The Meaning of a Team’s Record in Mid-May

The Cardinals had the best winning percentage (.686) on May 15 this year. (via Arturo Pardavila III)

The Cardinals had the best winning percentage (.686) on May 15 this year. (via Arturo Pardavila III)

We’re a month and a half into the season, and about a quarter of the games have been played. If you’re a Houston fan, you’re probably wondering whether the Astros are actually as good as their early-season record would seem to suggest. If you root for the A’s, you want to know whether it’s already time to give up on 2015. So, just how indicative are their records?

I looked at every season from 1901 to 2014, a total of 2,392 team-seasons and a combined won-loss record of 186,832-186,832. (That includes 1914 and 1915, when eight Federal League teams played 154-game seasons. There was no postseason in 1901, 1902, 1904, and 1994, but I looked at the regular season records anyway.)

The first thing to note is that winning percentage truly is normally distributed, at all times of the year.

But there is a wider distribution early in the year than late — extreme winning percentages tend to even out, as you would expect. No matter what happens in the first quarter of the season, teams are likely to spend the rest of the year regressing to the mean winning percentage, .500.

The second thing I’ll note is that I’m not going to answer the question, “When Do the Standings Matter?” Chris St. John addressed that here, and the answer is, “Somewhere in August.” So I’m just going to take a look at what the mid-May outlook actually means.

Of course, it’s better for a team to start out hot:

Avg Win% by 5/15 Avg League Rank by 5/15 Avg Win %, End of Year Avg League Rank, End of Year
Above .600 by 5/15 .662 1.7 .560 3.8
Below .400 by 5/15 .335 9.9 .433 8.0

(Let me make a few more clarifying points… First, throughout baseball history, the number of games played does not precisely equal the number of wins and losses. For example, games that have been called for rain or darkness still count as a team game played. However, I am looking only at wins and losses.

Second, in concert with league expansion, numbers and norms have changed. From roughly 1904 through 1961, every team played 154 games, but since then it is typical to play 162, though in 1972 the American League went back to 154. Moreover, thanks to the addition of playoffs before the World Series, the start of the season has become earlier in recent years: it used to be normal to begin in mid-April, but many seasons now start in late March. As a result, the percentage of games played before May 15 has increased greatly. Before 1962, on average, 16 percent of games had been played by May 15; since 1996, the proportion has been 24 percent.

Finally, throughout major league history, there has been an average of 10.6 teams per league: there were eight teams per league until 1961, but after six rounds of expansion and several other realignments, there are now 15 teams per league. I looked at “normalizing” league rank by multiplying a team’s league rank by 10.6 divided by the number of teams in the league at that time — so that a team’s rank would get higher and thus worsen in a smaller league and get smaller in a bigger league — but it barely changed the averages, over the past 114 baseball seasons. So I am just reporting the absolute numbers.)

Exactly three teams that posted a record below .400 on May 15 managed to finish with the best record in their league, and all three of them won the World Series: the 1906 Hitless Wonders White Sox, the 1914 Miracle Braves, and the 1979 We Are Family Pirates.

There’s a reason they called them the Miracle Braves: the Bostonians stood at 3-15 on May 15, but at some point over the summer, they basically stopped losing, going 49-14 in August, September and October, and sweeping the World Series over the mighty Philadelphia Athletics.

Only one other team started the year under .400 into mid-May and went on to win the league pennant: the 2005 Houston Astros, who began 14-23 before taking off in June and July to secure a Wild Card spot. They then sent the Braves home in an 18-inning, five-hour and 50-minute Division Series Game Four — only the second playoff series win in franchise history, matching their 2004 dispatch of the same Atlanta Braves — and then overcame the Cardinals, who had finished 11 games ahead of them in the division. But the Chicago White Sox swept them in four games, and the Astros have not been to the playoffs since.

Other than the Wild Card Astros, three other teams have won their division after starting the year under .400: the 1974 Pirates, 1984 Royals, and the 1989 Blue Jays.

(There’s a fourth “division winner,” the 1981 Royals, but they come with a major asterisk: that was the year that was interrupted by a summer strike, and so “division” champions were crowned for the first and second halves of the season, followed by a division series at the end of the year before the LCS and World Series. The Royals went 20-30 in the first half but 30-23 in the second half, and so they count as a division winner in the standings, but it’s an anomalous season.)

