Welcome to the first edition of Stats All, Folks.
This season-long series will provide a primer into the statistical side of baseball, focusing especially on analytics and advanced stats. Think of it as a crash course in sabermetrics.
If you’re a baseball fan who is fascinated by new ways of analyzing and thinking about the game, but can’t always keep track of every sabermetric stat and what it means, then you’ve come to the right place. Here you can dip your toes into the water rather than diving right into the deep end.
If you’re a die-hard traditionalist who will always hold a special place in your heart for batting average and pitcher wins, I encourage you to read on and keep an open mind. You might find sabermetrics to be much more relatable — and enjoyable — than you expected.
If you’re an experienced sabermetrician who is already well-versed in baseball analytics, this series might not tell you anything you don’t already know. But keep reading anyway, and feel free to check my work.
To get started, let’s take a look at the stat that has perhaps become most closely associated with sabermetrics: WAR.
What is WAR?
WAR stands for Wins Above Replacement, and — in a nutshell — is designed to approximate a player’s overall value by boiling it down into a single number. Specifically, it attempts to answer the question: How many games does a player help his team win compared to a replacement player?
A replacement player, in this context, refers to a below-average major leaguer, the kind of guy who can fill out a Triple-A roster just fine, but is overmatched in the big leagues. You know the type. Picture a journeyman minor leaguer who hops from team to team without ever making an impact in the majors; a player your club could pick up off the scrap heap at a moment’s notice. Every organization in baseball has plenty of these players lingering in their system (and occasionally in the majors) at any given time.
WAR uses this concept of a replacement player as a baseline for comparison. Last year, for instance, Baseball-Reference.com pegged Manny Machado as worth 6.7 WAR. That means that if Machado had missed the entire season and Joe Replacement had filled in for him all year, it’s estimated the Orioles would’ve won 6.7 fewer games.
It would’ve been really awkward for the Orioles to finish the season with 82.3 wins. But the point is that WAR helps quantify Machado’s importance to the team last year; without him, they might have been barely a .500 club rather than a postseason participant.
Players can also have a negative WAR. Last year, the Orioles’ lowest WAR belonged to catcher Caleb Joseph at -0.9. That implies that the club actually would’ve won about one extra game if they’d used a replacement player instead of Joseph in 2016. Yeah, he had a rough season.
How is WAR calculated?
Well, let’s just say it’s probably not something you can do at home. The specific formulas for calculating WAR are more complex than I have room to write about here, but here’s an extremely condensed explanation.
Remember that WAR is meant to estimate a player’s contributions in all aspects of the game. So for position players, WAR incorporates statistics that measure a player’s offensive, defensive and baserunning performances. These are called Batting Runs (which measure how well a player gets on base and hits for power), Fielding Runs (a player’s defensive range and arm) and Baserunning Runs (a player’s ability to steal bases, take an extra base on a hit, etc.). WAR also makes adjustments based on what league the player is in (AL or NL) and what position he plays. Then it compares that to what kind of production would be expected from a replacement player.
For pitchers, WAR integrates several components. At its core, it measures how well the pitcher prevented runs and baserunners. It also takes into account the pitcher’s league, his home ballpark, the quality of the defense behind him and other factors. Then, of course, it compares this production to a replacement player. In this way, WAR is a better indicator of a pitcher’s value than conventional stats such as ERA. After all, a pitcher who has a 4.00 ERA in a homer-friendly ballpark in front of a lousy defense is very different from a pitcher who has a 4.00 ERA in a spacious ballpark with Gold Glovers behind him. WAR helps better differentiate the two.
It’s also important to note that several prominent baseball sites have their own versions of WAR, each of which is calculated slightly differently. Two commonly used ones are Baseball Reference WAR (often abbreviated as rWAR) and FanGraphs WAR (fWAR).
So which one is better? That’s a decision that only you can make, my friend. Search deep within your soul. Meditate. Embark on a spirit walk.
