By In Baseball

The Timeless Game

Some years ago, my buddy Chardon Jimmy and I wanted to have T-shirts made that simply said: “Twenties Happen.” This is a Strat-o-Matic reference. In Strat-O baseball, when something is close to a sure thing — say a pretty fast guy scoring from second on a single with two outs — the game would often give you a “1-19” chance. That meant you would roll the 20-sided die, and assuming you rolled a 1 through 19, the runner would score.

But, we found, sometimes you rolled 20 (at which point the batter was thrown out — maybe he fell down or something). In fact, it happened five percent of the time. This has something to do with math.

That five percent, in case you are wondering, is also the same percentage of the time that baseball teams trailing going into the ninth inning come back and win the game.

There are two baseball phenomenons that are fascinating me these days. The first I’ve written about before: Teams leading going into the ninth inning have been winning 95% of the time more or less since the dawn of time. Yes, strategies change. Players change. Equipment changes. The use of relief pitchers evolves, the preparation of hitters evolves, the data used to set up defenses evolves, the game itself evolves.

In 1948, teams won 738 of 776 games they led going into the ninth. That’s 95%.

In 1968, that crazy year of the pitcher, teams won 1,315 of the 1,381 games they led going into the ninth. That’s 95%.

In 1977, when I was 10 years old and Duane Kuiper hit his only home run, teams won 1,788 of their 1,876 games. That’s 95%.

In 1989, when reliever Mark Davis won the Cy Young and Tony La Russa and Dennis Eckersley ushered in the era of the one-inning closer, teams won 1813 of 1890 games. That’s 95%.

In 2000, when the home runs were flying like balloons before a Super Bowl, teams won 2,081 of 2,190 games. That’s 95%.

Last year, teams won 2,032 of 2,137 games. And that too is 95%.

I have often thought that this all suggests that managers and general managers and players and writers and most of the rest of us are kind of fooling ourselves when making such a big deal out of closers or late game strategies or any of that stuff. The utter consistency of 95% suggests that the game has more or less drawn that line. Teams as a whole will win 95% of the games they lead going into the ninth. The rest is just rooting for good weather.*

*In case you want more involved numbers, going back to 1947 the win percentage is EXACTLY 95%. It rounds up or down to 95% every single decade since the 1950s — it goes as low as 94.6% and as high as 95.3%. But here’s the better point: There has never been a full season, not one since we have the numbers, where the entire league won 25 games over or under expectation. That’s in thousands of games. Baseball is so unpredictable on the micro-level, but you can say with certainty that at the end of this season, teams will lead 2,000 or so times going into the ninth and they will win 1,900 or so of those games. It happens every year.

Of course, there will be good and bad weather for individual teams. Some teams win every game they lead going into the ninth — that’s usually good for an announcer discussion in the postseason. Some teams blow an inordinate number of ninth-inning games. Then, even this often is beside the pint. Last year’s Florida Marlins won 98% of the games the led going into the 9th inning (wow!) which was a better percentage than Boston, St. Louis, Atlanta, Oakland, Pittsburgh, Detroit — you know, the best teams in baseball.

The Marlins still lost 100 games. Because they didn’t lead very much going into the ninth inning.

Yeah, those first eight innings still matter a lot more than the ninth..

Now, many people will say — and I understand this theory — that it’s only because of the game’s evolution that teams still win 95% of the time they lead into the ninth. In other words, they are saying that the newer strategies — the use of closers, the shifting of defenses, the study that now goes into the game — is the REASON that teams still win 95% of the time. This argument states that if you suddenly went back to the old days and stop using one inning closers and had your starting pitcher throw nine innings a lot and so on that your would be behind the times and would definitely lose more than your share of ninth inning games.

Maybe. But I have to say: I don’t think so. I tend to think more and more that all this ninth-inning scrambling (just like relentlessly watching pitch counts and shutting down starters after a certain number of innings) is more of a way for managers and teams to FEEL in control than it is about actually GAINING control. I think baseball, like every other sport, like more or less everything in life, follows the path of logic. it seems logical that teams using closers would win a la much higher percentage of the time than teams sticking with their starter or some three inning reliever. It seems like it HAS TO BE that way. But, some of these numbers suggest, it isn’t that way.

