Formulating an agrument

I could talk about the pitching but hey why not try to enjoy myself during this terrible downswing and go with something I know Ill love : picking on Boz.

Yesterday in his chat (still optimistic, the team hasn't quite beaten it out of him) he included this weird factoid.

Since 1983, 14 teams have won their division by 15 games or more. Five — more than a third — failed to repeat, including teams that had demolished their divisions by 21 ¹/₂ and 20 games.

The idea he was trying to get across was pretty straightforward. Even teams that seemingly are head and shoulders above their divisional competition can find themselves in 2nd place just a year later. But how he formulates the argument makes me question the whole point.

First off he uses two arbitrary endpoints in the argument. The first, "15 games or more", makes some sense. He wants to include the 2014 Nats, who won their division by 17 games, and dropping it down to 15 allows for more teams to be included (more data points) and gives some needed flexibility to the definition of dominant division winners. It's a perfectly reasonable choice. The second though, "since 1983", I can't figure out why he chose. You read enough about baseball and you get familiar with the years usually used as endpoints used for factoids. Unless you want to use 100+ years of questionable comparison, you pick division expansion (1969, 1994), years of team expansion (1969, 1977, 1993, 1998), other notable years (1946 post-war, strikes in 1981, 1994,  DH in 1973) to subset your analysis. But 1983? What is that measuring? Last time the Orioles were in the World Series? Only All-Star game grand slam?

Either you pick 1983 because it serves your point or you pick it because the data tells you to. Already the latter is questionable because since you are dealing with a division question you should probably only go as far back as these divisions existed. But if there were a big gap prior to 1983 of 15+ game leads but constant type leads since you could reasonably use 1983 as a stopping point. Well what does the data tell us? What years do we find these 15+ game leads? We'll use since 1969 since that is when divisions similar to 1983 started.

1969, 1970, 1971, 1975, 1983, 1984, 1986, 1988, 1995 (2), 1998 (2), 1999, 2002, 2003, 2008, 2011, 2014

Hmmm. The biggest gap is between 1975 and 1983, but the gap between 1988 and 1995 is almost as large. I just don't see the reasoning of looking before 1994**

OK for our sake let's restrict analysis from 1994 on since that makes sense. How many out of 9 (can't count Nats yet) division winners ended up losing their division the next year? 2. The 1999 Indians and the 2003 Giants.

Now here's another thing. There are multiple ways to be 15 games better than the team behind you. You can be great and they can be mediocre, or you can be very good with some luck and they can be good but with bad luck. Either can work. The Nats were great (Pythag 97 wins) and the competition was mediocre (Mets 2nd best with 82 pythag wins). That's the scenario we want to see. The 2003 Giants don't really fit that bill. They were more very good and very lucky turning a 93 win pythag team into a 100 win team. Talent wise they were more 10 games better than the Dodgers, which is a lot, but not enough for our purposes here. So we're left with the 1999 Indians. Over 15 games better in pythag than the next best team in the division. What happened to them in 2000?

It IS Boz's argument come to life. The Indians, a couple games lucky the previous year, got a couple games unlucky the year after. A division rival, in this case the White Sox had one of those years. Nearly everything went right. Only one hitter dropped while several got better, trades worked out, the entire pitching staff took a step forward. It's quite amazing to see. In the end they were just good enough, with everything going their way, to pass the Indians. It only lasted for a year, only happened to one team truly 15 games better than their opponents in 21 years, but this type of magical season does happen. This is what Boz was worried about at seasons start. It's what we worried about (albiet more with injuries costing the Nats a few games than getting unlucky)

Too bad that isn't what's happening here.

The Nats aren't facing a Mets team who is having everything go right. The Mets are playing like an 86-87 win team. A natural progression from the .500ish win team they were last year. They didn't catch the Nats with magic. The Nats have utterly collapsed. They are under .500 for god's sake. Has that happened before to anyone?

