In 2018, WAR is everywhere. Love it or hate it, the wins above replacement metric has changed the way we evaluate Major League Baseball players. WAR is an attempt to encapsulate all tangible aspects of a player’s value into one number. It allows for all players throughout MLB history to be compared on a single scale. It’s a grand idea that has firmly taken root.
Whether it’s fans, baseball writers, agents, or executives, just about everyone citing WAR understands the general idea. But I’ve long wondered how many of us are capable of pulling open a spreadsheet and accurately calculating WAR, with a reasonable understanding of each component. Furthermore, how many can explain the limitations of the current WAR calculation? And do we understand which subjective choices were made to get to the current formula?
For a long time, I’ve wanted to write this series. I’m a reasonable candidate: I’m not bad with numbers, nor am I especially talented. I know my way around Excel, but I’m not an expert. If I run into roadblocks as I try to understand WAR, perhaps you will too. If not, hopefully you can help educate me in the comments section. Let’s crack open the hood and attempt to understand WAR from a layman’s perspective.
As you might imagine, the WAR calculation differs for position players and pitchers. Plus, major sites like FanGraphs, Baseball Reference, and Baseball Prospectus have different formulas. For this exercise, I’m going to dig into FanGraphs WAR. That’s the version we use here at MLBTR. I don’t have any evidence of this, but I feel that FanGraphs WAR might be the most commonly cited version. Otherwise, I don’t have any justification as to why MLBTR should cite FanGraphs WAR and not someone else’s. By the end of this project I hope to have a clear understanding of the differences.
For simplicity’s sake, we’ll begin this exercise by examining position players. From what I understand, there is a little more room for subjectivity in the pitching formula, so we’ll leave that discussion for later. I’ll approach the subject by utilizing a case study, as that will keep us grounded in reality. I’ll attempt to see how if I can reasonably arrive at the known WAR figure that was compiled, examining lessons that arise along the way.
So, here’s a quick preview of what’s coming. Our preliminary subject will be the 2017 season of Chris Taylor of the Dodgers. It’s an interesting year to look at, as he racked up an impressive 4.8 WAR while making defensive contributions at five different positions. It’s easy to see that Taylor made positive contributions in offense, defense, and baserunning. We’ll examine each of the three components in separate installments, beginning next time with Taylor’s work at the plate.
I hope that this exercise will offer plenty of opportunities for dialogue on a notable, sometimes misunderstood subject. I’m looking forward to plenty of respectful debate along the way.
Thank you!!!
Who downvotes “thank you”?
I question why a lot of the comments have downvotes. Even simple “thank yous” and wishes for players to heal quickly always have a downvote or two. Now watch as this comment gets downvotes
X has a history on this site. People tend to downvote him just because it is him.
Cmon, Gare. You are better than that… Anyone can make multiple accounts..
Also pple can down-vote comments, which can be correct. This got 7 down-votes: (Feb 2018)
“Miguel being great in that sip of coffee, foreshadows a great 2018, which makes machado a luxury, much like Harper is now.”
“But a lot has to go right, and his glove — while improving— is not great.”
“I believe Miguel 3B job his to lose and Yanks will give him first crack” Or… (Same thread)
On side note, Andujar has 15 HR, .293 BA worth +1.4 WAR so far 2018
WAR incorporates players’ offense and defense, and Andujar plays atrocious 3B defense..( -14 DRS, -7.2 UZR…) yet +1.4 WAR is still good. Assume he plays league average 3B defense. How much better you think Andujar’s +1.4 WAR would be?
It’s important to note he’s learning the position on the fly^ These are growing pains of young rookie, and I think those 3B def #’s will surely improve.
Andujar is”learning [3B] on he fly”? He has played 670 professional games at the position, I would hope he has learned the position by now.
And people like matt chapman who’s an excellent defender and slightly above offensive player has a 6.0 war
He played DH one day, Went 3-for-5 with HR, tied game, and hit game-winning run in 13th (2 RBI in 4-3 W)
His WAR went from +1.4 to +1.5. Def went: -7.2 UZR to -9.2 UZR. Stayed -14 DRS
One game! 3-for-5 HR! 2 RBI! Feel DH almost completely negated his Fangraphs WAR gain.
Maybe MLBTR should just dump down votes.
JEEZ X, these guys wont cut you any break.
Thank you for this.
This is an excellent idea! I look forward to the breakdowns. Things like this are why this site is so much better than similar rumors sources.