Or don’t worry about choosing one. Both are plenty informative. The WAR glossaries at Baseball Reference and FanGraphs provide much more detailed explanations of how each site calculates its version of the stat.
What does WAR tell us about the Orioles?
This being a site for Baltimore baseball, it’s time to steer this discussion of WAR toward the Orioles. Let’s take a look at how WAR ranks the best players in club history.
Even though rWAR and fWAR present somewhat different values for each player, they’re in complete agreement about who the top five most valuable Orioles are:
That list shouldn’t come as a surprise to most Orioles fans. WAR simply reinforces what most already knew: those players were the cream of the crop. The first four are Hall of Famers; the fifth deserves to be but hasn’t been elected yet.
Keep in mind that WAR is a cumulative stat. In general, the longer you play, the higher your WAR will be. Each of those players in the top five spent at least 10 seasons with the Orioles, and the first three spent 19 or more.
So if you’re wondering why Hall of Famer Frank Robinson doesn’t crack the list, it’s because he spent only six seasons with the Orioles, which didn’t give him as much opportunity as the others to pile up WAR. As an Oriole, Robinson finished with 32.3 rWAR/33.4 fWAR, ranking 11th on both lists. For his entire major league career, though, Robinson amassed a stupendous 107.2 rWAR and 104.0 fWAR, the highest of anyone who has ever played for the modern-day Orioles.
What about the current squad?
It’s probably no surprise that the Orioles’ active leader in WAR is their most-tenured player, center fielder Adam Jones, who entered the 2017 season with a 27.7 rWAR and 26.9 fWAR. But you might be surprised how close behind him Machado is. The 24-year-old Machado started the year at 24.4 rWAR/23.0 fWAR, having nearly equaled Jones’ on-field value despite his Orioles’ tenure being only half as long.
Way back at the turn of this century, a venture capital guy named Tom Hicks,…
While the signings of players with major league experience to minor league contracts might not…
Question: Why wouldn't the O's grab both Max Scherzer and Justin Verlander since they could…
Tomoyuki Sugano, the 35-year-old Japanese right-hander who signed with the Orioles on Monday, appeared on…
Question: With the signing of Tomoyuki Sugano, it seems that if you prefer a smaller…
It’s been a busy offseason for the Orioles, and it’s not done. While the Orioles…
View Comments
WAR is a dumb analysis. If Machado couldnt play the rest of the season the assumption then is Orioles would use a minor league callup vesus a free agent or trade for a quality replacement. Furthermore it uses stats from an assumed weak composite player.
I'm old fashioned ... Give me BA, RBI, #of errors, ERA & W/L.
WAR doesn't assume anything about what a team would do if they lost a player. It simply compiles a number of individual factors in attempt to get the full picture of a player.
I think that WAR has its flaws, but pretty much any individual factor you look at will. If you know what they are you can make your own final determinations about where things stand. As for your stats of choice, RBI and W/L in particular are horribly flawed because they are dependent on what a player's teammates do as well, which doesn't help you know anything about the individual.
Yeah, I think the premise is simply for comparison sake to best evaluate players as a whole. I don't think who replaces whom is the point. It's to take a baseline and go from there. Frankly, you can poke holes in every stat -- even home runs have multiple variables -- but assessing value through specific formulas has its place, too. I'm a firm believer in using everything you can, but never discount the eye test.
I agree with Ben up to a point. I'm a traditionalist that believed I understood how to calculate all the old school stats until just this week when, via this site, I learned that Sacrifice flies, although not counting in batting average, DO count in On Base Percentage. (this was news to me). Thank you BBC for correcting me on that one. My Strat-O-Matic numbers will never be the same .. Hah!
Back to WAR ... I'm not a big fan, but I'm going to stop short of calling them worthless or stupid. They're not. However, considering there are multiple calculations used by different organizations, this indicates to me that it's all somewhat subjective. Now on top of that, I learn it's cumulative. (thanks again Dan) If we're to take this number seriously, shouldn't we then divide that number by games played or seasons or something to that effect? It's not a fair comparison when a player with 10 solid years gets penalized when comparing him to a player that played for 20.
