The 2018 Red Sox in Historical Context
As soon as any World Series is over, it’s fair to wonder how the most recent champion stacks up when compared to the history. And while sometimes the numbers are downright laughable, the 2018 Red Sox have been pretty extreme. In the regular season, they won five more games than anyone else. In the playoffs, they lost just once per series while eliminating the three other best teams in the game. Sometimes, you think about the history because you think you’re obligated. In this case, we look to the history because it seems like the Red Sox might’ve done something historic. It feels like this might’ve been one of the all-time greats.
I’ve run some numbers in order to see what we’ve got. I should acknowledge right here there’s no perfect, agreed-upon way to do this. There’s no ideal measure of a team overall. Does it matter how good a team is for seven months, or is it only the playoffs that matter, provided you do just enough to make it in in the first place? There are arguments to go in either direction, but for my purposes here, I’ve simply combined regular-season numbers with postseason numbers. The postseason sample, of course, is dwarfed by the regular-season sample, but that’s how I feel like it should be. You might have another opinion, and so you might trust your own analysis. Below, I’ll quickly present my own.
Just about all the information here comes from Baseball Reference, and I’ve pulled data going back to 1903. I’ve decided to focus on four team statistics. They are:
- overall winning percentage
- overall run differential per game
- overall winning percentage, z-score
- overall run differential per game, z-score
The first two, I think, are pretty obvious. Of course I should be looking at winning percentage, and then run differential is another marker of dominance. The latter two are less obvious, but by looking at z-scores — the number of standard deviations from the league mean — we’re able to get an idea of exceptionality, if you will. It’s more notable if a team stands out from a league otherwise characterized by parity. Here’s how the league winning-percentage standard deviations have moved since around the turn of the last millennium:
The league is less spread out than it used to be, but we also just saw the largest spread since 2002. Baseball in 2018 had very clear tiers. Here’s how the league run-differential standard deviations have moved since around the turn of the last millennium:
Same pattern, naturally. Used to be, ages ago, the difference between good teams and bad teams was more stark. We’ve seen a general trend toward parity, but, again, in 2018, the spread here was the biggest it’s been since 1956. In this most recent season, there were haves, and there were have-nots. The Red Sox were one of the haves. It’s fair to say the middle class was under-populated.
Anyway, that’s all the background. Now here is how the 2018 Red Sox measure up. I’m showing this in terms of percentile rankings out of all team-seasons. There are nearly 2,500 team-seasons in the sample. Spoiler: The Red Sox were very good.
We can go in order. Looking over almost 120 years of baseball, this year’s Red Sox ranked in the 99th percentile in winning percentage. They ranked in the 98th percentile in run differential per game. Looking at the same, except expressed in z-scores, the Red Sox ranked in the 99th and 97th percentiles, respectively. Tremendous season. One of the better seasons. The numbers here are in agreement.
But now let’s take a possibly controversial stance. What if we compare the Red Sox only to other champions? There have been 114 champions. What if we take the stance that, in order for a team to be truly historic, one of the all-time greats, it has to win it all? Here is the same plot, only this time, we’re comparing the 2018 Red Sox only to their title-winning predecessors.
The bars get lower, of course, because we’ve eliminated all the bad baseball teams. Still, among all previous World Series champions, the Red Sox look good across the board. Their percentiles are 86th, 80th, 86th, and 75th, respectively. There’s not quite enough to say the Red Sox are clearly one of the very best champions in history, but they’re in the best, say, 15 – 25%. Which is pretty damn terrific, when the peer group is all winning teams. There’s no shame in being far superior to the average champ. Especially in this modern era of super-teams.
The following table is split in two halves, even if it’s not very obvious. On the left, the 25 highest winning percentages of all time. On the right, the 25 highest winning-percentage z-scores of all time. In each case, the Red Sox are highlighted in yellow.