I’m excited. Love WAR. Would you be in favor of Fangraphs slightly amending WAR formula, to not apply negative defense for DH? Should be ZERO defense, not negative?
My favorite stat. I’m excited you decided to take this on.
I’m not going to act like I know WAR. honestly know as much as you guys.. which is why Tim deserves huge credit for taking on this project.
To ask my question in Layman’s terms.. Do you think DHs get abused via the WAR system? It’s my only (minor) gripe. Thanks for all that you do
To be precise, the DH does not get a “negative defense” but rather a *positional* adjustment.
This adjustment demands extra from a DH player because not only are they producing *no* defensive value, they are also forcing another player out on the field in their place.
It’s the same logic as why SS get a positive positional adjustment (premium defensive position that lets less defense-first players be ‘hidden’ in less demanding roles)..
They’re not “forcing another player out on the field in their place,” they’re hitting for the pitcher, who is going to be out on the field anyway. Shouldn’t the DH’s offensive production vs. replacement-level offensive production among pitchers cancel out the absence of defensive contribution?
Replacement level offensive production from a pitcher compared to a DH would be an absolutely ridiculous baseline. The worst hitting DH would still have a positive WAR number in that scenario.
I disagree completely. Raise the positional adjustments for the rest instead of subtracting from the DH if you want. It makes no sense to say that player being your DH hurts you in the field since obviously if he is your DH its because you are in a league where you have a player that bats and does not field.
This is a fundamental failure to understand the purpose of the positional adjustment. Fangraphs would do well not to lump it into the ‘defensive’ category, which they do because its tied to the position the player plays. It has nothing to do whatsoever with what the player does on the field
The penalty is essentially a credit or penalty in runs per 600 plate appearances that adjusts the amount of offensive value a player needs to provide in order to be average. A league average bat at catcher is very good – probably a top 5 or 10 catcher in MLB
A league average bat at DH is replacement level, because league average bats that can’t play defense are readily available in AAA baseball in the form of non-prospect players.
The positional adjustment has absolutely nothing to do with defense. It has to do with the fact that a designated hitter must cross a higher offensive bar to provide any value above replacement level.
With the positional adjustment bonus for catchers, one can be a replacement level catcher with like a 60 wRC+ (look at Austin Barnes in 2018, hes at 0.2 WAR with a 69 wRC+). If Barnes could not catch, that bat would not be rosterable or useful.
This adjustment is there to let you compare players across different positions in a useful fashion. Comparing to league average numbers is useless because of the relative rarity or abundance of league average bats at different positions.. Another good example right now is Tucker Barnhart vs Ian Desmond. They have identical offensive results to this point (86 wRC+). Barnhart has been worth nearly a win – 0.9 war. Desmond is negative -0.2 WAR, because he’s used as a first baseman and corner outfielder, and there are literally a couple dozen dudes sitting around in AAA that no one wants who could hit that well and play first base. None of them can catch.
Barnhart’s a useful, if not exceptional, player with identical results due entirely to the position he plays. This is what the positional component of WAR does.
Any player can DH, so the offensive bar for replacement level is the highest, and therefore the positional penalty is the largest. The population of dudes who can hit that well when defense isn’t a factor is dramatically larger than the population of dudes who can hit that well and play shortstop, or catch.
I know you weren’t reply to just me… but I understand the point of it. My point is that it would be better to use the easiest position to fill (one that doesn’t field) as +/- 0 and make your adjustments all positive to the rest. Seeing defensive skills as a plus makes more intuitive sense than the negative positional adjustment for a DH.
The problem is that WAR is dependent on average performances for comparison. Even replacement level is based first on league average (because league average determines the run environment). We are not capable of determining whether a shortstop or a catcher’s play has more impact on the game. So we first compare their performance (offensive or defensive) to the league average performance at their position, then we adjust it to compare to the replacement level, and only then can we compare them to each other
What you want is only possible if we could somehow deduce in an absolute way how many runs a given players’ contributions were worth. As it is, especially with actual defensive metrics, we can only compare to averages, and after determining that, compare across positions.
Replacement level itself exists to index performance to financial value; if we stayed comparing everything to average, many players who are below-average but valuable to teams would appear to have negative valuation (Barnhart is again a good example above). Since teams clearly value those players at greater than 0, replacement level essentially sets the floor of 0 value at the same floor of $0 value (freely available talent, eg minor league non-prospect journeymen who can be had for a pittance by any team at any time).