Give you another example, the second thing that fascinates me. I was absolutely convinced that quality starts were down in baseball. They HAVE to be down? I mean everything in starting pitching is down. Complete games are down. Shutouts are down. Wins are down. There have been 11 20-game winners the last five years combined. There were more than that in 1969 alone.

So, sure, the quality start HAS to be way, way down.

Except … it isn’t. Like with the ninth inning win, there a consistent statistical rhythm to quality starts. It’s not quite as consistent, but it’s close. Since the beginning of the live ball era, these two things have been true:

— About half of all starts end quality starts (that is: six innings pitched, three earned runs or less).
— Teams that get a quality start win about two-thirds of the time.

That’s the formula. Like I say, it fluctuates. In the 1960s, when pitching was king, pitchers threw quality starts 56% of the time. But because runs were so rare, those teams only won 64% of their quality starts.

In the 10 or so years of Bud Selig’s Power Hour — 1994 to 2004 — pitchers threw quality starts fewer than half the time but because so many runs were being scored they won 68.5% of the time.

So it will bounce around some. But in more normal times — if you assume the 1950s, 1970s, 1980s, and recent years are more normal — pitchers throw quality starts half the time (actually 52% or so) and teams win two-thirds of those. This was true in 1932, true in 1935, true in 1939, true in 1948, true in 1955, true in the expansion year of 1961, true throughout most of the 1970s and 1980s and 1990s until the strike. And it’s true now.

Last year: 52.8% of all starts were quality starts. And teams won 65.4% of the time.

The year before that: 52.1% of all starts were quality starts. And teams won 68.3% of the time.

The year before that: 53.6% of all starts were quality starts. And teams won 66.1% of the time.

All of this fascinates me because I”m fascinated about the idea of baseball timelessness. It’s something baseball fans talk a lot about — or at least the baseball fans I know. We talk about how 90 feet remains so perfect — a ground ball to short is an out in 1920 and today. We talk a lot about 60 feet 6 inches and how well that has held up (with a few alterations through the years to mound height and strike zones and so on). Baseball is the only game that pretends to share time; few serious people seem to believe that 1972 Dolphins or 1964 Celtics, as constructed, could play the game with today’s Seahawks or Heat.

But the 1965 Dodgers? With Koufax and Drysdale? You bet.

And one of the cool things about baseball is that the numbers often back this up. If tomorrow, every team in baseball decided all at once that every strategical advancement of the last 30 years is wrong and that everyone should basically run their team the way Earl Weaver or Billy Martin did in the 1970s, well, it would take some time to for all of us readjust. But the game itself, I think, would look the same.

Print Friendly, PDF & Email

43 Responses to The Timeless Game

  1. Jake says:

    forgive my ignorance – what happens when you roll a 20?

  2. MCD says:

    When I was a kid, there was no 20-sided die in a Strat-o-matic. It had a stack of cards, numbered 1-20.

    Obviously, a 20-sided die is a big improvement as not only did the cards require the work of shuffling, you pretty well knew that once the the 20 surfaced, it wasn’t coming up again immediately, even if you shuffled after every draw.

    • Dr. Baseball says:

      Actually, as I recall, Strat-o-Matic had 40 cards, each numbered 1-20 so that there always was the possibility of a 20, or a 1, coming up, even if you didn’t shuffle immediately. They called these cards “split cards.”

      • You probably mixed two decks to get 40. We did that so that a player didn’t KNOW a 20 wouldn’t come up again after it was out once. For that matter, the 18-20 or any other high numbers that players could count to know when to run.

    • Richard Aronson says:

      I knew a guy (when I was playing role-playing games a lot at CalTech) who would roll all his 20 sided dice 10,000 times, recording the results. The molding process used back then to create 20 sided dice left an irregular nub where the plastic was injected, and the size and shape of the nub made dice roll less true. He would use the dice that rolled lots of 10s and 20s and 1s and 11s as they were valuable in some circumstances, and other dice the rolled lots of high teens in other circumstances. Predicted results is ~500 of each number; IIRC his best die ever had 583 19s, which is statistically significant. So dice weren’t always better than cards. These days dice are made much better. Also, if you have a “true” shuffle (which almost nobody does) then five or seven shuffles are the best numbers for randomness.

