Why projection models are ‘low’ on the 2017 Phillies
Since the 2016 season closed, Matt Klentak has been busy upgrading and fine-tuning the Phillies roster. Poor corner outfield performances have been replaced by veterans Howie Kendrick and Michael Saunders, a bullpen that crashed and burned down the stretch has been renovated to include Pat Neshek and Joaquin Benoit, and the heartbreakingly outmatched Ryan Howard is gone, to name a few.
What these upgrades add up to, per PECOTA, is two wins.
That would bring the Phillies to a 73-89 record in 2017.
Why such a small gain?
Two wins seem like a scant gain, especially given the tweaks, upgrades and player development fans anticipate taking shape in the 2017 team. Some are already remarking that the team might vie for a .500 winning percentage. Others are even questioning what the team will do if they stumble into the wild card hunt. If the team is this much better than they were last March, how can it only manifest as two more wins and not, say, five or ten?
Part of the problem is assuming the Phillies were a 71-win team last year and not just a team that won 71 games.
By run differential, it can be argued that the Phillies should have been a 62-100 team last year. This is a Pythagorean winning percentage, which is one way of dividing luck from performance when reflecting upon a team’s record.
I know, I know: what happens on the field is more important than what happens in our supercomputers. Got it.
But PECOTA doesn’t care about last year’s record. It doesn’t start with a 71-win team and add up the improvements they’ve made. PECOTA is almost as cold and heartless as I am, and its projected records are based solely on how players will fare in the year ahead, adding them up into a team-level outcome.
The point is that starting with this building block – what the models say “should have” happened last year – makes the picture more in line with the expectations of many fans. It suggests the team will be 11 wins better than the talent level of their abysmal 2016 offense and inconsistent pitching were.
Where does 73-89 leave the Phillies?
At 73-89, PECOTA tells us, the Phillies would be baseball’s third-worst team.
They will have been passed by the Twins, Reds, Rays, Braves, A’s and Diamondbacks, all of whom were beneath them in last year’s standings. Falling behind the Braves again would hurt the most, I think, especially as both teams endure parallel rebuilds and might gear up for playoff pushes at the same time in future seasons.
Only the Padres and Royals would have worse seasons. With the way writers have presented the current Padres roster, the team could decide to just not show up in 2017 and no one would blame them (or be there to notice). The Phillies don’t want to think of themselves as being in the company of a team with such low expectations.
Are teams passing the Phils by?
With the club looking tuned up in key places going into next year, the expectation should be moving up the standings, not down them. What gives?
While it’s possible that some of these teams will improve more than the Phillies have – the Braves, for example, are getting a lot of attention for the young talent they’ve added in the last year – another explanation is that the Phillies were worse than a lot of these teams in 2016, too, and it just didn’t show in the standings.
PECOTA has three means of projecting the record a team “should have” earned, two of which agree that the Phillies had the talent level of a team with a .376 winning percentage in 2016, worth 60 or 61 wins.
The Phillies won many more than 60 games, including starting the season 14-10 in the month of April. They were touching the top of the NL East for a few days, even. After April, though, the story changes.
While April was a positive month for the club, it was also the outlier; in no other month was the team .500 or better, and few would argue that they were an above-.500 team last year.
Taking away the 24 games they played that month – the 14 wins and 10 losses – the team was 57-81. That is, their winning percentage was exactly .375, matching what PECOTA describes as fitting their actual performance. This, being a larger sample size than a single month early in the season, is more likely to be representative of the team’s real merits and shortcomings.
It is possible that some of these other teams made more improvements than the Phillies did. It’s also possible that the Phillies will move ahead of other teams they were behind last season. PECOTA would has us describe the team’s mobility as going from baseball’s worst team to its third-worst, not starting as the eighth-worst and plummeting further.
In that sense, PECOTA, too, agrees that the Phillies have improved— quite a bit, in fact.
The good news: baseball isn’t a spreadsheet
The response to this methodology is easier to predict than a season of baseball.
But I’m not trying to take away any of the accomplishments (for lack of a better term) of the 2016 Phillies or say we’re living in a version of the Matrix where the “real world” is the one described by the computers spitting out projections for us year after year.
The point of this exercise – just like the point of projections to begin with – is to consider more ways of looking at and understanding things like player performance, talent, and even luck and random fluctuation.
There’s good news in the fact that even PECOTA, which was so down on the Phillies as a team last year, confirms that they should be expected to far exceed that performance.
There’s even better news in the fact that projections aren’t a crystal ball. They determine probabilities, not fate, and how the Phillies do in the year ahead is yet to be determined. They could, just as they did last year, have stretches where they look like a much better team than anyone thought, perhaps even passing 73 wins.
Data is important for how it helps us understand the game and even find ways to try to predict future performance. It is not as important, as many will point out, as what happens on the field.