Well we see it didn't happen since 1994, so let's go back through those 8 times from 69 to 93. The last four times it happened the teams did in fact fail to win the division the following year. Were any of these complete collapses?

The 1983 White Sox. Totally. They were a good team in 1982 who made it all work in 83 only to come crashing down in 84. The starting pitching saw three guys pretty much turn bad at the same time, while the hitting kind of took a combined step back along with their DH, Greg Luzinski, hitting the wall.They weren't quite the Nats talent level though, imo.

The 1984 Tigers? I don't know if I'd call it a collapse. They continued to play well in 1985. There were also not 15 games better than anyone by Pythag, the AL East being very competitive in the mid 80s. The 1984 team was more of an aberration to me. One of two standout years in a sea of decency that lasted from 1978-1988, powered by 2 things that are ultimately very variable - a great bench and a great pen.

The Mets in 1986 and 88 present the last two examples and are probably fueling Boz's comments. In both 1987 and 1989 they were still good but just passed. In 1987 Ron Darling became the average arm he'd be for the rest of his career, while Ojeda got injured and the bullpen stepped back. As you could guess that wasn't enough to make them bad but the Cardinals watched Jack Clark have a monster year and Coleman and Pendleton go from bad to good. In 89 they got a bit unlucky. Carter and Gooden got hurt. Cone wasn't awesome. Meanwhile the Cubs saw Grace develop and Sandberg bounce back, got a career year from a rookie named Dwight Smith, got Mike Bilecki's only quality year as a starter and Rick Sutcliffe's last good season...

Where does this leave us? Well Boz's initial statement, that 15+ winners can easily lose their lead the next year is probably a bit overblown. While it happened 4 out of 8 times from 69-94 it's only happened 2 out of 9 since then with the divisions set up as they currently are. Plus when you try to factor luck out of the records it even gets less likely in the initial pass of the data* because a couple of those teams drop "true 15 games better" status.

However unlikely doesn't mean impossible as we see. Boz (and really all of us - we did say it was possible) is right in a more macro sense, despite the terrible way he formulates the argument and his wrong facts and off conclusions. It's hard to play high 90s good and when you drop back into the low 90s, a little bit of bad luck for you, a little bit of good luck for them, and you can totally be passed.  In the biggest lesson to the current Nats, the late 80s Mets only won two divisions when they really could have won six divisions in a row. They were that talented, but luck/timing/heart/etc wasn't on their side.

To me looking at that team - the late 80s Mets is most informative. We've thought that "being good for a while, win your division a few times, make playoffs. then hope" was the best strategy you could put together for championships. But unless you can put together a worthy high 90s win team (which we did think the Nats could have been this year, to be fair) over and over again the playoffs aren't guaranteed. Not even close. So many little things can happen that can keep you out that playing for it all when you have the chance... it just makes more and more sense to me.

The other thing I take away is that what the Nats are doing right now, this total collapse into mediocrity? It's crazy. While not unprecedented it really only happened once before with that 1984 White Sox team and it's tough to argue they had the Nats talent. Still the Nats have some time to get back and either win the division or find themselves a more normal path to missing the playoffs, being a good team that's just not good enough by a game or two. Otherwise this Nats team will go down as something historic. An incredibly talented squad with limited competition who completely imploded and lost the division in a single season.


*to be complete we need to go back through the years to see who might have been 15 games better by pythag but won their division by fewer than that and see how they did the following year.

**At that point we have more divisions with fewer teams in them. We should see more 15+ game leads during this time. Is that the case? Well if we see 8 such times in 25 years from 1969 to 1993 with 2 divisions per league we should see 12 in the next 25 just by division increases alone... but it hasn't been 25 years yet (21) so scale that back to 10... and we see 10!  That's actually interesting because I would expect the decrease in number of teams in the divisions to lead to more instances. It's probably something that can't be seen with this analysis - probably just shows up as greater average divisional leads.