Replacement level itself was agreed between FG and BR to be a team that won a certain (very, very low) percentage of it’s games. In order to determine that, though, you first need to know what a league average team does (as the value of runs varies depending on the overall run environment; you have to know how many runs a 81-81 team scores before you can determine the RS/RA of a .294 team, as it will be different in 1968 than it is in 2004).
Positional adjustments are, basically, hacky but necessary because our measurement of baseball is not precise enough to support the kind of measurement you’re asking for.
Think about what you see with statcast: we have an outfield defensive stat, but its again, above average, because comparing how many raw outs of value a center fielder produces to a shortstop is basically impossible, and mostly useless; what we’re interested in is how a CF compares to other center fielders. that a shortstop may handle 250% more balls would distort any direct comparison to the point of uselessness.
And, more imporitantly, the point is that to be a replacement level DH you have to be an -above average hitter-. A league average hitter at DH is BAD. that’s Ryon Healy. It doesn’t help your team win.
By penalziing the DH position in run value, you move the bar to achieve averageness upwards. You pretty much have to be like a like a 110 wRC+ hitter to be valuable as a DH. that’s intuitive because anyone can DH. And the positional adjustment makes that so.
Being a DH-only or corner-only bat significantly harms a players value. You see this with prospects all the time, as they are treated as less valuable with the same performance if they’re seen as 1b only or corner OF only, because their path to being a worthwhile major leaguer is narrow – rake.
WAR is an economic measurement more than anything else, and so it reflects this reality.
Junts1. Amazing. Makes me even more excited for Tim’s WAR “for dummies” series, hope by Layman’s attempt.. means “for dummies” 🙂
Beautifully written. It shows so much goes into this stat. I love it. Thanks man. Bookmarked 🙂
DiamondNRG: “My point is that it would be better to use the easiest position to fill (one that doesn’t field) as +/- 0 and make your adjustments all positive to the rest. ”
You *could* index every position to a 0 run adjustment (e.g. DH: 0, 1B: +5, SS: +25, etc.), but then you would subtract a constant 17.5 from everyone anyways and be right back to where you started. 🙂
hahahahahaha
Oh man I can’t believe you’re doing this. Can’t wait to read up on this entire series. I’m not going to lie, I find some of the longer, non news articles on this site hard to read sometimes, but this is one that I’m definitely not going to skip over.
Thank you for being honest
Sounds awesome, hopefully someone horrible at math like me can understand this
The WAR stat and others like it, are hurting the game. They are too one dimensional and don’t allow consideration for a players intangibles. Plus hearing about them, makes the game boring to watch.
Lol I think it makes the game way more interesting, as long as you don’t only rely on it.
I’d much rather he announcers and analysts talk about stuff like WAR rather than having to face palm any time they say how great a player is due to their win-loss record or batting average. And why is WAR to one dimensional? Can you name one dinosaur age stat that takes intangibles into account?
Your issue with WAR is that it doesn’t value something that by definition is incapable of being valued? I’m curious to know what stat or stats you believe measure a player’s intangibles.
There are no stats that measure a player’s intangibles…that’s what makes them intangible. You are of the mind that there should be a universal stat. I think it is impossible to create that universal stat.
People treat baseball like it should be math or science class and be black or white. I disagree. I think it should be more like a literature or history class because there are many shades of gray. In math or science, there are true right and wrong answers. In history, there are no right answers, only reasonably argued points. Sure some things can be argued much more easily than others, but they are still not facts.
Watch, someone will come up with a presidential WAR at some point…
No stat measures intangibles; but managers and such are placing way to much emphasis on these new generation stats to make decisions. The game itself is one dimension. It’s either strike out or homer. No bunts, hit run, or sacrifice plays anymore. I’m a life long fan who had season tickets for many years and I get bored with the game today. Way too many pitching changes. It’s being over managed. The game has evolved in a bad way. Listen to the players from the 70s and 80s, they feel the same way.
I don’t really care what those players think, and I disagree that it’s evolved in a bad way.
No, not that at all. The problems inherent with WAR are the problem inherent with all mathematical models. The purpose of building a mathematical model is to use the past as a means to predict the future. Whether the model actually does this is how you know if it is explaining anything.