  3. Jeff A. says:

    Joe I think the argument about teams continuing to win at the 95% rate is that they continue to do so because they follow the same strategy as all the teams in MLB.

    I would venture a guess that if 10 managers started using their starting pitchers for 9 innings per game, and the remaining managers continued to utilize closers and relief pitching, that those 10 managers would lose significantly more than 5% of the games they led going into the 9th. (I suppose it’s possible that the 10 could win significantly more than 95% – my point is it would only be reasonable to judge the efficacy of these different management strategies if there were significant samples each year of less homogenous managing).

    • Doug says:

      My intuition is that this doesn’t quite work, because only one team can lead going into the 9th, and so only one team can be in a position to give up a lead in the 9th.

      So you have your one group of managers who use old-school tactics, and they lose a certain percentage of games where they led going into the 9th. And you have your other group of managers, who use the shiny new tactics, and they lose a certain percentage of games where they led going into the 9th. But when one group loses, that doesn’t mean the other group wins. It means the team that was behind wins. The two sets of games and the two percentages are unrelated. The modern group does not gain a competitive advantage in a game vis-a-vis holding on to the lead from the fact that the other group is using outmoded tactics.

      As far as I can see, the only way that the group using old-fashioned tactics is going to have a lower percentage is if those tactics are less effective at getting three outs without giving up runs. And we know that they used to work at a 95% effectiveness rating, so they were effective at doing so at one point. So, as far as I can see, they should still be effective unless baseball offenses are just fundamentally better than they used to be, strategically or in terms of talent or whatever. Which is an open question – but like Joe, I kind of doubt it.

  4. Jake Bucsko says:

    Does someone have the means or the inclination to figure how many games the Yankees won after leading going into the 9th inning in the Mariano era? If the number is 95 or even 96 percent, it’d be interesting to see.

    Obviously Rivera was a great relief pitcher. His longevity and consistency are unquestioned, and his postseason record is extraordinary. But how differently would we view his career if his entire stat line were the same, only without the saves? Say instead of pitching the 9th inning (which apparently is 95% guaranteed anyway) he only pitched in the 6th, 7th, or 8th. Would he still be viewed the same way?

    It’s not an exact comparison, as closers tend to only come on with a 3 run or less lead. I’m sure the percentage in a 1 run game goes down significantly, as the percentage of winning with a 5 run or more lead in the 9th is probably 99.

    • I don’t have the data, but I have seen this exact analysis before. And the Yankees in the Rivera era won about the same percentage of their games when leading in the 9th as all the other teams in baseball – about 95%.

      • Dave says:

        Yeah, I’ve seen the same info but don’t have it handy. I do believe though that in the ’50s the data had the Yankees actually winning a slightly higher percentage of such games (2% maybe), but the Sparky years, Gossage years, and Rivera years were within a percent of each other. Anyway, that’s my memory, maybe from Baseball Prospectus or some SABR journal article.

    • Mark Daniel says:

      In the playoffs, Mariano Rivera pitched in the 9th inning with a lead 59 times (if I counted correctly), and the Yankees only lost 2 of those games (96.6%).

      If the Yanks had lost 3 of those games instead of 2, then that rate would have been 94.9%. So, you could argue that Mariano Rivera added 1 win to the Yankees playoff total over his entire career compared to the baseball average of 95% (only counting games in which he pitched in the 9th with a lead) .
      But, the two losses that did occur were huge – the DBacks game 7 victory in the 2001 WS, and the Red Sox game 4 victory in the 2004 ALCS. These two games are bona fide baseball all-time classics. Why are they classic? Because it’s rare to come from behind in the 9th to win a game.
      Moreover, those two games prevented the Yanks from having one more WS title (2001) and one more WS appearance (2004), if not another WS title.
      What would have happened had Rivera blown a 3rd save and brought the success rate down to the lower but still impressive 95% level?

  5. Polish1 says:

    I used to use a double deck of the 20 cards, so that there was still a possibility of that 20 coming back to back – not a big one but it still could happen.

    I’ve heard the 95% win stat before and am still amazed by it. Great article.

  6. TWolf says:

    I would like to see the comparative percentages over time of taking leads of one run, two runs, three runs, etc., into the ninth inning.