| Team | Year | Champ? | Win% | Team | Year | Champ? | Win%, Z |
|---|---|---|---|---|---|---|---|
| CHC | 1906 | No | 0.747 | DET | 1984 | Yes | 2.77 |
| NYY | 1927 | Yes | 0.722 | CLE | 1995 | No | 2.69 |
| PIT | 1909 | Yes | 0.717 | NYM | 1986 | Yes | 2.66 |
| NYY | 1998 | Yes | 0.714 | NYY | 1998 | Yes | 2.62 |
| CHC | 1907 | Yes | 0.712 | SEA | 2001 | No | 2.55 |
| NYY | 1939 | Yes | 0.710 | CIN | 1975 | Yes | 2.38 |
| NYY | 1932 | Yes | 0.703 | OAK | 1990 | No | 2.36 |
| CLE | 1954 | No | 0.703 | BAL | 1970 | Yes | 2.29 |
| SEA | 2001 | No | 0.698 | DET | 1968 | Yes | 2.26 |
| PHA | 1929 | Yes | 0.697 | STL | 1944 | Yes | 2.25 |
| NYG | 1904 | No | 0.693 | MON | 1994 | No | 2.18 |
| PHA | 1931 | No | 0.692 | CHC | 2016 | Yes | 2.18 |
| STL | 1942 | Yes | 0.692 | PHA | 1929 | Yes | 2.12 |
| NYG | 1905 | Yes | 0.690 | LAD | 2017 | No | 2.09 |
| CLE | 1995 | No | 0.686 | NYY | 2009 | Yes | 2.09 |
| BOS | 1912 | Yes | 0.686 | ATL | 1997 | No | 2.08 |
| PHA | 1910 | Yes | 0.684 | OAK | 1989 | Yes | 2.07 |
| CIN | 1919 | Yes | 0.682 | ATL | 1995 | Yes | 2.03 |
| STL | 1944 | Yes | 0.681 | CIN | 1976 | Yes | 2.01 |
| CHC | 1909 | No | 0.680 | CHW | 2005 | Yes | 2.00 |
| NYY | 1961 | Yes | 0.677 | NYY | 1927 | Yes | 1.99 |
| BAL | 1970 | Yes | 0.676 | BOS | 2018 | Yes | 1.98 |
| BOS | 2018 | Yes | 0.676 | LAD | 1974 | No | 1.98 |
| BRO | 1942 | No | 0.675 | CHC | 1906 | No | 1.97 |
| BOS | 1915 | Yes | 0.673 | CHC | 1907 | Yes | 1.94 |
On the left, in terms of raw winning, the Red Sox haven’t been surpassed by many champs in what we can consider the more modern era. On the right, more recent teams are better represented, as the leagues have become more even. There are reasons one might prefer either measure. The one on the left is elegant in its relative simplicity.
To whatever extent the Red Sox haven’t often been regarded as one of the best teams ever, during and after the season, it’s probably because of the stratification of the leagues. There were a number of very good teams, in part because there were a number of very bad teams, and among those very good teams, the Red Sox weren’t extraordinary. They got somewhat lost in the crowd, as some attention would’ve gone to the Yankees, and the Astros, and so on and so forth. Still, the Red Sox were legitimately terrific, and obviously, they strengthened their case all the more by breezing through difficult competition in October. The 2018 Red Sox were great. They wound up very deserving champions. They’re not the best champion ever, and they might not be all that close to being the best champion ever, but even as champions go, few teams have been so good for so long. You couldn’t really ask for a more impressive seven months.
Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.
Interesting that a few spots above the Red Sox on the right side of the table are the 99-win 2005 White Sox.
Of course, they went 11-1 in the playoffs, though definitely not beating three teams as good as the one’s the Red Sox just beat. Just a view that the ’05 White Sox might be underrated historically.
IIRC the 05 White Sox weren’t even the best team in their division according to BaseRuns, Cleveland was. So like Boston they surprised the metrics by going through their postseason opponents like a buzzsaw.
It’s also very interesting to see that they’re above the 116-win (in 154 games!) 1906 Cubs, who have the best winning percentage of all time.
The 1906 World Series was a, possibly the, historic upset, but it’s crazy to think that the team the Sox beat was, by that measure, at least, not even as good as the oft-forgotten 2005 champs, or the 2016 curse-breakers.
Interestingly, although the 2018 Red Sox appear to be solidly around the 22nd or 23rd best team in history, there are only 7 teams ahead of them in both Win% and Win% Z-score: 1927 and 1998 Yankees, 2001 Mariners, 1929 Athletics, 1995 Indians, 1944 Cardinals, and 1970 Orioles.
Is it possible that the Red Sox’s true talent level was underestimated?
The Astros and Yankees I think probably had more proven talent, but there was a lot of untapped (or undertapped) upside on Boston’s roster. Devers was touted prospect that had a replacement-level rookie season but then an excellent postseason. Kelly had been an excellent reliever for a season out of the past season and a half. Price had been a Cy Young winner turned pumpkin that turned back into a Cy Young winner.
For all the things that broke right during the regular, I think you had a lot of players actually underperforming their potential on the roster (Bradley Jr., Devers, Barnes, Kelly, Price, to name a few) that were able to finally meet it on the biggest stage. Those reserves of talent helped push them over the top, even when the stars faltered a bit.