Models are all based on assumptions about how the system you are measuring functions. These assumptions have to be challenged and tested repeatedly or the model never improves. If you simply accept the number cranked out by the model as objective truth without examining how that number was created and whether it accurately predicts the future, you won’t know whether you are reading signal or noise. You also have to rigorously examine you data collection method for flaws (AKA, garbage in, garbage out). I’m not sure any of the required rigor is being applied to calculating WAR.
So, bottom line: In theory, anything can be modeled. But in reality, complex systems can be very, very difficult to model. Just coming up with a number is not enough.
BlueSkyLA….Very, very well said.
Thanks. If you build a model to predict the weather, you know how well it works by how well it predicts… the weather. Does WAR predict anything? Maybe it does, maybe it has been ground-truthed somehow, somewhere, but not that I”ve ever seen. I’m hoping some of these basic statistical concepts can be explored in this series.
Your issue with WAR is that it doesn’t value something that by definition is incapable of being value? I’m curious to know what stat or stats you believe measure a player’s intangibles.
Pitching WAR undervalues soft contact…..
Tim this is going to be awesome! Thank you for doing this. OK, I’ll get out of the way now and enjoy the ride…..
I am looking forward to this. I use some of the other basic sabremetric stats for analysis, but I haven’t really delved into WAR. I’m not sure I like it, for the same reason it is used, because it takes every facet of a player and tries to distill it down to one number. Most importantly, it invites more bias into the discussion, by determining what is most important. That said, I am looking forward to better understanding it. One question, is it Fangraphs WAR that is heavier on defense? Thanks Tim. You guys do a great job here.
WAR has value, just like any other baseball stat has since Batting Average. Fans just tend to over-estimate their value. In MLB, you have the top of the heap (about 20%), the bottom of the heap (another 20%), and 60% of a bunch of players who are all the same guy. Arguing one is better than another because he has +1 more WAR is just silly.
Not to mention the unbalanced schedule almost by itself negates the worth of any stat. If you feel differently, then you think pitching against the Royals in Tropicana (in front of 8,000 retirees) is the SAME EVENT as pitching against the Red Sox in Fenway (in front of 35,000 lunatics). And you HAVE to believe that.
Again, they all have their worth, but it’s only in extremely narrow ways.
The simple fact that there is not one universally accepted formula for WAR is what makes it both interesting and unappealing to me in terms of evaluation. I wanna see bunts against the shift and more squeeze plays too so what do I know?
I hate math, and I despise Bill James.
I enjoy the game the way it was meant to be enjoyed.
There’s an entire generation of fans who grew up living off of box scores in the paper, stat lines off of baseball cards, the baseball encyclopedia… baseball has always been a math game.
Okay then……I hate new math.
Awesome!! Thanks for taking this on. I love all the new metrics to calculate a player, and going through the exercise will be great!!
Chris Taylor is why WAR is flawed. He has the highest bWAR of position players on the Dodgers not including Manny. But he really has not been close to their best player
As of writing for position players on the Dodgers: Max Muncy has 3.4 WAR; Yasmani Grandal has 2.9 WAR; Chris Taylor is third at 2.1 WAR, See: fangraphs.com/teams/dodgers
Thats why i said bWAR. Fangraphs is the best source imo
Apologies, missed that detail/assume you were talking about fWAR.
That said, the only major difference between fWAR and bWAR for position players is how they value defensive contribution. by about a win. I’m sure you could find counter examples in the other direction too.
One thing I genuinely don’t know if anyone has roughly calculated is what the “error bars” are on WAR. I wouldn’t be surprised if someone told me that by season’s end the expected error is +/- 1 whole win.
So in that respect, fWAR and bWAR for Chris Taylor maybe aren’t that far apart after all!
WAR is ultimately a tool to compare players to one another. WARmongerers get too caught up in the number itself and stuff like $$/WAR valuations, and the WARbashers discard the whole thing entirely. Both kind of miss the point, I think.
Definitely. People who use it on its own as an end all be all stat will never appropriately value players. And people who write it off entirely are just scared of a change and likely still value awful stats like RBI’s and Saves
A great endeavor, though I do have a question. If Fangraphs is the typically referenced source, why does the hyperlink to players lead to BRef?
I would like to see that changed too!
just tell me how to figure it
I am not a huge sabermetric fan, but I don’t mind WAR that much. My biggest problem with it was mentioned in the article. There is no universal formula for calculating WAR. It seems like anyone can just start a website, make their own WAR calculation and say it is right.