    • Richard says:

      Dave Smith at Retrosheet has done that work for you:

      “Teams leading by one run after eight innings have gone on to win 85.7 percent of the time. That number goes up to 93.7 percent when leading by two runs, and 97.5 percent when leading by three runs.” –

      In that article, Jim Caple further goes on to note that “basically, Rivera was not used except in games the Yankees were going to win 88 percent of the time anyway. Actually, the percentages were usually higher than that. According to Elias, of Rivera’s 652 career saves, just under a third (210) were with a one-run lead when he took the mound while 216 were with a two-run lead, 180 with a three-run lead and 46 with a lead of at least four runs.”

  7. Brent says:

    Are we saying that your manager can be Ned Yost stupid (more on that in a second) or Earl Weaver Smart, and you still will probably win 95% of the games you lead going into the ninth? That seems to show that managers are overpaid. As for my Ned Yost reference, the Royals won 2 of 3 in Baltimore this weekend, they had 4 innings in the 3 games that occured in the 9th or later, the 9th on Friday, the 9th and 10th on Saturday, and the 9th on Sunday. The Yost pitched his best bullpen arm 2 of the 4 innings, with a 5-0 lead in the 9th on Friday and with a 9-2 lead in the 9th on Sunday. He did not pitch Holland on Saturday in either the 8th or 9th, with the game tied. Of course, he won the two games that he would have won anyway had Holland pitched or not and he lost the other game. Does it matter though? Should I stop cursing his name for stupid pitcher usage, b/c it really doesn’t matter?

    • Doug says:

      If a manager’s job is primarily about on-field tactics, then yeah, they’re hideously overpaid. Because the difference between one manager and another is going to be marginal, and you could probably throw up a chart in the clubhouse that would tell you what to do and have it be correct most of the time. And, yeah, most of the time, the manager won’t make the team win or lose in the 9th, the players will.

      However, a manager’s job isn’t primarily about on-field tactics. And saying that discounts the importance of a manager in developing players and maintaining good chemistry and relations within the team, which are both pretty hard to evaluate or put a dollar figure on, but probably more important than on-field tactics.

      But feel free to curse Yost for dumb reliever usage anyway. Cursing managers for dumb reliever usage is our god-given right as baseball fans.

      • Chris Smith says:

        Managers are like symphony conductors. If you put a few dozen well trained, very talented musicians in front of me, with sheet music in front of them, I could swing a little stick and make them sound pretty good to the normal, everyday ear. Good conductors, though, can make them come together as a unit and shine amongst groups of other similarly talented and trained musicians.

        It’s hard to believe that team chemistry and shared goals make much of a difference in a game so built on having one man with one bat hitting one ball from one pitcher, but it absolutely matters. Just as much as it matters that a real orchestra has a good conductor.

    • Ian R. says:

      Remember, Joe is talking about the league as a whole, not individual teams. It’s entirely possible that teams with good managers win more than 95% of the games they lead going into the ninth, but they’re balanced out by the bad managers.

      So, yes, you can keep cursing Ned Yost’s name for stupid pitcher usage. It probably does matter, just maybe not as much as you’d think.

  8. Lars says:

    1) I’d totally buy that t-shirt. 2) baseball is beautiful.

  9. jess says:

    you’re the best Joe.

  10. What are teams win percentages after leading in the first inning? After each inning? Does it slowly increase as the game gets deeper (and there are less outs in which to make a comeback) or does the percentage stay the same for several innings?

  11. Herb Smith says:

    Doug brings up some good points about managers. And Yost isn’t the only anti-intelligence skipper in the game. The ones I find fascinating are guys like Ron Washington; his in-game maneuvers are truly awful. And yet, one of the most sabremetrically-inclined front offices continues to employ him.

    That leads me to believe that Joe’s basic theory (that the game is gonna play out in a similar fashion no matter WHAT strategies you use) is somewhat correct. And the other part o9f managing is far more important: keeping a smoothly-running locker room, instilling hope and desire and confidence in your players, intelligently dealing with the media, etc.

  12. Damon Rutherford says:

    If you’ve been paying attention, I’ve been alive and well for years within Joe’s message board (and at Baseball Think Factory).