I'm with Ben ... until something changes .. give me BA, RBI, Slg. Pct. & ERA, etc. (wins vs. losses is another argument) Defense & base running still fall under the eyeball test.
Like I said above, give it all to me and I'll make my own conclusions. And that's what we're trying to do here. Break off an advanced stat in an understandable morsel. We're not looking for converts. Just thought it would make for interesting discussion. And, with our first installment, it has.
Yeah ... and Wieters went yard last night. Geeeeeze ...
Nice piece., though I guess I knew most of the information. I find WAR useful, but take it with a grain of salt. The defensive component is particularly nebulous, as even the humans and robots that compile those UZR's and DRS's admit that it's a young art, perhaps unfinished in its ultimate effectiveness. I try to use some combination of the eye test, traditional stats and WAR to come up with my own ridiculousness. Other new-fangled stats I do find more helpful: BABIP, FIP, WOBA, etc. Not sure why. Probably the fun acronyms. Or the fact that none of 'em lend themselves to yet another round of Edwin Starr-related hilarity.
Claude: I imagine we'll hit more or all during Paul's series this year. And I'll say once again. I like to think I'm open minded. But I fully distrust defensive metrics. Way, way, way too many variables for me to feel confident in those.
I think the unfinished nature of these sabermetric stats is realistic, but largely because these are based on models that seek to quantify things that are difficult to apply numbers to. The original baseball stats are pretty much not even really analyzed numbers but raw means, which in and of themselves have problems. They don't really look at variations from the mean or regressions back to it. They're more designed to capture a snapshot in time. This is one reason I like the model-based stats is because if one looked at Machado based on 11 games, we'd see he's hitting below .200. That would be misleading of his overall value. You need a larger sample for these averages to mean much, and you cannot make any kind of predictions based on them. Advanced stats do allow for some degree of extrapolation to establish expectations for a season. They're definitely not perfect, and they'll be improved upon as models continue to be refined. They have value as a piece of a larger picture.
I'm selling jeans. And right now my denim distribution instinct tells me a certain shortstop with a fun to chant name has become a liability.
Hmmm. Always a slow hitter to start. But I can see where the concern about defense is starting to manifest. Everyone was has a blip in 162. Maybe it's just getting out of the way early?
I've been watching his arm lately, and it hasn't seemed as strong as I remember. Is it just me?
Great article, Dan. I knew what WAR was but I learned a lot of new things from this article. I trust WAR not as an exact number of games, but a general indication of how much a player means to a team. Plus, there's waaaaaaaaaay too many intangibles that WAR can't take into account. Leadership, etc.
Yep. That's why you have to be careful with sabermetrics only. Nick Markakis' WAR hasn't been good in years. But for steady influence, steady leadership and steady everything, he could have helped the past two years.
I find hit hard to believe that Manny was *only* worth 7 wins more than a minor leaguer.
Why don't they measure it against the Mean or the Average MLB player? That stat would at least have more meaningful value since it attempts to compare actual MLB performances rather than a fictional set of stats by position that probably have correlation problems between positions, at least.
Seven team wins. That's pretty momentous for one guy who gets 4 ABs and touches the baseball maybe 5-8 times a game. And, though I can't speak with 100 percent knowledge, the baseline is a baseline. So that it is uniform, it doesn't really matter whether it's replacement or mid-level. You just need a jumping off point.
Part of the reason for using replacement players instead of "average" MLB players is that if a regular starter gets injured, the guy who replaces them is more likely to be a replacement-level player than an average MLB player.
For instance, if Manny got injured, who would replace him? Probably Ryan Flaherty or someone of his caliber. Flaherty, with all due respect to him, is a below-average MLB player. He's closer to a replacement-level player. (His career rWAR is 1.8 in five seasons, or about 0.4 WAR per season.)
