How the NL East was won... And was it baseball's worst division?

The Washington Nationals won the NL East by a massive 17 game margin.  Before we break down some specifics on run differential and how the Nats dominated MLB's weakest division (which was not the NL East) to create some separation, let's sum up the real reason that the Nats won the NL East in one sentence:

The Washington Nationals are the most talented team in the NL East.

Gee... That was easy.  Now that we know why the Nats won the NL East, let's see if we can examine how they won it.

Pythagorean W-L expectations leveled out

While the final standings show that the Nats overwhelmed the competition, the outcome wasn't always a certainty.  At the All Star Break, the Nationals led the Atlanta Braves by just percentage points, as each team was 9 games over .500.  Of course, the two teams' win-loss records didn't really tell the whole story either.  Let's have a look at the standings after July 13:

Since I decided to include it, let's start with a little something about the Pythagorean Win-Loss Expectation.  It uses run differential to determine what a team's expected win-loss record should be.  The link above will tell you a bit more about the methodology behind it.  It also includes a passage about why it shouldn't be used as the sole way to evaluate a team's true talent level.  The theory certainly does have some holes.

Perhaps the primary concern is run distribution, which the article mentions.  For an extremely small sample, we can look at the Nationals final regular season series with the Marlins.  The Nats went 3-1, winning their three games 4-0, 5-1, and 1-0.  The game that they lost was 15-7 Marlins.  The Nationals outscored the Marlins 17-16 in the series for a run differential of +1.  Their pythagorean expectation in those four games would have left them with 2.06 wins (rounded down to 2).  Since the one loss was a blowout, it skewed the run differential significantly.

Game situations can have an effect as well.  While teams always continue to try and win ballgames, there are times when they're not always doing everything that they can to win each particular game.  If a starter blows up and doesn't get out of the third inning, teams will often go to their long reliever.  This is often a player who is a swing man who isn't quite good enough to make the rotation, but has value as a long reliever who can eat some innings.  Even if that reliever gives up a run or two, their manager will occasionally stick with him for a few innings just so that he doesn't use his better relievers in a game that's out of hand.  This can tack on negatively to a team's run differential.

At any rate, I subscribe to the theory that a team's pythagorean expectations have at least some correlation with a team's true talent.  As the sample gets larger, the likelihood is that a team's actual record will trend more towards it's pythagorean record.  While I agree with the article above that 'luck' is often used as a crutch by those who put too much stock into a pythagorean record, I do believe that it's one of the variables that affects why a team's actual record differs from what their run differential says it should be.  That is the variable that will tend to even out the most over time.

Let's skip forward to what the final standings tell us happened.

With a larger sample, we see a greater correlation between their season long run differential and their actual records.

• The Nats outperformed their expected record by 1.52 wins in the second half.  They finished two wins shy on the year, which leveled out a bit from being 3 (technically 3.39) wins short of their expected record in the first half.
• The Mets were actually outscored by 8 runs in the second half, but outperformed their run differential.  In the first half, they were five games short of their expected record, so they finished closer to their season-long pythagorean expectation.
• The Braves finished one game better than their run differential over the full season after being three games better in the first half.
• The Marlins and Phillies actual records both matched their pythagorean record for the season.  The Marlins were one game above their expected record at the break.  The Phillies were even both at the break and at the end of the year.

So in the first half, the Nationals and Mets both underperformed as compared to their pythagorean expectations. Both overperformed in the second half, though they didn't quite get back to even over the full season.  The Braves overperformed in the first half.  They underperformed a bit in the second half, but stayed a hair over their expected record.

Run Differential Improvement/Decline

• The Nats had a +61 run differential in 93 games at the break, which led the National League.  They had a +70 run differential in 69 games after the break.  This was a jump from +0.66 runs per game to +1.01 runs per game.
• The Braves had a +12 run differential in 95 games at the break.  They had a -36 run differential in 67 games after the break.  This was a huge decline from +0.13 runs per game to -0.54 runs per game.
• The Mets had a +19 run differential in 95 games before the break.  They had a -8 run differential in 67 games after the break.  This was a moderate decline from +0.20 runs per game to -0.11 runs per game.
• The Marlins had a -19 run differential in 94 games before the break.  They had a -10 run differential in 68 games after the break.  This was a minor improvement from -0.20 runs per game to -0.14 runs per game.
• The Phillies had a -47 run differential in 95 games before the break.  They had a -21 run differential in 67 games after the break.  This saw a moderate improvement from -0.49 runs per game to -0.31 runs per game.

Let's explain this a bit.  Each team plays 76 games within their division (19 against each team, for 190 total divisional games), and between 32 and 34 games against each of the other two divisions in the same league* (combined 66), and 20 interleague games.  We can see that the NL West was under .500 against the NL East, the NL Central, and in interleague play.

*Two teams from each division end up playing 34 against one and 32 against the other.  It seems weird, but I would guess this has something to do with the fact that there are an odd number (5) of teams in each division.

The NL West had the worst overall record (389-421) by a wide margin.  In fact, they performed so poorly outside of their own division that every other division finished above .500.  The NL East and NL Central were both over .500 against the West, while the AL's weakest division (the AL Central) drew the NL West in interleague play and dominated.

The NL West also had the worst run differential among the six divisions.  Much like the records within each division even out to 190-190, the run differentials within the division even out (every game that the Dodgers won against the Rockies provides them with a +X and the Rockies with the same -X), so that -80 run differential came from the 430 games outside of the division.

The fact that no team in the NL East other than the Nationals finished above .500 was more due to balance than anything else.  The Marlins and Rays (77-85) finished with the best record among fourth place teams.  (The average win total for the fourth place teams was 73.2.  The Phillies and Cubs (73-89) finished with the best record among last place teams.  The average win total for a fifth place team was 69.6.  While the Mets and Braves finished tied for second place in the division, the NL East lacked a doormat.  Both the NL West and AL West had two.

You can't really judge a whole division by its top two (or even three) teams.  Because each team plays every opponent in their division 19 times, if you don't have balance, you're going to have teams boost their win totals by beating up on the lesser teams in the division.  We can see this in the NL West.

The Rockies and Diamondbacks finished with the two worst records in baseball.  Arizona finished four wins below their pythagorean expectation, but their run differential of -127 was hideous.  Colorado's case may be a bit strange, since they finished an insane nine wins below their pythagorean expectation with a run differential of -63.  The Rockies are always hard to judge though.  No team in baseball has anywhere near as pronounced a home/road split.  The Rockies finished 45-36 (+56 RD) at home this season and 21-60 (-119) on the road.

It would be expected that the Dodgers and Giants should get fat against both of these teams, right?  The Dodgers cleaned up against both of them, going a combined 28-10.  The Giants dominated Arizona (13-6), but actually lost the season series to Colorado (10-9).  That's still a combined record of 22-15 (.595) against the two worst teams in baseball.  The third place Padres would have finished tied for fourth in the NL East.

The Nats actually had two teams that they beat up on just as much as the Dodgers beat up on the two doormats in their division.  They went a combined 28-10 against the Marlins and Mets.  The difference is that the Marlins were 71-72 against everyone else and the Mets were 75-68 against everyone else.  That's one .500 team and another that probably would have been fighting for the wildcard into the final weekend.

Would it have made the division look a bit stronger had the Mets or Braves won four more games and finished 83-79?  Maybe.  Just because they didn't doesn't mean that the NL East was a weak division, though.  It was certainly better than the NL West.  It should be interesting to see how it shapes up for next season.  The Mets and Marlins both look to be on the rise.  The Phillies roster is old and expensive.  The Braves front office is in a period of transition, so it's difficult to project what to expect from them.