCT Diff
Tampa Bay 29 25 .537 38 70 .352 .185
Seattle 26 23 .531 43 70 .381 .150
Boston 27 15 .643 68 52 .567 .076
NY Yankees 27 16 .628 68 51 .571 .057
Chicago Sox 35 19 .648 64 44 .593 .055
Kansas City 18 30 .375 38 76 .333 .042
Detroit 22 26 .458 49 65 .430 .028
Oakland 26 24 .520 62 50 .554 -.034
LA Angels 33 26 .559 62 41 .602 -.043
Texas 24 29 .453 55 54 .505 -.052
Minnesota 27 30 .474 56 49 .533 -.059
Baltimore 14 25 .359 60 63 .488 -.129
Toronto 16 31 .340 64 51 .557 -.217
Cleveland 22 36 .379 71 33 .683 -.304
Were it not for the one-run games, the post-season would have had a different cast of characters. In the AL, while the Yankees and Angels would still be division winners, Cleveland would be the Central champs and the White Sox would have taken the Wild Card. (In the NL San Diego and St. Louis would have won their divisions; but Philadelphia would have been the best in the NL East, and Atlanta would have won the Wild Card.)It is certainly interesting that St. Louis and Cleveland, two of the worst teams in one-run games, also had two of the best pitching staffs in all of baseball. Tampa had dreadful pitching and did rather well in one-run games. This could also be said of Cincinnati and Kansas City, although in Kansas City's case "rather well" only means comparatively better than their overall performance.
But what are the implications? Does quality of pitching not matter in determining the results of the close games? That seems extremely doubtful - and we all know that the White Sox pitching was as good as anybody's, and no team was better in close games.
So here are some rhetorical questions we may all ponder. Why do these things happen? Why do some teams do better than others? Why do teams improve upon, or fail to match, the level suggested by the rest of their games.
When I originally examined this issue, I suggested four possible explanations for performance in one-run games. They were: 1) luck, 2) the bullpen, 3) the bench, and 4) the manager. I have since been inspired to consider a fifth factor - the ability to hit home runs.
After all, what kind of games are we talking about? We're talking about low-scoring games. In the AL, the teams combined to score 8.00 runs in one-run games, as opposed to 9.44 runs in all games, and 10.09 runs in non-one-run games. This is what you would expect to see - after all, 2-1 games are considerably more common than 12-11 games.
It's well known that teams that hit home runs (as opposed to other kinds of offense) have an advantage in the post-season, when scoring also tends to be depressed. It's not because more home runs are hit in the post-season, but because the home runs that are hit account for a larger share of the runs that are scored. Generally, long-sequence offenses don't work quite as well in the post-season because it's harder to put together a series of hits against quality pitching. Could regular season one-run games be similar? These would also tend to be lower-scoring games, and so the home runs that are hit similarly ought to have a larger impact on the game result.
I was inspired to add home runs to my list of possible factors because... because I am a shrewd and savvy student of the game? You're not buying that? OK. Looking at the charts above, I noticed among the teams that performed best in one-run games were outfits like the White Sox, Yankees, Red Sox (not to mention the Diamondbacks). What do they have in common? They all hit lots of home runs. Those are four of the top eight HR hitting teams in baseball. It's the one thing their offenses truly have in common. While Boston and New York were the two highest-scoring teams in baseball in 2005, neither Chicago (8th in the AL) nor Arizona (10th in the NL) had a particularly impressive offense. The one thing they all did well was hit home runs.
Well, could this be it? Could this be the key? Could it be that simple? Not the ability to play for a single run, not the quality of your relief pitching, but rather the number of home runs you hit? Fat chance. Hitting lots of home runs doesn't seem to have helped Texas or Philadelphia much in their one-run games. Cleveland and St. Louis, who also hit lots of home runs, were among the very worst teams in one-run games. Hardly a reliable indicator: back to the drawing board. Let's return to my original Gang of Four:
1. Luck. These results are for the most part a matter of random chance, the luck of the draw. The mind resists this, but there are good reasons to regard it as the most likely explanation. You don't lose by eight runs because you didn't catch a break - you lose by eight runs because you weren't very good that day. But you can win (or lose) by one run because the game is so close that the impact of random chance is sufficient to overwhelm the impact of overall quality.
