What the )&*^(* Happened?
Through August 5th, the Phillies were 63-48, in first place in the NL East, with the second best record in the NL.
Had they continued at that pace, they would have won 29 of their final 51 games. Instead, they won only 17 of 51.
How did 12 games turn from wins to losses? Breaking it down by phase of the game, here is one way we can apportion the blame for The Collapse:
- 4 losses can be attributed to the offense — they scored fewer runs, from 4.42 per game, to 3.65 (0.78 runs per game worse)
- 6 losses from pitching/defense — they gave up lots more runs, from 4.09 per game to 5.37 (1.28 per game worse)
- 2 losses because the success (or “luck”) they had been having in one-run games regressed back to normal
More specifically, when comparing the first 111 games to the final 51, what changed?
The Offense was 40 runs worse than previous trends (i.e. 0.77 per game for 51 games). By position:
15 runs worse in CF (primarily Herrera): from .277/.334/.458 (.792 OPS) to .199/.259/.323 (.581)
11 runs worse in LF (primarily Hoskins): from .261/.370/.518 (.888) to .211/.299/.437 (.736)
7 runs worse at 2B (Hernandez): from .261/.367/.368 (.736) to .231/.321/.338 (.660)
7 runs worse at 3B (Franco et al): from .261/.319/.462 (.781) to .220/.292/.382 (.675)
7 runs better at SS (Kingery/Cabrera): from .223/.272/.344 (.616) to .261/.303/.429 (.732)
6 runs worse at all others combined
40 runs worse in total
The “3 H’s” that were so instrumental to their early success were much worse down the stretch. Odubel and Cesar had been slumping for a while already, but nevertheless, a big part of the reason the first part was good was that it included their hot streaks.
The July-August acquisitions made up for some of those declines, but not nearly enough.
Also included above but not quantified is the impact of injuries. Other than JP Crawford, the team was lucky to be fairly healthy for most of the season, but as the season wore on, nagging injuries plagued Franco, Williams, and possibly Hernandez down the stretch.
Run prevention was 65 runs worse (i.e. 1.28 per game for 51 games). First, to simply split the additional runs allowed by who was on the mound:
+14 runs Velasquez: from 4.15 runs allowed per 9 through Aug. 5th, to 8.45 RA/9 after
+12 runs Eflin: from 3.93 RA/9 through Aug. 5, to 6.65 after
+4 runs all other starters
+35 runs Bullpen: from 3.99 RA/9 through Aug. 5, to 5.76 after
+65 runs Total
As for splitting this by pitching vs. fielding vs. luck/timing, we can make some educated guesses:
20 of the 65 runs can be attributed with some certainty to pitching, based on the change in FIP, and most of that (~15 of the 20), can be assigned to Nola/Arrieta/Pivetta/Velasquez (Eflin’s FIP improved slightly).
According to the advanced fielding metrics, the defense did get worse from August 6th on, by something in the range of 10 to 27 runs. Let’s call it 20-25 from fielding.
That leaves 20-25 runs that are tougher to categorize. We know that three things went the wrong way from August 6th on:
- While the expected BABIP allowed, based on batted ball profiles, remained about the same (.286), actual BABIP shot up to .327, meaning that given how hard opponents were hitting the ball, many more hits than would be expected were falling in.
- wOBA against through August 5th was .300, slightly below what would be expected based on batted balls (.305), but it was much higher than expected after August 6th (.339 vs. .312).
- The Phils’ strand rate went from league-average through August 5th (73.2%), to much lower after August 6th (66.8%), meaning that the hits that were given up were more bunched than they had been, allowing more opposing runners to score.
The Phillies finished with the 10th best record in the NL, and both runs scored and runs allowed ranked 11th.
The offense was below average, and their run prevention combined very good pitching with terrible fielding for end results that were also below average:
BsR = Fangraphs’ Baserunning Runs metric
After all their ups and downs, they ended up with a record and stats that were not terrible, but below average.
Not to ruin anyone’s day, but you may notice that their final stats are uncannily similar to those of the Mets:
Runs vs. Average
The graph below shows Runs above average for each team in the NL, by each aspect of the game, as measured at Fangraphs.