The only other major problem is with the attitudes of some proponents of WAR and other metrics. I still like RBIs, wins, saves, batting average and the rest. I’m not saying they are indicators of overall value or anything other than comparison between everyone in specific events. So I don’t really care whether you like the stats or not. If you see someone mention RBIs, just ignore it. You don’t need to use words like antiquated or dinosaur to make yourself feel superior. There are lots of things I don’t like, but I don’t treat people like they are idiots for liking them.
I suppose… but when you say you like pitcher wins… it makes it very hard to continue in a serious conversation… I mean heaven forbid your guy goes 6 innings and gives up 6 runs and you win 7-6 … versus the pitcher that goes 8 innings and gives up 1 and gets a L. The dumbest stat ever. Wins are for teams, not pitchers.
If they meant nothing, no one would keep track of them. When Clayton Kershaw won the MVP despite playing in only 27 games, everyone loved to make the argument that the starting pitcher is so important to the outcome of a game that it didn’t matter that he only played in 16% or so of his team’s games. But when wins are mentioned, it doesn’t matter at all because a pitcher is only part of the team.
I will be the first to agree that wins for relief pitchers are completely based on luck and utterly useless. But if you have two starters that go 7 innings and one gives up 1 run and the other gives up 2 runs and the final is 2-1, then there is a winner and a loser. One guy that was better than another in a specific game. It may not seem fair to tag a guy a loser despite only surrendering 2 runs, but that’s why they have the term “hard luck loser.” The same goes for a guy that gives up 6 runs in a 7-6 win. The team still gets the win despite winning ugly, so why is so hard to accept a pitcher getting that win? Does a team deserve a win if they win a 1-0 game when the only run scores on an error? It doesn’t matter HOW a win is achieved, it only matter THAT a win is achieved. Same with losses. I would much rather have a guy that goes 18-6 with a 3.20 ERA than a guy that goes 11-12 with a 2.20 ERA. You don’t go to the playoffs based on looking good or winning pretty, you go based on winning period.
And you win as a team. In a 6-7 game, neither pitcher helped his team win… one of them did it in spite of their pitching. The other didn’t because of it. Maybe we should give batters a win or a loss depending on if they hit over .250 for the night? Come on…
That doesn’t make any sense. The guy who threw the 2.20 ERA gives your team a better chance to win. The only reason you would pick the other pitcher is because he plays on a better team?
You are looking at it in only one way. The guy with the most wins did enough to win, the other guy didn’t. There is no magic number to win a game, you just have to do better than the other guy. I’d rather win a 7-6 game than lose a 1-0 game. Apparently you would rather look good in a loss than look not so good in a victory. Having a so-called “better chance to win” is not actually winning,
Besides, you don’t even know that one guy is a better pitcher than the other. Maybe the guy that went 18-6 pitched 18 0 or 1 run outings and got really blown out in the 6 losses and the guy that went 11-12 gave up 2 or 3 runs in every game, win or loss.
One pitcher helped his team win because he gave up one less run than the other guy.
That’s terrible logic. The only thing that you are arguing is that you would want pitcher who play on better teams.
Based on your logic, you would want 2015 Drew Hutchison, who went 13-5 despite a 5.57 ERA. What about Jacob Degrom? Is 2015 Hutchison better than Degrom even though he has a 1.85 ERA and a losing record?
Some old school stats are worth defending, but a win-loss record is not one of them. It is garbage all the way around and doesn’t show if a player is good/valuable/bad or not.
I never once said that wins should be the primary factor in judging a pitcher and with starters going fewer innings now, they are probably not as important as they once were. I am just arguing that they are not USELESS. 2015 Drew Hutchison is a clear outlier as is 2018 Jacob deGrom. Asking that question just makes you look foolish. We are not talking about picking a guy like Hutchison over deGrom. I am talking about 2018 Max Scherzer over deGrom.
And I do want pitchers that play on better teams if they are part of making that team better. Jose Quintana puts up good numbers, but he doesn’t make a team better. A pitcher has to have a killer instinct and that is not always quantifiable in any way other than wins. Some players thrive on not being the best in the league, but being better than the other guy on that day because in reality, that is all that really matters.
I actually built an online WAR calculator years ago.
For those interested in playing with it, check it out at ryantansey.com/simplewar.html
Ah yes the condescending nerds who call RBI’s and ERA “prehistoric stats”
We get it, you read fangraphs.
