The Cleveland Indians are insanely hot, currently riding an 18-game winning streak after sweeping the Orioles this weekend. And the Dodgers are all but falling apart, having lost 10 in a row (and 15 of 16) after being swept by the Rockies, the worst 16-game stretch of any team in baseball this year.
As we near the midpoint of September, two of baseball's best teams are threatening to put themselves into the record books -- just for different reasons. But which unexpected run is actually a bigger surprise?
You can see the case for each side. Cleveland was always projected to be good, but "fourth-longest win streak ever" isn't something you'd expect from any club, much less one that had struggled for months to truly hit its stride. And it was only a few weeks ago that we were discussing Los Angeles in terms of possibly being the "best team ever," in part because the Dodgers had gone all year without a losing streak more than a mere three games.
In order to answer that question, we need to ask others first. Since calculating the odds of any winning (or losing) streak relies on using winning percentage as part of the mathematical input, we need to consider how we think about the "true talent" levels of these teams. Are they "really" who they are right now, streaks included? Or what they were when their streaks began? Or is their true talent level what the advanced projections say they are going forward, which is often the most effective?

It's a complicated question. As good as the Indians are, they're not going to win every game for the rest of the year, nor are the Dodgers going to lose every game. Plus, if we wanted to account for the varying talent levels both within the team (Los Angeles is more likely to win with Clayton Kershaw starting than, say, Hyun-Jin Ryu) and opponent (Cleveland has been fortunate to have had eight games against the rebuilding Tigers and White Sox), it would get really complicated.
We'll avoid those issues for now and keep this relatively simple. So let's split those questions up and run some numbers.
If their true talent is their winning percentage at the start of the streak ...
When the Dodgers started their 10-game losing streak by dropping a doubleheader to the Padres on Sept. 2, they were 92-41, a .692 winning percentage, and the odds of a team playing that well dropping 10 straight are a mere 1 in 130,164. Meanwhile, the 69-56 Indians were playing .552 ball the last time they lost before facing the Red Sox on Aug. 24, and the odds of such a team ripping off 18 in a row are a still-high but less-unbelievable 1 in 44,166.
Another way of saying that is that despite Cleveland's streak being longer, the far higher Dodgers winning percentage when this all started makes what's happened to Los Angeles much less likely -- not that this is any comfort to Dodgers fans.

The other thing about the Dodgers' streak is that it could actually be worse. Sure, 10 in a row sounds bad, but if not for a narrow 1-0 win over the Padres in Kershaw's return from the disabled list on Sept. 1, we'd actually be looking at a nearly-unbelievable 16 straight losses -- which would be tied for the longest in franchise history. While that win kept them off that particular record, it also seems unfair to act as though this bad stretch didn't really start 16 games ago, as they've won just once since.
So if we're to look back to what the Dodgers were 16 games ago, they had a .717 winning percentage (91-36). If they'd dropped that single 1-0 game and were riding 16 losses in a row, the odds of such a team losing that many in a row would have been -- deep breath -- 1 in 590,768,094. It didn't happen. By a single run.
If their true talent is their winning percentage now …
Of course, you can't toss out the streaks as mere luck and act like they haven't added new information. The Dodgers, for example, are obviously better than they've been playing, yet concerns about Corey Seager's right elbow, Alex Wood's velocity, and the hitting struggles of Yasmani Grandal, John Forsythe and Curtis Granderson are real issues headed into October. If the team isn't this bad, it was probably never as good as it seemed at its peak, either.
So if what these teams are now is closer to their true talent level, then what we have is a Cleveland team that upped its winning percentage from .552 to .608, cutting the odds of an 18-game streak from 1 in 44,166 to 1 in 7,757, or about seven times as likely based on the big improvement in winning percentage.
Since the Dodgers (.643) now look worse than they did both when the 10-game streak started (.692) and the 15-of-16 skid did (.717), their odds are cut, too, but it's still more shocking than the Tribe's win streak. A .643 team would have had a 1 in 29,737 shot of losing 10 in a row -- and a 1 in 14,365,096 chance of losing 16 in a row.

If their true talent is their projected record from here on out ...
There's another way to look at this, and that's in advanced projections. That is, by looking at each player's expected performance and expected playing time, you can project how well a team is expected to play. You may remember we did exactly this in January, raising some eyes by saying the Dodgers already projected to be as good or better than the Cubs, before they had a second baseman, Yu Darvish or knew anything about Chris Taylor. At the time, Cubs fans revolted, but projections are often worthwhile ways of taking emotions out of the equation and looking ahead.
This is likely to be controversial, but it won't be: Cleveland will lose games before the season ends. Los Angeles will win games, several of them. Based on the past two weeks, these are the best and worst teams in baseball history, and yet you can't simply throw out the information we learned over the first four-plus months of the season.
Taking all of that into account -- the elite Dodgers playing awfully, and the good Indians playing unbelievably -- the current projections, which don't factor in opponent strength, see both clubs as being about equal the rest of the year. Cleveland is projected with a .597 winning percentage, well above the .552 it was at when it started its run; Los Angeles is projected at .593, well below the .717 or .692 it was at for the start of its pair of terrible stretches.
Those numbers may be the best expression of their true talent, and it's worth noting they're not dissimilar to the .597 (Dodgers) and .577 (Indians) projected numbers we saw all the way back in April. If so, Cleveland's run of 18 wins from a projected .597 team would be 1 in 10,775, and for once that's less likely than the 1 in 8,017 odds of a .593 Dodgers team losing 10 in a row.
Again, though, that single 1-0 Dodgers win changes everything. If not for that, the odds of a .593 team losing 16 in a row are only 1 in 763,962. Or perhaps that ought to be "only," really. No matter which numbers you use, these streaks are nearly unprecedented. The only thing we know for sure is that they won't last forever. Probably.