The irony is that for over a decade, FG was accused of chronically over-rating Red Sox talent. It was a ridiculous accusation (as if FG controls the algo behind Steamer), but one that had a surprising amount of support, bc people love a conspiracy theory.
I think one of the things that is maybe slighting the perception is that even after their dominance in the postseason, the Red Sox’s combined postseason & regular season RDiff is still less than Houston’s combined RDiff. 263/270, respectively. Houston was just so good this year during the regular season.
Furthermore, Red Sox were no doubt an all time team, but isn’t it possible that we had 4 all time teams in the postseason this year by the measures you showed here (BOS/HOU/NYY/LA)?
It would be interesting to try this study again, but instead of generating z-scores with the mean being any team in the league, generate it from only against playoff teams. Who cares how much better the best team is than the worst team? I think it would be more revealing to see how much Boston stood out from the other great teams.
Rightly or wrongly, the Sox were actually underdogs in the majority of their post-season games, supporting the point that the richest — er, best — teams were particularly dominant this season.
This is skewed a bit since they also played more road games than home games in the playoffs.
Fully agree they were seen as underdogs against Houston, but I’ve seen some MSM types tout that they were underdogs against LA which just isn’t the case.
“Who cares how much better the best team is than the worst team? I think it would be more revealing to see how much Boston stood out from the other great teams.”
Jeff did that, though.
If a team plays in a poor league, that’ll both increase their win percentage and their win percentage z-score, right? So, suppose a team is really good but plays in a year in which the league on the whole isn’t so good, won’t their win% z-score be kind of artificially propped up, but the same team in a year where the league has more good teams will look worse by the z-score method? Isn’t that suggestive of the z-score method just telling us how unusually good relative to the other teams that year (not historically) said team is? Is that very useful knowledge?
Well, there is no way to compare team from two different years, so all you have is how they were relative to the competition.
That said, an adjustment for schedule should be applied to any serious analysis of team quality — esp. in seasons with historically bad teams. That is, if you happened to play the Orioles or Royals more than most teams, it might change your schedule-adjusted run differential. We apply this same logic to college football rankings, why not MLB?
I made the same point below. It doesn’t seem to be useful.
Thanks for addressing something i was thinking about! Three comments:
(1) i think the main reason i had the feeling that the 2018 red sox looked like one of the best teams ever is that in the playoffs they convincingly beat all three of the other teams who seemed to me to be their strongest possible opponents. No one cleared the way for them. They got no luck of the draw. And they still won all three series decisively. It seems to me that playoff games should count more than regular season games, say twice as much, and strength of opponents in the playoffs (relative to other teams that year) should be considered too.
(2) i don’t see why z-rating should be considered. I might be convinced but it isn’t clear to me. Why should the relative historical value of a team be measured by overall variation in the league? The comparison should be with teams from other years. It seems to me that what’s relevant here is the quality of the teams which a given team played in a given year. Before 1969 there wasn’t any basis for comparison, because there were two leagues, everyone played the other teams in their league an equal number of times, and their was no interleague play so it was impossible to tell which league was stronger. Since then, though, there has been measurable variation in quality of opposition, and that should be considered. The 2018 red sox played in a division with one very strong team, one good team, one poor team, and one abysmal team, and that influenced the strength of their schedule, whereas, for example, the 2016 cubs played in a division with one good team and three poor teams. Things like that should be considered.
(3) run differential is just one way of stripping away the luck factor resulting in a team’s won-lost record. Base runs is another, because it neutralizes sequencing. Something to try to neutralize babip luck is another (that was a huge factor for the 2001 mariners and 2016 cubs).
Never in the history of MLB have so many teams not cared about winning. (Sorry. This was not supposed to be a reply to the message above but to the article).
To clarify the first part of point (2), the purpose of the study is to rank teams from different years with respect to each other, and in such a situation it doesn’t make sense to refer to the overall performance of the rest of the teams that year (it might make sense to refer to the specific performance of the teams which the team investigated played the most, but that’s a different question). If the rest of the teams are more balanced, that means that there are fewer good teams who might actually beat the team in question a lot of the time, but there are also fewer bad teams to beat up, while the reverse is true if the rest of the teams are less balanced. Neither one of these situations would systematically favor better or worse stats than the other, so there is no reason to take this into consideration. If the question were ‘which team was the most dominant?” rather than ‘which team is the best?’, it would be a different matter, but that’s a different question. If judging quality of teams in different years, it hardly makes sense to consider the quality of the second-best team in each year, as that doesn’t affect the quality of the best team.