It’s not unusual Kenneth. About a month or so ago I mention that I found it amusing that a poll was done of 70 players (35 pitchers and 35 position players) asking what stat the pay attention to the most, the amusing part was that not 1 mentioned WAR. The first response given was (I am paraphrasing) “most players are not smart enough to understand it”
Do you not understand that there is much better stats to use?
Do you not understand that some people don’t care about those stats and “better” is in the eye of the beholder?
JDGoat…..Do you not understand that all stats have value and when your reply to those who view things differently is insipid or condescending you come off as……..ummmmmmm…….. let’s just say you remind one of a feminine product a woman may use on a summer’s eve .
There’s no need for people to call others nerds for understanding that there is much better stats than RBI’s and ERA to judge players.
And there is no reason for you to jump into every conversation with your brand of condescension when someone does not toe the line of your choice, using terms such as prehistoric, old, stupid etc. You give off the impression that you are lurking, rubbing your hands together with a mad scientist grin just waiting. I can almost hear you saying “Oh goody….another one I can insult and prove to them I am the smartest.” Get over yourself, you aren’t the GOAT.
Have a nice day.
My problem is that people are fine pointing out the problems and downfalls of advanced stats and then completely ignore the same problems with the old ones. I’m actually someone who’s in the middle and uses both but it’s terrible to argue with when someone completely writes off one or the other.
“when someone completely writes off one or the other” you mean like you do when you attack anyone who mentions ERA, BA or RBI.
And by the way, just because someone has a few years under their belt does not make them a dinosaur, antique or stupid. I had the privilege, years ago, of living a few doors away from HoF’er Judy Johnson. I was blessed to spend a few evening sitting on his front porch talking baseball. Even at his advanced age he knew more about the game than you or I will ever hope too.
If you truly are “someone who’s in the middle and uses both” then phrase you comments that way instead of always being a “D”
I’m done, this has run it’s course…have a nice day.
Why are we arguing about stats. Stats are the past.
No matter who’s hitting or who’s pitching or who’s fielding or who’s running
The right plays just have to be made.
Sometimes it’s the 200 hitter driving in a key run
Sometimes it’s the poor D corner OF making a play
Sometimes a Molina steals a base or hits a triple
Sometimes a crappy P gets a key out
No stat can predict the future. It’s why we as fans watch the game.
Imo all stats have a place tho.
Avg. Because you need to get hits to score runs. Growing up watching John Olerud chase 400 was exciting. I wouldn’t care if it was all bloop singles or crushing double after double in the gap watching a player hit 400 would be amazing
RBIs. Because you need runners to score
Saves. Because Mariano Rivera showed how effective it is to play 8 inning games
Win/loss to pitchers. I get it’s a “bad” stat to show how effective a pitcher is. But still is fun to watch good pitchers chase 20 wins and bad pitchers will still rack up more Ls than good Pitchers
War is funky too. And certainly also has its place. It’s better to have a unique all encompassing stat to measure players quickly and easily
And I get the arguments above about the “old” stats. But that’s why we as fans shouldn’t care about stats as much. We know from watching games if player A or B isnt “clutch” or gets hits with no one on and struggles with runners on.
I guess you only need the stats when you want to follow the entire league.
Just enjoy the games. Watch for those improbable things to happen. Leave the stat stuff to the front offices who get paid for it.
“RBIs. Because you need runners to score ”
Let’s take the argument to the ultimate extreme: There is always two outs when the batter comes to the plate and never a runner on base. The batter always hits a triple but the next batter pops out on the first pitch. Let’s say the batter has 600 at bats during that season, so his slash line is 1.000/1.000/3.000/4.000, but he has no RBI and no runs scored. He has just had the most fantastic season ever seen at the plate but he has contributed nothing towards helping his team score runs.
I’ve been playing with baseball stats for more than 60 years. When I first started I had box scores from newspapers and their Sunday listing of player stats (the usual data that everyone thought important for so long) and the score cards from games I listened to on radio. Everything I did was by hand.
As time went on, home computers appeared and when I got one I had to write a program to handle baseball data analysis as well as still inputting the data by hand. The internet blossomed and I could download data for analysis and eventually I could read what others had done with their research. But even with that, so much of baseball’s history has been lost that it is not possible to truly compare today’s players to those in baseball’s past. We don’t know how many times Babe Ruth batted with runners on 2nd and 3rd in 1924. We don’t know the speed/break of Walter Johnson’s pitches.
If you have used stats much you would realize that an argument using an anecdote that’s about 10 sigma from the mean really isn’t valid. The problem with WAR is it never recognizes anything outside of 1 standard deviation (sigma) let alone 10 standard deviations.
