[Our old friend Gregi Gross, with whom I discussed baseball in Germany in a three-part series entitled "Eine Kleine Chin Music" (parts I, II, and III) some time ago, has conducted a study on what it takes to be a winner in baseball. I am proud to present it here. Enjoy.]

Heading into the 2002 season, so says Michael Lewis in Moneyball, then Oakland A´s assistant GM Paul DePodesta faced a serious task. With Jason Giambi, Jason Isringhausen and Johnny Damon the team was going to lose three core players to free agency, and Athletics GM Billy Beane wanted DePodesta to determine exactly what influence those three players had on the 102 wins the team had compiled in 2001. The idea was to offset these losses by signing other players with different skill sets that together would bring to the table what free agency had robbed the team.

Paul DePodesta decided to start by judging how many wins a team would need to make it to the play-offs.

"There aren´t a lot of teams that win ninety-five games and don´t make it to the play-offs," he said. "If we win ninety-five games and dont make the play-offs, we´re fine with that."

I wanted to find out if Paul DePodesta assumptions were right. Is 95 the magic number of games a teams needs to win to safely enter the postseason? And which run differential had historical teams brought into the play-offs?

Since for the last ten years we have enjoyed the wildcard, I first looked at the seasons from 1995 until today. Looking at each division individually, I prorated all seasons to 162 games where needed (in 1995 the season was shortened, for instance). Since some teams in some years clearly had franchise seasons (like the 1995 Indians, the 1998 Yankees or the 2001 Mariners) or in some years an entire division struggled, I decided to also include averages without those peaks:

AL East

AL Central

AL West

AL Wildcard

And for the National League we get this picture:

NL East

NL Central

NL West

NL Wildcard

For Paul DePodesta's team, the Oakland Athletics, the average number of wins needed to win their division was 95. It was 93 if you discount the 2001 Seattle Mariners and their 116 win campaign. If the A´s would face serious competition in their division, they could also qualify via the wildcard. There they would need 94 wins on average, and only 93 if you discount their own 2001 campaign, when they trailed the Mariners but took the wildcard with 102 wins (still the most wins for a wildcard team in its short history).

What else can we see? Judging by the number of wins needed to win a division, the weakest division in baseball is the NL Central, followed by the AL Central and the AL West. The best division then would be the NL East, followed by the AL East. Then again, both eastern divisions probably feature very good and very bad teams, so the division winners run up high win totals against weak competition. And in the NL Central there is tough competition among almost equal teams, so their win totals stay low.

So overall the average of the average (of the average...) for all of MLB looks like this:

So yes, Paul DePodesta hits the nail right on the spot. With 95 wins you stand a good chance to enter the postseason. On average, your team needed close to +130 run differential to win that many games. But what about before? When there was no wildcard? Or even before that, when there were no divisions? Let´s have a look at divisional play dating back to 1969:

American League

National League

On average you get these figures:

Even before the Wildcard, your team needed 95 wins to enter the play-offs. However, from 1969 to 1993, the run differential of all the play-off teams was a bit smaller. On average, the teams accumulated +121 runs more than they allowed. So, judging purely from wins and run differential, the wild card made things difficult for teams, but not by much.

And before that? Here are the prorated stats:

The average?

Now that is interesting: Before divisional play started in 1969, you needed 104 wins to contend. And the earlier you look, the higher the number gets (as you can see in the Prorated Wins graph). In fact, the first play-off team with no more than a 100 prorated wins were the New York Giants in 1891 - and prorated they won 99 games. The first team to win their league or division with less than 95 prorated wins were the 1904 Brooklyn Dodgers. And it's around that time, that as many teams started to pour into the play-offs with fewer than a 100 prorated wins as ones that had more than that figure.

Lets look at some graphs showing the prorated wins and prorated run differentials of all playoff teams in the history of major league baseball:

As Jim Albert and Jay Bennett show in Curveball, there was a wider spread in winning percentages in the early years of baseball. This has as much to do with the many changes in the basic structure of the major leagues given the small number of teams as well as the short schedules. It's easier to win eighty percent of 60 games than it is to win eighty percent of 162 games. And the number of wins needed for contending teams tells us the same. According to the numbers above, there is a difference of nine wins between the average needed for a playoff team before and after 1969. Today, with all the fuss about competitive balance, you need just 95 wins to be a play-off team. 50 years ago that figure was nine wins higher. You had less competitive balance back then, I´d say. As Mike points out, that also has a lot to do with the fact that before 1969 you had two playoff spots for 16 teams (12.5 percent) and today you have eight playoff spots for 30 teams (26.6 percent). And you could take this line of thought even further – because of expansion and free agency you have a different type of talent distribution today that also influences competitive balance.