Another factor is that replacement players are much easier to find than average players are. They're available all over the place, as minor league free agents, non-roster invitees in camp, etc. You can shake a tree and five replacement players will fall out. So that's why they're used as the baseline for the WAR comparison.
Thanks, Paul, for this series. I hope I can read every article.
I've been watching the Orioles for 58 seasons now, and I guess I'm a statistical traditionalist. I still believe RBI is the most important offensive statistic, for example. But I wish that percentage of inherited runners who score was a big stat for relievers.
My trouble with WAR is that every time I read a "primer" article on WAR, like yours here, it always shows me, like you did, what I already knew. Why should I take the trouble to understand and follow WAR just to know that Brooks, Frank, Cal, Eddie, Palmer, and Mussina were the best Os ever? Tell me that WAR shows that Dauer was better than Grich, or Melvin Mora was better than DeCinces, or Lee May was better than Boog, or Bumbry was better than Jones, and then maybe I'll be more interested in this stat.
Whatcha got with WAR that'll start a good discussion and make me want to dig deeper into how this stat is calculated?
Here's one that might interest you: according to rWAR, the sixth most valuable Oriole of all time was...Mark Belanger. (40.9). You can debate whether that's the proper ranking for him, but the fact that WAR incorporates Belanger's excellent defense helps reflect how valuable he was to the club. I think WAR does a better job of capturing his value than most traditional stats, which probably wouldn't put him even in the top 10.
Interesting you mention Grich. He's a guy who was criminally underrated as a player and didn't get the Hall of Fame consideration he deserved because his "conventional" stats weren't eye-popping. But WAR helps show his value. Grich had a career 70.9 rWAR, which is higher than many players who were elected to the Hall of Fame (including Jim Palmer and Eddie Murray).
And since you asked about Mora and DeCinces, yes, Mora has a higher rWAR as an Oriole than DeCinces did (29.0 to 22.8). Mora is another player who didn't get the recognition he deserved, partly because he never played on a winning Orioles team. But he had a few outstanding offensive seasons and played several positions well. Again, that's not something you could see just by looking at RBIs or other basic stats. RBIs aren't a very good indicator of hitting performance, since they depend so much on how often a player's teammates get on base in front of him.
Paul, Boog Robinson mentioned this earlier but I didn't see a response from the group. Can an overall career yearly average be calculated by dividing the overall WAR by the number of years played? For historical discussions and evaluations, this seems more useful than cumulative stats.
Technically, yes, you could do a yearly average WAR, but none of the sources I've come across do it that way. So it would require a lot of legwork.
Thanks, Paul. You really did a great job of breaking it down for an old, mathematically challenged guy like me.
So basically what WAR is telling me are things I already knew just from other less complex stats and what I've seen with my own two eyes. WAR looks to me like it tries to quantify the unquantifiable. If it is too complex to explain and too detailed a calculation, it is not worth my time to understand, especially when it doesn't tell me anything I do not already know.
The problem with the eye test is that it's very limited. We watch the Orioles every day, but we probably don't watch most other major leaguers more than a handful of times a year. So how do we truly determine the value of players we don't see as much? Or compare modern-day players with players of past eras? And even for the players we do see daily, we might not always be paying attention to smaller things such as how they run the bases, etc. That's the goal of WAR.
Yes, the calculation of WAR is very complex. But that's because it's incorporating many, many different aspects of the game and trying to paint a complete picture of a player. I think that's a point in its favor, not a flaw. It's not a perfect measure by any stretch, but it's a lot more complete than just using the most basic stats or the eye test.
Paul,
Given that WAR is a cumulative stat, what would be a fair way of determining which of two players had the better Orioles career if they played for the team for a different number of seasons, e.g., one played for the team for 10 years and the other five? As you said, "you could do a yearly average WAR, but none of the sources I’ve come across do it that way."