At the end of last week, Ben Carsley gave insight into Baseball Prospectus’ projection for the Red Sox, which, derived from PECOTA, has the team going 87-75 and claiming top spot in the American League East. But BP isn’t the only projection system in town. For example, the folks at FanGraphs have the Red Sox pegged for an AL East winning 91-71 record; slightly more favourable than BP’s projection. The FanGraphs projection comes from the Steamer and ZiPS systems, which, given the four-win difference at the team level, must differ from PECOTA for at least a few of the players. As such, I explored which Red Sox players have the largest discrepancies between the systems.

In order to make fair comparisons between systems, it is necessary to evaluate them for a common playing time. While it is certainly interesting if the projection systems expect different amounts of playing time for a given player, that difference will be reflected in their counting statistics (e.g., wins above replacement (player) [WAR(P)]). This can make it appear as though there are large differences in expected performance, when there actually are not (other than accumulating playing time). As such, I think lining things up makes comparisons between projection systems a little cleaner. With this in mind, for what you see below the hitting projections are scaled to be per-600 plate appearances (per-450 for catchers), and the pitching projections are scaled to be per-200 innings pitched for starters and per-65 innings pitched for relievers. These playing time figures can be debated, but they are intended to represent a season’s worth of time and therefore a decent baseline.

We’ll start on the hitting side. In most cases the three systems project the players in very similar ways (i.e., within one WAR(P)), but here are the Red Sox players (who will be regulars on the big league squad) for whom there is at least a one-win difference:

It turns out the differences for these players are largely driven by how the systems view the players’ defense. This makes assessing the differences considerably more difficult, as the defensive metrics that go into these systems can vary wildly in how they evaluate actual on-field performance (e.g., Dustin Pedroia last season), let alone anticipated performance. For example, all three systems expect Xander Bogaerts to be an above-average hitting-shortstop to similar degrees, but while Steamer and ZiPS project him to be a slightly below average defender (-1.8 and -2.5 fielding runs, respectively), PECOTA is really down on his defense, projecting -8.8 fielding runs above average (FRAA). The opposite situation explains the variation in Andrew Benintendi’s projections: PECOTA essentially views him as a center-fielder playing left-field, and therefore has him contributing 17 FRAA, but ZiPS, which also likes the outlook for his defense, only has him at 5.8, and Steamer projects him to be slightly below average (-0.7). Oddly enough, Jackie Bradley Jr., the team’s actual center fielder, is given the worst defensive projection by PECOTA (1.0 FRAA), while Steamer (7.1) and ZiPS (4.9) expect much better. Finally, nobody expects Sandy Leon to continue his Babe Ruth act from 2016, but only PECOTA projects his defense to be below average.

Obviously it remains to be seen which system will project the Red Sox’s (and other teams’) hitters most accurately, but in the meantime we can get an idea of the range of possibilities by looking at the average projection, and the best (highest) and worst (lowest) cases. Here is that exact thing for what I expect to be the everyday lineup (last year’s performance is included as a reference point):

There is almost a 10-win swing between the best and worst case projections, just in the everyday nine. Getting 25 wins above replacement from this group would be great. Keep in mind that a replacement level team (i.e., 0.0 WAR(P)) is expected to win around 52 games, so the best case scenario from the everyday lineup already gets the Red Sox in the neighbourhood of a .500 record. And that is without considering the bench, or the pitching.

With that said, we should consider the pitching side of things. The pitching comparisons are especially tricky due to the fact that the WAR metrics are calculated quite differently between FanGraphs (FIP-based) and BP (DRA-based). Regardless, here are the Red Sox hurlers for whom there is at least a 1-win difference:

The rotation! While to this point I have been focused on relative differences, it is worth noting that these projections are strong. The worst of the group is Eduardo Rodriguez’s at 1.8 WARP, which is fine. For the most part the differences are between PECOTA and ZiPS. As you can see, for each player ZiPS is the high-man, with PECOTA as the low-man and Steamer in the middle. This pattern is a result of the way the metrics are calculated, and makes examining the differences hard. With the exception of Fernando Abad, these three projection systems have very similar expectations (i.e., < 0.5 win difference) for the probable Red Sox relievers. But as the table above shows, when it comes to the rotation, Red Sox fans should be rooting for the ZiPS projections to best represent what will happen this coming season.

In the end, barring major injury – I am not including David Price in that category yet – the Red Sox appear to have a clear path to the division crown. But as is evident from this little exercise, the range of outcomes is considerable. We can keep this all in mind over the next six months, but regardless of which system turns out to be most accurate, it will be nice to have the players out on the diamond playing again.

Photo by Steve Mitchell – USA TODAY Sports

While so much focus is put on them before the season starts when we’re starving for baseball, we all sort of forget about them once the games start. Projections are not something that most people continuously go back to look at, myself included. However, considering how disappointing this Red Sox season was, I thought it would be worth it to look back at what PECOTA said about them before the season to see what the system got right and what it got wrong.