In all, only 1.7 percent of teams that were at or under .400 on May 15 actually made it to the playoffs. In fact, very few teams below .500 on May 15 actually make it to October, but your odds improve considerably the closer you are to the break-even point. Just 4.6 percent of teams that were at or under .500 on May 15 made it to the playoffs. However, of the teams that were below .500 but managed to stay above .400, 6.7 percent made it.

In addition to the three miraculous world champions mentioned above, nine teams have won the World Series despite a May 15 record below .500, but only two pulled it off after the season expanded to 162 games: the 1969 Miracle Mets and the 2003 Florida Marlins. The Mets started 15-18 before kicking into high gear, earning their nickname, and collecting the first championship in franchise history. The 2003 Marlins were so moribund coming out of the gate that they fired Jeff Torborg on May 11, after he had led the team to a 16-22 record; new manager Jack McKeon was a spry 72 years old and hadn’t managed in three years, but he led the team to a Wild Card berth and a World Series victory.

Total teams % made playoffs % won World Series
Above .600 by 5/15 450 40% 14%
Btw. .400 and .500 by 5/15 637 7% 1%
Below .400 by 5/15 450 2% 1%
Overall 2316 17% 5%

In other words, if a team is truly in the cellar right now, it probably doesn’t have a lot of hope. Nearly 60 percent of World Series winners — 61 of 107 — were above .600 on May 15, and 41 of them had a winning percentage over .650, which is a 105-win pace.

So, what about the teams that started out at a scorching pace? On the whole, they do pretty well: no matter what happens after May, they still get credit for all the wins they banked in the early weeks. For example, the 1984 Detroit Tigers were 28-5 in mid-May — that’s an .848 winning percentage, a 137-win pace. The Bless You Boys actually stretched their incredible start to 35-5, so even though they fell off quite a bit, going just 69-53 (a .566 winning percentage) the rest of the year, they still finished with 104 wins and won the World Series.

One of the more notable recent examples of a team that started out hot and then simply cratered was the 1995 Phillies, who were still the reining National League champions after the 1994 canceled postseason, and who sat at 13-5 on May 15. (Due to the strike, the 1995 season didn’t begin until late April.) But they went 9-20 in July and ultimately finished 69-75.

Fun facts about that team: the closer was Heathcliff Slocumb, whom they had obtained two years prior by trading away Ruben Amaro, Jr., the current Phillies general manager. Over the 1995-1996 offseason, they traded him to the Red Sox, in return for Glenn Murray, Ken Ryan and Lee Tinsley, who combined for 161 plate appearances and 148 innings in a Phillies uniform. A few months later, in May 1996, the Phillies re-signed Amaro. A year later, in July 1997, the Red Sox traded Slocumb to the Mariners for Derek Lowe and Jason Varitek.

In fact, teams that start out with a .600 record on May 15 almost always finish with a respectable season. In the expansion era, there have only been 21 cases of a team finishing the season below .500 after starting the year above .600, and two of those occurred in strike seasons.

Generally speaking, the better a team starts, the better it ends up:

% won World Series % made playoffs
Above .800 by 5/15 36% 55%
Above .750 by 5/15 32% 54%
Above .700 by 5/15 25% 46%
Above .650 by 5/15 18% 46%
Above .600 by 5/15 14% 40%

Winning a World Series is never a foregone conclusion, but if a team wins the vast majority of its games in the early going, it usually will waltz into the playoffs.

The 2002 Red Sox were an unfortunate exception. On May 9, they were in first place at 27-9, four games ahead of the 25-15 Yankees. But after going 66-60 the rest of the way, their record at the end of the year was a merely respectable 93-69. That left them 10.5 games behind the 103-58 Yankees, and six games behind in the Wild Card race, losing out to the 99-63 Angels, who went on to win the World Series. (The Angels were at 22-16 on May 15, a .579 record.)

The 1972 Twins started out 16-6, but that was effectively the high-water mark for the season, as they finished 77-77. A midseason managerial change can’t have helped: the Twins fired manager Bill Rigney in early July, when their record was 36-34. Under new manager Frank Quilici, the team went 41-43 the rest of the way. Quilici ultimately amassed a 280-287 managerial record with the Twins from 1972-1976. After that, he was fired, and never managed again.

It’s too early to know what will happen the rest of the way; as they say, that’s why they play the games. But after 40 or so games, there’s enough information for a pretty accurate cull. There will be a group of teams that has almost no chance of a sniffing October, and there will be a group of teams that are likely to be relevant all year.