Luck, I think, accounts for the remarkable performance in one-run games by the Arizona Diamondbacks. Arizona went 28-18 in one run games - the rest of of the time, they were 49-67. They gave up 160 runs more than they scored - the second worst mark in the major leagues. This is partially because of their uncanny ability to get beaten senseless - in 12 games where the margin of victory was 8 runs or more, Arizona went 1-11, scoring 35 runs and giving up 155. This is the tell-tale sign of a bad ball club, and that's what they are. Arizona was a lousy team, who only managed to go 77-85 because of a remarkable run of good fortune in some close ball games. I think it was mostly random luck, but they may have had a couple of other things going for them.
2. The bullpen. We tend to assume that relief pitchers are the ones on the mound when these games are decided. It is by no means certain that this is actually the case. But if that is true, logic certainly suggests that the teams with the best bullpens would have an advantage. This might particularly apply to the team with the best relief pitchers, as opposed to the team with the deepest bullpen. The Toronto bullpen, as a group, is much deeper and stronger than the New York Yankees bullpen. But no Toronto reliever is better than Mariano Rivera.
None of this will explain why the Cleveland Indians were so bad in one-run games. And they were very bad indeed - so bad that it clearly cost them the AL Central. If Cleveland had played .500 ball in their one-run games, they would have wound up with 100 wins, which would have been enough. If the White Sox had played .500 ball, they would have finished with 91 wins. Cleveland wins the division if either of those things had happened - and in the event of both, they would have won easily, by nine games.
Cleveland had the best team bullpen in baseball in 2005. Their relievers went 22-18, 2.80, and theirs was the only relief corps with an ERA below 3.00. It didn't seem to do them much good in the close games, did it?
The three worst bullpens in baseball? Glad you asked - it was Arizona, Boston, and Tampa Bay. We've already noted how well Arizona did in one-run games. As it happens, the Red Sox were one of the best teams in the AL in these games, and Tampa played much better in the close games than they did the rest of the time. At any rate, quality of bullpen does not seem to be a reliable indicator as to how teams perform in close games.
3. The bench. A team with depth, a team that can bring in talented pinch-hitters at key moments, ought to have an advantage in the late innings of a close game. This might be more important in the National League, where there tends to be more pinch-hitting and in-game substitutions. NL teams got 195 to 272 at bats from pinch hitters - Kansas City led the AL with 103 pinch hit at bats.
Obviously, there is much more to having a good bench than having good pinch hitting - but it does seem to be the one specific aspect of the bench that is likely to have an impact in a close game. The Blue Jays got excellent production out of their pinch hitters in 2005 - they hit .324 in 102 at bats. Among the teams with lousy pinch hitting, we do find some teams who struggled badly in close games - Baltimore and Cleveland. But we also find some teams who did just fine in close games, or at least better than their overall performance - Boston, Kansas City, the Cubs.
4. The manager. One likes to think that superior use of in-game strategies should have an impact in close games. I don't think too much of this myself - while I think there are enormous differences between managers, I think game management might be the area where there is the least to choose between them. It is very far from being the most important skill needed to manage a major league team. While I see managers make in-game decisions that I disagree with all the time, I almost never see a manager make a decision that I can't see any reason for whatsoever. It's led me to believe that if I don't understand a manager's decision, there's probably something the manager knows that I don't.
I don't think there's a real consensus on which managers are best at in-game strategy. The most experienced managers are Tony LaRussa, Bobby Cox, and Joe Torre. Torre's team did very well in close games, Cox's team did OK, and LaRussa's team did not play well at all.
I don't have any good ideas for talking seriously about the impact of managers when the game is close in the late innings. So let's peruse instead the ESPN splits and look at how teams performed when it was close and late.
