Going into the season, I thought the Cleveland Indians would be better than the Boston Red Sox. Not a lot better, a little better. But better.
My opinion on this matter was based on numbers. and it should be acknowledged that numbers, even carefully chosen ones, aren't always right. That's especially true when you're using numbers to forecast things that haven't happened, as opposed to making sense of things that have already occurred. Every year I create an objective forecast of the coming baseball season. Every year, there are hits and misses. When it comes to a comparison of the Indians and Red Sox, that was clearly a miss.
The thing is, after the Indians beat the Red Sox on Saturday, Cleveland sits 18½ games behind Boston in the American League standings. That's fine. If I were to do a comparison of every possible combination of two teams in the majors, I could easily find bigger misses. Heck, I'm 20 games off of the projected Baltimore Orioles against the reality Baltimore Orioles, because even though I thought they'd be bad (74 wins), it was impossible to foresee them being a team that might win fewer than 50 games.
What's disconcerting about the Red Sox-versus-Indians miss is that despite what the standings show, I'm still not entirely convinced that I'd pick Boston in a postseason matchup between the two teams. We could get that matchup in a few weeks, in the American League Championship Series, so maybe it would be instructive to look at some questions regarding these teams. Would a Boston-Cleveland matchup in October be more even than the standings suggest?
1. Why did MLBPET think the Indians would be better?
MLBPET, if you've only just tuned in, is my system for projecting, evaluating and tracking big league baseball. It's not a very good name and I wish I could think of another, but I'm kind of used to it by now. Besides, when I started building what is ostensibly a system of interconnected spreadsheets, I didn't really know I'd be sharing its output with anybody in the outside world.
As I mentioned, the ratings were close. In the previous run of simulations I performed before Opening Day, the Indians won an average of 93.5 games, the fourth-best total in the majors. Boston averaged 92.3 wins, ranking sixth. Right now, the Indians look like they are going to come up a little short of that forecast, with current simulations of the last week of the season putting Cleveland at an average 89.6 wins. That's worse than five other AL teams, including Tampa Bay, and roughly equal to Seattle. Boston, on the other hand, has blown its preseason forecast out of the water. I've got the Red Sox winning an average of 110.2 games in current simulations.
While in-season simulations fold in as much data from the actual schedule as possible, in addition to those initial forecasts, the preseason forecasts are based only on two factors: The rating of each team's roster, and how that roster fares in the schedule the team is preparing to play. Simply put, MLBPET thought the Indians had a better roster. For that matter, it still does. The roster rating MLBPET assigns to each team, based on its current depth chart, has the Indians at 97.8, ranking third in baseball. Boston is at 95.6, ranking fifth. This, as much as anything, is why I'm still not sure I'd pick the Red Sox to win a best-of-seven series against Cleveland.
Perhaps that is best illustrated by a position-by-position comparison of how the system would forecast the teams today if we were starting a new season. I'm going to list the forecasted WAR value for each position, along with the MLB rank. The thing to remember is that these are projected numbers that included 2018 results as part of their formula, not actual 2018 results.
First base: Indians, 1.9 WAR (13th) vs. Red Sox, 1.7 WAR (21st). This is mostly Yonder Alonso, with a little Edwin Encarnacion, ranking ahead of a combination of Mitch Moreland and Steve Pearce.
Second base: Indians, 4.6 WAR (2nd) vs. Red Sox, 2.6 WAR (9th). Remember, the projections are based on current depth charts and right now, I've got about two-thirds of Cleveland's playing time at second assigned to Jose Ramirez. The rankings for both teams at this position are markedly better than their actual 2018 results because of in-season upgrades.
Shortstop: Indians, 6.0 WAR (1st) vs. Red Sox, 3.3 WAR (9th). Boston's Xander Bogaerts is a very good player having a very good season, but let's face it, even the hardiest of Red Sox fans would trade him straight up for Francisco Lindor without blinking.
Third base: Indians, 5.3 WAR (1st) vs. Red Sox, 2.1 WAR (17th). Josh Donaldson still has enough in his track record to swamp Rafael Devers in this comparison.
Catcher: Indians, 2.3 WAR (13th) vs. Red Sox, 1.4 WAR (21st). Not the sexiest part of this comparison, but MLBPET sees both Indians backstops (Roberto Perez and Yan Gomes) as better than anybody on the Red Sox.
Left field: Red Sox, 3.9 WAR (3rd) vs. Indians, 3.0 WAR (10th). Andrew Benintendi over Michael Brantley, with some other guys sprinkled in.
Center field: Red Sox, 3.8 WAR (5th) vs. Indians, 0.9 WAR (24th). Jackie Bradley Jr. is a clear winner over the hob-glob of options the Indians have at this position.
Right field: Red Sox, 6.5 WAR (1st) vs. Indians, 1.7 WAR (17th). Mookie.
Designated hitter: Red Sox, 3.2 WAR (3rd) vs. Indians, 2.4 WAR (7th). I'm not sure the projections are capturing the true difference between what is mostly a comparison between J.D. Martinez and Encarnacion.
Rotation: Indians, 19.6 WAR (2nd) vs. Red Sox, 16.8 WAR (5th). Chris Sale may be ahead of any of the Cleveland starters, but Corey Kluber, Trevor Bauer and Carlos Carrasco all live in his neighborhood.