The more extreme teams there are, the more unbalanced schedules become. No idea off hand if the Red Sox played an easy or tough schedule, but it should be investigated, esp in the era of tanking.
” Neither one of these situations would systematically favor better or worse stats than the other, so there is no reason to take this into consideration. ”
You are assuming that talent given up by a tanking team will help a good team by the same amount as it hurts the tanking team, in terms of run differential. This really isn’t the case, though. Good players — esp. pitchers — can make a huge difference on a bad team and only marginally improve a good team — due to the vastly differently replacement levels of each team’s roster. The more extreme tanking becomes, the more impactful this effect becomes.
Imagine for a second if half of MLB teams gave all of their talent to the remaining teams and fielded teams of random high schoolers. All the real teams would be guaranteed ~81 wins as a floor. A mediocre team (among those fielding real teams) would have 120 wins and a massive run differential. A good team would have 130+ wins. Also, scheduling effects would become a critical part of determining the most dominant team.
I think that tanking is only a recent phenomenon (that’s my impression) so it wouldn’t really be important to evaluating the relative strength of teams over the last 115 years. If the purpose is to evaluate how this specifically affected the Red Sox this year, where tanking may be a factor, then the study would need to be focused on that problem.
Once you go down the rabbit hole of (3), you might as well just ask, which team would analytics predict to be best? The problem is that this assumes that no team can affect run sequencing, BABIP, etc. Wins and run diff. — provided they are adjusted for context (schedule) — are IMO the most reasonable things to look at for historical analysis of team quality.
BTW, based purely on player-level predictive analytics, the Astros were better than the Red Sox. I.e., if they played a series tomorrow, the Astros would be favored. I think that’s an interesting footnote, but not a valid way to assess a team’s historical dominance.
Re: (2), I think his use of Z-score is an attempt to control for the fact that tanking factor has emerged and could be argued to be inflating win totals for the top teams, rel. to the past. As an extreme example, one could not compare win pct. of a MLB team to that of an NBA team if looking to measure dominance. Baseball is still baseball, but it is changing.
Run differential is already a rabbit hole, just a shorter one. If you’re going to take the trouble to go down one, you may as well go down them all.
In terms of (2), there is indeed a possibility that the 2018 Red Sox’ stats may have been helped by tanking teams more than other teams on the list here were. But (a) that wasn’t Jeff’s stated reason for using z-scores, and (b) if that’s really the issue, the study should be designed around addressing the question of the effect of tanking in a systematic way.
I believe there is a flaw in simply equating post-season games with regular season games, particularly since the article seeks to compare average performance across eras. Modern teams total games include up to 19 post-season games — always against the highest level competition. Pre-playoff era teams played at most seven. This clearly advantages the older teams in the presentation in this article.
I think a similar comment was made above, but how would things look through a more context-neutral lens? For instance, OPS (or adjusted-OPS) differential or wRC+ (by hitters) minus wRC+ allowed (by pitchers). Jeff often cites those type of underlying metrics in his AL vs. NL posts, seems like they could be useful here as well.
The 2018 Red Sox are at least in the running for the best Red Sox team in franchise history. I think it’s down between the 1912, 1915, and 2018 teams.
“while eliminating the three other best teams in the game. ”
I only went back to 2000, but looking at the average wins for the eliminated teams:
99.3 BOS 2004
98.3 BOS 2018
98.0 STL 2011
97.3 LAA 2002
96.3 FLA 2003
96.0 HOU 2017
94.0 BOS 2013
93.3 BOS 2007
93.3 STL 2006
93.0 CHW 2005
92.7 SFG 2010
92.3 NYY 2009
92.0 ARI 2001
92.0 NYY 2000
91.0 SFG 2012
90.8 SFG 2014
90.7 CHC 2016
90.3 PHI 2008
89.7 KCR 2015
So, by this measure, the Red Sox had a hell of a hill to climb…
And i think’s that’s an underestimation because the dodgers were better than a 90-win team.
An ELI5 on standard deviation, please? I know it’s a representation of, for want of a better term, the difference between a sample and the whole, but specifically in regards to the 2 charts above (SD:W% and SD:RDiff/G): are the points on the graphs the differences between the best team(s) and the worst team(s), or between the best team (s) and the average team?
Thanks for improving my perspective on this team, Jeff. One observation: When you say “since around the turn of the last millennium” in a few places above, you mean something like “since around the turn of the previous century” – right?
Keep up the great writing!