CLOSE & LATE AB R H 2B 3B HR RBI BB SO SB CS SH SF HBP GDP BAV OBP SLG OPS RC RC/27 Boston 700 133 209 30 3 29 130 85 123 8 1 3 6 10 25 .299 .380 .474 .854 126 6.45 NY Yankees 734 127 191 35 3 25 121 107 142 10 0 10 6 7 16 .260 .357 .418 .775 117 5.34 Cleveland 852 128 222 48 5 30 122 86 205 5 7 16 9 7 19 .261 .330 .434 .764 122 4.75 LA Angels 923 124 249 36 3 22 115 88 154 30 3 21 3 6 17 .270 .336 .387 .723 127 4.64 Texas 841 119 206 32 5 34 115 66 197 15 2 3 4 6 16 .245 .303 .416 .719 107 4.26 Detroit 909 107 236 38 9 19 102 81 195 11 4 15 12 7 18 .260 .321 .384 .705 116 4.21 Oakland 899 123 228 45 2 21 113 91 156 4 3 13 8 11 21 .254 .327 .378 .705 114 4.19 Baltimore 743 114 189 43 2 20 106 71 122 13 5 19 9 8 23 .254 .323 .398 .721 95 4.18 Seattle 939 127 235 50 5 19 117 101 183 20 4 15 4 8 26 .250 .327 .375 .702 118 4.16 Toronto 962 120 252 51 1 18 112 85 181 9 5 13 11 12 24 .262 .326 .373 .699 119 4.12 Chicago Sox 937 120 231 39 3 32 117 88 201 18 11 21 5 11 26 .247 .317 .397 .714 116 4.04 Tampa Bay 861 108 214 40 7 16 101 70 195 34 8 13 7 17 18 .249 .315 .367 .682 104 3.98 Minnesota 1034 127 241 52 7 21 118 130 223 22 9 19 4 13 21 .233 .325 .358 .683 127 3.97 Kansas City 813 93 181 28 2 15 84 67 182 5 3 15 7 11 17 .223 .288 .317 .606 77 2.98 AVERAGE 867 119 220 41 4 23 112 87 176 15 5 14 7 10 21 .254 .322 .389 .711 113 4.38This doesn't provide anything we could call a reliable indicator either. The Red Sox and Yankees, who do well in close games, obviously hit very, very well late in close ball games. Actually, those two teams hit very well pretty much all the time anyway. And once again, we look at Cleveland and the White Sox and we scratch our heads, we furrow our brows, and look generally perplexed.
Maybe it's the pitching, after all?
TEAM G IP H SO BB HR ER BAA SLG ERA Seattle 104 216.3 217 154 88 29 53 .264 .420 2.20 Minnesota 120 289.7 253 214 85 25 76 .239 .350 2.36 Cleveland 116 262.0 222 195 74 29 72 .231 .368 2.47 Baltimore 110 195.0 192 180 88 28 54 .259 .423 2.49 Detroit 99 223.7 207 151 79 20 66 .248 .359 2.66 Toronto 114 247.0 234 173 100 28 74 .252 .384 2.70 Kansas City 101 193.3 204 179 86 15 59 .272 .399 2.75 Oakland 108 207.7 181 177 74 22 64 .232 .357 2.77 Chicago Sox 113 274.7 239 219 112 26 89 .236 .380 2.92 LA Angels 113 265.7 221 246 102 26 95 .226 .357 3.22 NY Yankees 100 183.3 164 133 54 16 67 .237 .368 3.29 Tampa Bay 109 218.0 229 129 100 25 87 .275 .433 3.59 Texas 114 230.3 240 171 96 21 102 .269 .402 3.99 Boston 104 190.7 213 133 75 18 106 .284 .441 5.00 AVERAGE 108 228.3 215 175 87 23 76 .018 .387 2.99No, I don't understand how Seattle's pitchers could give up that many hits and walks and home runs late in the game, and still give up so few earned runs. Perhaps a lot of timely errors created a lot of unearned runs. And once more, the Cleveland-Chicago comparison is simply bewildering. When the game was Close and Late, Cleveland both out-hit and out-pitched the White Sox. It very clearly had no impact whatsoever on how the teams did in games decided by a single run.
Well, when in doubt, change the subject. Let's run at this question from a slightly different angle. What specifically was different about team performance in one-run games, compared to how they performed the rest of the time? What happened to the offense, what happened to the pitching? Let's make another data table!
Team Overall One-Run Not One-Run W L R RA W L R RA W L R RA Chicago 99 63 741 645 35 19 195 179 64 44 546 466 Boston 95 67 910 805 27 15 207 195 68 52 703 610 New York 95 67 886 789 27 16 196 185 68 51 690 604 LA Angels 95 67 761 643 33 26 229 222 62 41 532 421 Tampa Bay 67 95 750 936 29 25 227 223 38 70 523 713 Seattle 69 93 699 751 26 23 198 195 43 70 501 556 Oakland 88 74 772 658 26 24 186 184 62 50 586 474 Minnesota 83 79 688 662 27 30 184 187 56 49 504 475 Detroit 71 91 723 787 22 26 172 176 49 65 551 611 Texas 79 83 865 858 24 29 256 261 55 54 609 597 Cleveland 93 69 790 643 22 36 212 226 71 33 578 417 Kansas City 56 106 702 935 18 30 215 227 38 76 487 708 Baltimore 74 88 729 800 14 25 135 146 60 63 594 654 Toronto 80 82 775 705 16 31 174 189 64 51 601 516 LEAGUE AVG 82 80 771 758 25 25 199 200 57 55 572 559Because my readers are keen-eyed Bauxites, sharp and perceptive, you have surely noticed that the league totals do not balance. Let me explain! This is because of inter-league play. The AL had a winning record against the NL in 2005, but the NL did a little better in the one-run games between the two leagues.