So if your team is trying to decide whether to be a buyer or a seller, it probably should already know by now.


Alex is a writer for The Hardball Times, and is an enterprise account executive for The Washington Post.
newest oldest most voted
JR
Guest
JR

Thanks for all of the digging here Alex. Any way to account for playoff expansion effects on these percentages? I’d assume that a .400 team has a slightly better chance of attaining a second wild card spot in 2015 than they did winning their division in 1985.

Charles Ballaro
Guest
Charles Ballaro

They played roughly 154 games in 1972 because of a player’s strike. that wiped out a week’s worth of games.

Matthew
Guest

If anything, this article understates the severity of the collapse of the ’95 Phillies. They were actually 37-18 on June 25, with a 4.5 game lead in the division. They then went 3-15 over the next 18 games. Amazingly, they were still over .500 as late as Sept. 14, at 65-64. No matter how you cut it, starting 37-18 and going 32-57 the rest of way is just mind-boggling.

Tom Au
Guest
Tom Au

Few teams manage to win their division after a bad early start, but two teams that did were the 1974 and 1979 Pittsburgh Pirates.

Marc Schneider
Guest
Marc Schneider

The ’92 Braves were 20-27 and ended up 98-54 and won the division by 10 games.

MGL
Guest
MGL

Of course if you actually know something about a team, other than it’s win/loss record in the dark, that completely changes the calculus. Sort of like when a team is down 3-1 in a playoff series and you see a graphic that shows that only 1 out of 12 teams down 3-1 have ever come back. Well, if the team down 3-1 was actually the better team, the real odds might be 4-1 or 5-1 and if they were the worse of the two teams, it might be 14 or 15-1. The Astros likely ROS record and chances of making… Read more »

Roger Miller
Guest
Roger Miller

Quite interesting as I had completed a study on the effect of April covering teams from 2000-2014 which incorporated all 30 current MLB clubs. In my study I compared whether April or the prior year was a better predictor of the last 5 months of the season. Prior year (.94 Std Dev) turned out to be better than April (1.32 Std Dev) at forecasting remainder of season and which teams would make the playoffs. But good or bad starts combined with prior year results yielded some very predictive outcomes. Teams who were good the prior year (top 30%) and also… Read more »

Jfree
Guest
Jfree

You say a team should probably know if they are a buyer or seller by now. What if they are a bad team because they are actually incompetent off-the-field as well as on-the-field? Is there any hope at all for fans of that team? Or is it better that they cut their losses at this point and switch to an alternative sport like competitive knitting?

BillyF
Guest

I think “dividing” the historical teams into groups of every 0.5 win% (.600, .550, and so forth) is not depicting the real picture, drive of team morale, and fan expectation (projected income!) that pushed a club (owner) to be buyer or overachievers to boost a win or not. What is the “Win% of making to the postseason TODAY?” That’s the real question that will address the author’s original theorem and sources. After all, we know there’re not only 4 teams, but 5 teams that are in the postseason (two Wild Cards). Based on that win% (*), let’s say if it’s… Read more »

jss
Guest
jss

Your first chart is missing information for context: win pct for rest of season. See http://tangotiger.com/index.php/site/comments/regression-toward-the-mean-example

Michael Bacon
Guest
Michael Bacon

Question for Roger Miller.You write, “In my study I compared whether April or the prior year was a better predictor of the last 5 months of the season. Prior year (.94 Std Dev) turned out to be better than April (1.32 Std Dev) at forecasting remainder of season and which teams would make the playoffs.”
There is a huge discrepancy in the sample size between one month and a full season. What if you compared the first month of the season to the last month of the previous season?

Henry
Guest
Henry

I think the point is a bigger sample even if slightly more out of date (last season) and presumably with substantially more roster turnover is better than a smaller more recent sample (April).

Roger Miller
Guest
Roger Miller

Certainly sample size is one reason that prior year (bigger) is a more predictive than the current April (smaller). The other factor is that not many teams have radical changes of fortunes year to year. Only 20% of all teams had a 15+ win variance from the prior year. Which is why good or bad Aprils can often send off false signals of change. However for those 20% of teams who do change dramatically year to year, April has been an interesting indicator. Teams who improved 15+ wins over the prior year had an average April win % of .558… Read more »