  13. jim merritt says:

    So the one thing i’m sure about here is that 100% of managers have been trying to win their games in the 9th inning sice the beginning of time. If that has meant starters for nine, mutiple inning relievers or modern closers work it, they were going with what they thought was their best. What I reject is that if you throw some “slappy middle reliever out there” (john wasdin, burke badenhop, shawn camp, etc)in the 9th and think that will work consistently because JOPO says it has always been 95%, you will be disappointed quickly.

  14. Could it be that a strategy that improves a team’s odds of holding a lead in the ninth inning has a concurrent equivalent benefit to retaking the lead in the ninth?

    • Brett Alan says:

      No, that absolutely can’t be. Why would a team trailing in the ninth score runs because they use a closer when they have a lead in the ninth?

      Perhaps it could be that there’s a concurrent, different strategic advancement made by offenses, although I’m not sure what that could be. I don’t think the use of pinch-hitters, for example, has progressed much. I suppose you could cite the DH–I wonder whether the numbers in the closer era are different between the AL and NL? If the number has actually gotten better in the NL–if, say, it’s been 97% in the AL and only 93% in the AL–that would suggest that the closer was a very useful advance which was canceled out by the advent of the DH. But that seems unlikely to me, partly because a lot of the DH-type players in the AL would have been the ninth-inning pinch-hitters under the old rules.

  15. gogiggs says:

    Well, it probably won’t work consistently, depending on your definition of consistent, but it can certainly work for a season, maybe even two. It happens almost every year: Jason Grilli, Bob Wickman, Jim Johnson, Heath Bell, Chris Perez, Dave Aardsma, Bobby Jenks, etc, etc, etc,

    • Dave says:

      Even Chris Davis threw 2 innings of shutout relief for an extra innings victory about two years ago.

      In Baseball Prospectus a few years ago, there was a discussion of Joe Nathan. For the previous 3 or 4 years, he actually had better stats than Rivera. And for some time spans Hoffman was better too. Soria with KC for a couple of years. What distinguished Rivera was that he did it for so, so long (with no DL stints), and that he, because of his team, had so much playoff exposure and was usually shutdown then, even 2 inning stints which he rarely did during the regular season. (Trivia: in 2001, Rivera faced only one batter twice in the same outing, only one batter had actually seen his speed and cutter movement on that day in an at bat (from one day to the next, it could vary). It was Gonzo in the 9th inning of the 7th game of the WS. Because of what he’d seen in the previous at bat, Gonzo moved in the batter’s box from where he normally stood. Even then, his hit was only a flare.)

      • To expand upon your comment regarding Mariano Rivera and Luis Gonzalez: In 141 post-season innings, Rivera gave up 2 home runs. Moreover, he did it while pitching in the teeth of the steroid era, when utility infielders were hitting the ball over the fence. Case in point—Luis Gonzalez experienced a mysterious power explosion after his 30th birthday. In 2001, Gonzalez walloped 57 home runs, and yet in facing Rivera in Game 7, he opted to choke up on the bat to punch the ball over a drawn in infield.

  16. Tom Pareti says:

    “In 1948, teams won 738 of 776 games they led going into the ninth. That’s 95%.” What am I missing here? There were sixteen teams playing 154 games each (granted they were against each other) so 8×154=1232. Is it possible that 456 games were tied going into the 9th innning?

  17. Kuz says:

    I did not carefully read all the comments on this post so forgive me if I’m repeating a previous comment. If I understand this correctly, historically 95% of teams leading GOING INTO the ninth inning win the game. Given this, I think (maybe) the proper analysis would be to calculate the average LEAD going into to the last inning. Then calculate the average number of runs scored in each 1/2 inning of all games. Then set the null hypothesis as average lead going into last inning is greater than average lead going into the last inning minus average number of runs scored in each half inning all games. I would suspect the null hypothesis would be confirmed with a probability of approximately 95%.

  18. Brett Alan says:

    It’s funny…I would have thought that the percentage of “quality starts” would have gone UP over time. Correct me if I’m wrong, but if you give up, say, two runs in the first six innings, but stay in the game and give up two more runs in the eighth, that’s NOT a quality start, right? I’m thinking that pitchers giving up two or three runs are much more likely to be pulled from the game after six or seven innings than in the old days.

Leave a Reply

Your email address will not be published. Required fields are marked *