This chart gives us a kind of base-line. We can see that in the AL, teams scored 3.99 runs per game in one-run games, 4.76 runs per game overall. In other words, they maintained 78% of their offensive output.
As you would expect, the pitching improves accordingly - not quite as much, thanks to inter-league play, but almost. AL teams allowed 4.01 runs per game in one-run games, 4.68 runs per game overall. The opposition, therefore, maintained 80% of their offensive output.
Having done that, we can now pick out how individual teams deviated from this overall pattern. And folks - this is where things get as clear as mud. But I've been peering and peering at these numbers, and shapes are beginning to emerge from the fog:
Team Overall One-Run Not One-Run R/G RA/G R/G RA/G R/G RA/G Chicago 4.57 3.98 3.61 3.31 5.06 4.31 Boston 5.62 4.97 4.93 4.64 5.86 5.08 New York 5.47 4.87 4.56 4.30 5.80 5.08 LA Angels 4.70 3.97 3.88 3.76 5.17 4.09 Tampa Bay 4.63 5.78 4.20 4.13 4.84 6.60 Seattle 4.31 4.64 4.04 3.98 4.43 4.92 Oakland 4.77 4.06 3.72 3.68 5.23 4.23 Minnesota 4.25 4.09 3.23 3.28 4.80 4.52 Detroit 4.46 4.86 3.58 3.67 4.83 5.36 Texas 5.34 5.30 4.83 4.92 5.59 5.48 Cleveland 4.88 3.97 3.66 3.90 5.56 4.01 Kansas City 4.33 5.77 4.48 4.73 4.27 6.21 Baltimore 4.50 4.94 3.46 3.74 4.83 5.32 Toronto 4.78 4.35 3.70 4.02 5.23 4.49 LEAGUE AVG 4.76 4.68 3.99 4.01 5.11 4.98Observe! Overall, the 2005 Blue Jays had an offense slightly better than league average, and pitching that was a good bit better than average. Their problem in one-run games were on both sides of the ball - the better than average pitching becomes just league-average in this new context, and the offense is significantly worse than average - again when it is measured against league averages in one-run games.
It's a different offensive context - all AL teams reduced their runs allowed, surrendering just 80% of the runs they allow overall. But Toronto didn't match the overall improvement - they allowed runs at a 90% rate compared to their overall performance. The effect was their normally better than average pitching performed at a level roughly equal to the league average. And while other AL teams scored 78% as many runs in these games as they normally did, the Jays fell off even more, scoring just 71% as often in one-run games.
And finally, we see something that suggests what happened to Cleveland in their one-run games. They didn't hit. No other team in the AL lost as much from their offense in one-run games. It's probably very, very unwise to expect anything to correspond across the board with success in one-run games. Boston hit very well in close and late situations; the White Sox didn't. But both teams did well in one-run games.
When it we got to the late innings of a close game, Cleveland hit better than the White Sox and pitched better than the White Sox. They had a better bullpen. But in their 58 one-run games, Cleveland's offense let them down. Despite the fact that their hitters performed quite well when it was Close and Late. That much I am sure of. But how or why or whatever... beats me.
In the end, I suppose I keep going back to Random Chance because, unlike the other explanations, it's impossible to find fault with it. (It's always hard to quantify the fog.) Everything else we look at simply doesn't satisfy, at least not as a reliable indicator. For every team whose success might be attributed to a strong bullpen, we find another team with a better bullpen, who performed dreadfully in one-run games. And this applies to pretty well everything I've thought to examine. There's no single explanation. Which really isn't all that surprising; but there isn't even a single thing in common, which is slightly surprising.
In overall play, we see much the same thing, of course - there are many, many different ways to be a good team, to win 90 games (and just as many ways to be bad.) But all teams that play .600 ball actually do have one thing in common - they're all good baseball teams. (There are exceptions, of course, but they are pretty rare.) That simply does not apply in one-run games. The margin is too small for overall team quality to have the same impact.
By looking very, very carefully at each individual team's performance, we can isolate what happened to that team in those games. We can find what went wrong,or what went right. What we are unlikely to find is a reliable indicator that will tell us what kind of team will do better in these games than another.
It's just One of Those Things - when it happens, you can always find specific things that happened as part of the phenomena. But I don't think that you can identify what created the phenomena in the first place.