Anecdotes of extreme conditions are fun but don’t move a discussion forward when looking at statistics.
Also in baseball virtually all the statistics are too short a sample to be valid for much use anyway.
In your ultimate extreme argument, that batter has essentially done nothing. It’s all flash and no substance. Because he looked good without contributing to a team winning. It’s the equivalent of a basketball player scoring a lot of points in garbage after his team is already well out of it.
I’m looking forward to finding out if there is ANY reason to consider WAR a valid stat. I’d like to understand if it’s retrospective only or is it prospective too. Is it simply a comparison tool and if so who gets included in the averages used.
For any of us fantasy players there is always a question of determining which player is farthest from the mean. If you are in a league with 10 teams then you must decide if the mean is calculated by using the top 10 players at a position, or maybe the top 20 players at a position. There is no right answer just a preferred answer by each person doing the calculation. Is WAR similar? Is the average range in CF calculated by ALL CFs, the top 30 CFs or using a different sample between the two. Are positions in the shift differentiated so if TBs OF plays closer to each other than LAA’s OF does that hurt their numbers?
So much of WAR is averaging averages and the constants are based on the opinions of the people performing the calculation that as mentioned there are three sources of the number who don’t agree on the constants or the measurements so how valid can it be?
It seems like a nice science project to estimate value of players but suggesting a cumulative WAR should determine if a player should be in the Hall of Fame seems ridiculous especially when comparing a Babe Ruth WAR to a modern day WAR. How can there be sufficient data to provide fielding WAR? So if you can’t truly calculate the fielding component of WAR, why include it?
The other question I hope to have answered is how does the strength of positions affect the overall WAR numbers each year. For years there were very few great shortstops and center fielders and lots of great first basemen and corner outfielders. Should your WAR exceed a better player because you play a week position? Trout vs Miggy is a great example. Back when Ozzie Smith played should his WAR have been higher than great 1Bs of that time simply because he had limited competition at the position?
I hope the series answers all these questions!! Tim, this was a great idea!!
The average range in CF is calculated based on the average of all center fielders. UZR and DRS do account for shifts in their numbers, although this is not a simple or perfect task admittedly.
Offensively, the weights are almost all agreed upon, so the main differences in WAR arise from the two main methods using a different statistic for the defensive component (UZR for fangraphs, DRS for Baseball Reference). The two use a similar but slightly different ideology, and over the long run the differences between the two almost always even out. The main difference is that DRS tends to have a higher range of values (higher for the best players, lower for the worst). Moreover, the fact that there are different yet simultaneously valid approaches is something that I believe is a feature of WAR, not a bug. It demonstrates that WAR is not perfect and that reasonable people can disagree on how good a player is, but gives a reasonable range of values to get a sense of the player’s value. No sabermetrician worth his or her salt would ever tell you to rely strictly on WAR, because it’s inherently a very broad tool, but it’s the best broad tool we have right now.
Strength of position does not affect the WAR calculations. There’s a certain amount of WAR dished out each year (I think around 1000), a number which was determined based on a lot of discussion about how good a replacement level player is (and thus how good a team filled with replacement level players would be). So league-wide strength could affect WAR. However, WAR is not rationed out evenly amongst each position every year. The positional adjustments are made based on relative strength of positions over a huge sample (think decades).
Hope this helped answer a few of your questions, I’m sure Tim will do a better job addressing them than I did.
the fact that there are multiple versions of how you calculate WAR tells me it’s an irrelevant stat. Correction it’s not a stat. Stats are consistent across all platforms.
You can put your hands over your ears and pretend it doesn’t exist if you please. It’s the best way to calculate a player’s value. There’s some sort of error bar you need to account for but everyone should understand that.
If war is so great, why is the game losing popularity? It’s because people are bored with the nerds telling the players how they should hit. Baseball is a great sport if they would allow it to be played multi dimensional. Right now, it’s unwatchable.
I dislike the current versions of WAR as they rely of too many generic assumption in rating players, especially when used to compare players that play different positions.
It’s like a presidential election or the polling for one. It lacks important context. Yes you may get more votes nationally but it’s the context of where you got the votes that determines the electoral college and who wins. And WAR is about wins.