Yet for the last 35 years or so, Paul DePodesta's assumption was right: By winning 95 games, you almost assure your fans of following your games into October. And history shows that you need a run differential of +120 to do so. So what happened to the 2002 Oakland Athletics? They would score 800 runs and allow 654 of them. Their run differential of +146 would secure them 103 wins, enough for the division crown and four more than the wild card team from Anaheim had. Billy Bean and Paul DePodesta, based on the right assumption, took the necessary steps.

But what about the run differential of +135 needed for 95 wins? If you look at the formula used by Bill James, you´ll see that the number of runs scored and allowed – the number of overall runs in the teams games – greatly influences the run differential needed for a certain amount of wins:

To make it more clear, lets have a look at a graph depicting the situation for 95 wins:

During the last ten seasons, the average playoff team scored 845 runs. Looking at the graph, you see the average playoff team then should allow no more than 710 runs for a run differential of exactly +135. But if you, say, allow just 588 runs, you need to score just 700 runs for a run differential of +112 to have an expected W-L record of 95-67. And if your team happens to play in Denver, you better prepare for a better run differential.

Finally just for fun, let´s have a look at teams that followed that rule but somehow managed to miss their goal. Throughout baseball history we find teams that missed the play-offs despite accumulating these 95 wins or even more.

• The last team were the 1999 Cincinnati Reds, which won 96 games and had a run differential of +154. Unfortunately for them, the Mets had a slightly smaller run differential (+142), but beat their pythagorean W-L record by two games to finish at 97-66, thus taking the wild card spot.
• The 1993 San Francisco came up empty with 103 wins, trailing division winner Atlanta by one game. Their run differential, carried among others by Barry Bonds and his .336/.458/.677 line, was +172.
•The 1954 New York Yankees also finished with 103 wins. They ended up having exactly the same run differential as the division winning Indians (+242), yet trailed the Tribe by 8 games.
• In 1942, the Brooklyn Dodgers managed to avoid postseason play by winning 104 games. That season, 106 games were needed to take you to the Fall Classic. Appropriately, their run differential of +232 trailed that of the St. Louis Cardinals (+273) by a wide margin.
• Another team had experienced the same fate as the '42 Brooklyn Dodgers, when in 1909 the Chicago Cubs finished the pennant race 6 games behind the Pittsburgh Pirates. Their difference in run differential was almost the same as in wins (+245 to +252 for -7).

How far could you take your run differential without danger of entering the play-offs? If we prorate their stats, the 1898 Baltimore Orioles put up a run differential of +326. They would finish safely in second, trailing the Boston Beaneaters by six games. The 1886 Detroit Wolverines scored a +374 (prorated), finishing 3 games behind the Chicago White Stockings. Yet those two teams were not to be compared to the 1872 Philadelphia Athletics. If we prorate their 44 games to a 162-game schedule, they outscored their opponents by a whopping 655 runs. Unfortunately for them, the Boston Red Stockings did better and it wasn´t even close. Their prorated run differential came in at +961 and it enabled them to distance the Philadelphia Athletics by 7.5 games.

On the opposite site of the spectrum, we find the 1984 Kansas City Royals. The team was outscored by 13 runs, yet managed to win AL West by 3 games. They were trailed by the Minnesota Twins, who finished with a .500 record. Yet the fact that they had a better Pythagorean W-L record than the Royals seems to have made the Twins tick. Because three years later, in 1987, they won 85 games to win the AL West two games ahead of the Royals. To add insult to injury, the Twins let themselves be outscored by 20 runs. By the way, they still beat their Pythagorean W-L record regularly by a few games. As of today, they have won 21 more games since the start of the 2001 season than they should. And of course, this season they own a 32-26 record but have been outscored by 11 runs.

(All stats from Baseball Reference and Sean Lahmans Database)