For the Phillies:
Offense: -19 runs (i.e. worse than average)
- Below average hitting: -20 runs
- About average baserunning: +1 run
Run prevention: +7 runs
- Well above average starting pitching: +40 runs
- Above average bullpen: +11 runs
- Awful fielding: -44 runs
The net total is that as a team they’ve been 11 runs worse than average:
Again, the team that looks the most like the Phillies is the one that followed them in the standings — for both the Mets and Phils, their key strength was the starting pitching, and their biggest flaw was their defense.
Compared to last year
The Phillies in the end scored fewer runs than in 2017, but that was in an environment where runs were harder to come by across the league.
They averaged 4.18 runs per game this year, compared 4.26 per game last year, but relative to the NL average, they improved from 93.0% of the average in 2017, to 95.7% this year.
The underlying stats were a mixed bag. As bay_area_phan pointed out (with I thought a touch of glee), the Phils finished 2018 with a batting average title of sorts, inching out three other teams in the battle for lowest average in MLB:
They supplemented that low(est) average with a high walk rate (9.5%, 3rd in the NL), and average power (6th in HRs, 8th in ISO).
Overall their underlying stats were below average — using Fangraphs’ metrics:
Hitting (wRC+): 91 = 10th
Baserunning (BsR): (2.9) = 8th
Overall Offense (hitting/baserunning): -63.3 runs = 10th
That Overall Offense number was exactly at the league average as late as September 19th:
Or, for another view, this is how the Phillies tracked relative to the NL average for Overall Offense:
For the table below, the first columns have the 2018 full season stats, and then how those compare to last year’s. The columns to the right show the stats through August 5th, and how they ranked at the time, followed by the stats since August 6th, and how those stats would rank over the full season (small sample rankings are prone to lots of noise).
*ISO, short for Isolated power, is the difference between batting average and slugging percentage, and essentially measures the average extra bases per at bat (1 for a double, 2 for a triple, 3 for a HR).
The Phillies were second in the NL in pitches per PA (4.02), behind only the Dodgers.
They were patient and worked counts, two things we’ve wanted from Phillies teams for years, but in the end they couldn’t capitalize often enough. They finished dead last in both making contact (74.7% of their swings), and in the quality of contact when they did connect: they were last in both Hard-hit rate (29.1%), and Line drive rate (19.9%).
It seemed that they were caught looking at the third strike an awful lot, and this isn’t included below but they were 6th in the NL in the number of strikeouts looking. However they led the league in strikeouts swinging, and in terms of strikeouts looking as a % of all their Ks, they were closer to the bottom, ranking 12th in the NL.
Progress over time
The graphs below show how some key stats have progressed. On the left is the weekly performance (with a dotted line showing the NL average for each one), and on the right is the cumulative season-to-date number. Batting Average and BB% together drive a team’s On Base Percentage, and Batting Average and ISO combine to form Slugging Percentage.
Batting — Individual
Below are the Phillies hitters ranked by OPS, along with how they compare to the MLB average OPS at their position. To the right are their OPS over the last 30, 14, and 7 days.
Below is how each hitter’s OPS has progressed over time.
These graphs are very busy — start with the names to see how they currently rank, and then follow each line to the left to see how they got there.
And while in most cases OPS is a reasonable proxy for the more accurate wOBA, below is the same graph for wOBA. The biggest difference is that Santana was third on the team in OPS, but second in wOBA, trading places with Franco:
The Runs Created stat, which has gone through various iterations, starting with Bill James in the 1980s, takes all of a hitter’s stats and estimates how many runs they should translate to. Fangraphs calculates a version of that called weighted Runs Created (wRC) which is a counting stat version of wOBA.
The table below shows wRC for each Phillie, by week.
And of course, the various stats all agree that the Phillies’ fielding was really, really bad:
Pitching - Individual
Below is a high-level break down of each starter’s games:
QS is the typical definition (6+ IP, 3 or less ER)
“bad” is any start with more ER than innings pitched, and
“other” is all the rest.
Through August 5th, the Phillies led the NL in Quality Starts, with 57 (51% of their games), 3 more than any other team:
Lively/DLS 1 each
From August 6th on, they were next-to-last in the NL, with only 14 Quality Starts (27% of their games).
And below are each pitcher’s games, with the Game Score for each one on the right (highlighted green if in the 60-79 range, dark green if 80+).