The biggest problem is the generic positional adjustment. A CF gets about a 0.9 boost over a corner OF. In a generic way this is true but it misses the context of WHO you play for and the composition of the team. To use Trout as an example, he plays CF in Anaheim because he is the best OF on the Angels and gets the CF player position boost because of it. If he were to be traded to say, Tampa or Toronto or Boston he would be a corner outfielder because Kiermaier, Pillar and Bradley are all much better defenders. And he would get docked on WAR because of being a corner OF not because he was any worse of a player. Same as a player could be a top 5 SS but if traded to LAA or CLE He would end up playing 3B. AS there is a park factor used in WAR there also needs to be a Team factor.
The second problem I have is using the same generics as to the value of a walk. It’s overrated to winning. If you look at the Astros of 2017 or the Red Sox of 2018 one of their keys to winning is putting balls in play alot and striking out and walking very little.
People discount RBI as it’s supposedly a team stat vs an individual stat and granted total RBI is heavily dependent of a players position in the batting order and the quality of the rest of the other players. But there are “RBI like” stats that are controlled by the player and are player stats that should be but aren’t used. One I like is OBI% that Baseball Prospectus uses. It’s what % of runners on base does a player knock in. Basically percentage of RBI minus HR divided by chances. In looking at the top 3 players in WAR in the AL this year. Betts has knocked a runner on base 17.2% of the time, 186 ROB and 32 OBI.
Ramirez was at 17.1%
Trout although he has had 234 ROB only knocked in 30 for a 12.8% OBI%
The League average OBI% is 13.8% so Trout is actually below average in knocking ROB in for runs.
Trout draws a lot of unintentional walks with ROB that don’t create runs or wins but all the WAR calculations give too much value to those walks because of the lack of context to them.
Like park factor modifies WAR there needs to be context as to teams and results added to WAR for it to have much more value than say pitcher wins.
I like what you’re saying, and I also agree War is way too simplistic to be the be all stat everyone tries to make it.
The smart thing to do is just use WAR as 1 of many statistical evaluations. Problem with that is every stat is really subjective.
Are we gonna start discrediting certain singles bc they happened against a team that is or isn’t shifting, or bc a player was on base leading to a bigger hole on that side of the field?
Personally I wish there was more statcast type stats. Regardless of position how many 40% catch prob balls is player A recording an out on vs the league average.
Based on exit velocity how many times is player A giving himself a better chance for a hit on a ball in play then league average.
I think if the general public had more numbers given to them that front offices around the league use I believe someone eventually would come up with a better formula, but until that happens Idk if there will ever be any 1 stat we can use to really judge true value.
Someday maybe, but until we get a really 100% accurate assessment of what each outcome is truly worth there is no end all be all stat.
Example…
What is the real value of a triple? Some triples are actually more valuable then others. A triple with 2 outs is honestly not much more valuable then a 2 out double. Is a 2 out walk with men in scoring position from the 8hole hitter really as valuable as a single?
In a game where a ground out that moves the runner from 2nd to 3rd is more valuable then a walk it really becomes difficult to assess a players true impact.
I once ask a nerd of an organization about War. He struggled to give an answer. Finally, he said that it was about a player being replaced by a Minor Leaguer. Then I asked him the following, “when Wally Pip got hurt and he was replaced by Lou Gehrig who had not had one at bat, I asked him what was Lou Gehrig’s WAR. He had no answer. The fact is that WAR is very subjective and you do not know what that replacement will do until he is given a chance in the big leagues. War can only be done with comparing players in the big leagues. Question, what was Fowler’s War when the Cardinals gave him all that money and what is his War today?
What if the guy got drilled in the back by Randy Johnson and couldn’t hardly put on his socks in the morning yet still suited up…..and was terrible. Sample size is a big factor that is often overlooked when mentioning ANY stat.
Also, part of WAR is the baserunning adjustment.
Should players that face Yadier Molina and Wilson Contreras 20+ times a year be penalized if they get excellent jumps and have plus speed but still get caught because….well those are two great defensive arms behind the plate.
That same player might have the same leadoff, same 1st to 2nd time and be 20/20 on attempts in another division….just for example.
There are so many examples of this included in almost all stats that they need be taken with only so much weight. A balance between stats and the eye test is usually the safest method.
Does WAR figure Jon Lester couldn’t hold a runner on base? How about when a runner steals 2nd base uncontested because the catcher chose to hold the fast runner at 3B to keep him from advancing home….is that saavy baserunning and inflated stats?
All of these instances are minor and very random but they ALL add up to eliminate any true stat from being “end all, be all”.