Before we dive into the individual player projections, let’s note that PECOTA was quite high on the overall team back in April. The day before the season started, Boston was projected to win 88 games, enough to win the East and give them the second best record in the American League. Of course, PECOTA wasn’t the only entity that was high on the Red Sox before the season. Which specific players did it get wrong that swung the season in such a poor direction?

One last qualifier before we begin: I only used players that both were projected to play something close to a full-time role and actually played something close to a full-time role. So, guys like Eduardo Rodriguez, Blake Swihart and Henry Owens aren’t included as PECOTA only projected them for a handful of starts/games played. Now, onto the offensive players!

Starting at the top, it’s clear that the two star young players in Boston’s lineup outproduced what PECOTA foresaw back in April. The strange thing about Bogaerts’ season is, despite the huge strides we all saw him make at the plate, his True Average was actually worse than what was projected for him before the season. Now, if we were to break this down by all of the triple-slash components, we would see some major differences. This, of course, is because of how average-heavy his 2015 was, as I talked about a few weeks ago. The huge difference for Bogaerts this season has been his defense. Coming into the year, we figured that he would be an average-at-best shortstop, and even that may have seemed optimistic. He outperformed every reasonable expectation, and that’s where we see the big increase in WARP.

If you go back in time to April, there was no player’s projection who caused more of a stir than Mookie Betts. Many thought he was just another overrated Red Sox player, and there was no way he was going to hit as well as the projection systems (PECOTA certainly wasn’t the only one so high on him) said he would. Well, after all that, he wound up out-performing the projections. Even I, who was expecting a solid season from Betts, couldn’t have predicted him to blow PECOTA out of the water. Beyond his offense, which was slightly better than expected, his defense in center field was tremendous. So, that’s an extra three wins from the two young players. If you told me that would happen before the season started, I would’ve thought there’d be no way the Red Sox didn’t make the playoffs.

Of course, now we get to the two veteran free agents, who were also two of the most disappointing players in the league. It’s not a huge surprise that PECOTA was so far off on these guys, as basically everyone in the world was off. In fact, it was the biggest reason the Red Sox were so much worse than projected, as these two players caused a negative-eight game swing. Ramirez in particular was a disaster, as PECOTA saw him as both a great hitter and an average defensive player.

The two veteran mainstays in the organization had some lofty projections coming into the year, and while they both had good seasons, they both fell short of what was projected. Pedroia’s projection was particularly off, but that’s just a playing time issue as he missed a good chunk of time with injury.

Finally, we have All-Star Brock Holt, who obviously outproduced what was expected from him before the year. That’s not a shocking revelation, but he deserves a mention for being the only good thing about the first half. Oddly enough, his second-half was poor enough that playing time was really the only difference between his actual production and his PECOTA projection.

Now, to the pitchers!

Oddly enough, all of the starters who threw at least 100 innings outproduced their PECOTA projections, although most of them were coming from extremely low starting points. If you remember correctly, PECOTA was a harsher critic of Boston’s rotation than any other one. It’s worth noting that these actual WARPs are based on FRA, not the new DRA model, as DRA hadn’t been released at the time PECOTA came out.

Porcello’s actual performance would have been a lot uglier had it not been for a strong run to end the season. In the end, he was only a little worse than everyone expected, although he did it in a completely different fashion. Instead of being a ground-ball machine, he relied on strikeouts and a lack of walks for his success. It’s a good strategy, but it’s tough to succeed when you give up more than a home run every nine innings.

Next we have Joe Kelly, who PECOTA hated more than just about any pitcher who was set to get a significant amount of playing time. For much of the year, they were right, but he also had a solid run to end his season. Kelly, to his credit, was able to induce more strikeouts than PECOTA could have ever predicted.

Wade Miley’s projection vs. actual production is easily the most confusing on this list. PECOTA pretty much nailed his season, as they weren’t significantly off in any area. However, the 1.5 difference in WARP is anything but insignificant. While his projected line represented a replacement-level player, it actually ended up being worth more than a win.

Finally, we have Clay Buchhoz, who is just about the hardest player to project every season. PECOTA saw him being an average pitcher, which any sentient being could have seen as being a mistake. Buchholz is never average. It’s either way above-average or way below-average. This season, he was outstanding, of course. His 2.5 WARP was far and away the highest in the rotation despite him throwing just 113 innings before injury.

For as bad as the rotation was all year, it really wasn’t the reason Boston played below their expectations. The main culprits were the bullpen and the veterans. Whereas previous disappointing Red Sox seasons happened because too much faith was put into young players, it was the old ones who failed this time.

Photo by Winslow Towson/USA Today Sports Images