Bullpen: Red Sox, 5.03 WAR (5th) vs. Indians, 5.02 WAR (7th). Added an extra decimal here to show that there is a difference, but it's basically a dead heat. My formula for bullpen performance incorporates anticipated usage by leverage situation. Boston has a big edge in high-leverage spots thanks to Craig Kimbrel. The Indians have more and better middle-leverage options, even if many of those pitchers have underperformed in 2018.
To sum up, the Red Sox only really have a decisive edge in center and right field. That's why MLBPET likes the Cleveland roster a little better, both now and the way they stacked up entering the season. I'm not the only one that sees it like this, either. Check out the rest-of-season projections at FanGraphs, and look at the winning percentages under the projected numbers. They see the clubs as virtually dead-even.
Of course, this comparison is more about why I'm asking the key question at hand -- why are the Red Sox 20 games better? -- than about actually answering it.
So ...
2. Have the Red Sox overachieved?
Any time you're talking about a team destined to win somewhere in the neighborhood of 110 games, of course they have overachieved. Only six teams have ever won that many games in a season.
But that's not exactly what we're talking about, is it? One way to look at overachieving is to consider whether a team has an unusual number of players enjoy career seasons. Let's look at a comparison of how the players on the Red Sox's current depth chart are performing against expectation, with current numbers prorated for the full season.
Boston's roster has performed a little better than expected, but it's nothing crazy. Mookie Betts may not truly be a 10-win player -- no one is, really -- but it's also entirely possible that as he approaches the age of 26 Betts' game has gone to a new level that he'll occupy for the next few years. That's not overachieving, that's improving.
The other facet of overachieving is plain, old luck. Is Boston's win total exaggerated by sheer random chance? To a certain extent, yes. The Red Sox are on pace to win those 110 games, but their run differential is closer to that of a 103-win team. Still awfully good, but that win total doesn't break franchise records, nor does it climb into that all-time win leaderboard listed above.
The Red Sox have gone 24-12 in one-run games this season, and 42-19 in games decided by one or two runs. With Kimbrel at the back of the bullpen, you might expect Boston to be strong in those categories, but not to that extent. Based on Boston's overall run differential, you'd expect the Red Sox to have won about six fewer one- and two-run games than they actually have.
No matter how you slice it, Boston has played great baseball this season, and if they are playing over their heads, they aren't doing it by much. However, they are probably in reality more like a 100-win team than a 110-win team.
3. Have the Indians underachieved?
I'm going to skip the chart on the Indians' performance vs. their expectation because it says even less than Boston's does. Basically, the players on the Indians' roster have together been about 1.5 WAR less effective than forecasted. Let's move onto the luck portion of the exercise.
The Indians have outscored opponents by 154 runs this season, which over 162 games translates to a win total of 97 games. Cleveland is actually on pace to win 90, and they're getting 89 in my simulations of the remaining schedule. Only four teams have a worse luck factor by this measure -- the Dodgers, Orioles, Nationals and Astros.
There are a couple of reasons for this. First, we return to the topic of close games. Cleveland is 19-22 in one-run games and 13-14 in games decided by two runs, or 32-36 combined. But based on the Indians' overall run differential, you'd expect them to have won about 39 of their one- and two-run decisions, not 32.
When you combine the Boston's six-win surplus in close wins and the Indians' seven-win deficit, you've made up much of the ground in the nearly 20-game gap in their overall records. Things are a little more clear.
I should mention that there is another component to this that might suggest Cleveland's run differential has been slightly skewed by blowout wins. The Indians have played in an MLB-high nine games decided by 10 or more runs this season. They've won eight of them, two more than any other team in baseball. By contrast, the Red Sox have gone 3-1 in such extreme routs.
That may be more of an oddity than anything. Cleveland is just 29-17 in games decided by five or more runs, a mark that includes those double-digit affairs. That's solid, but consider that in five-run games Boston is 36-15. Team strength is better reflected in easy wins than the close ones, and Boston has had more of those. Still, I'd state Cleveland's baseline talent as being more like a 95-win team than a 90-win team.
4. What have we learned?
In terms of baseline talent, I don't think there is a huge difference between the Indians and Red Sox, an opinion reinforced by the going-forward projections. There has been a little bit of luck baked into each team's record, but Boston has certainly outplayed Cleveland by a comfortable margin this season, especially if you factor in that the Indians' AL Central schedule has given them the majors' easiest slate of games.
Given what we've seen over the past few months, I wouldn't make the claim that the 2018 Indians are better than the 2018 Red Sox. However, I am comfortable in stating that the gap between them is considerably less than what the standings suggest. If we do end up with a Boston-Cleveland ALCS, I remain unsure which team I'd pick. Given that the Red Sox are nearly 20 games ahead of Cleveland, Indians fans should take some solace in that.
On the other hand, it's certainly possible that the projections are not yet picking up on what is a widening gap in the talent level of the teams. After all, in Betts, Bradley, Bogaerts, Benintendi and Devers, the Red Sox have a lot of high-level talents whose best days may lie ahead of them. That, Boston fans, is where your solace can be found.