/cdn.vox-cdn.com/uploads/chorus_image/image/52211555/676431.0.jpeg)
We here at Lookout Landing have a proud tradition of being at the forefront of the advanced stats movement in baseball. Developing and utilizing the most sophisticated means of statistical analysis is a crucial part of doing in-depth analysis. As a lifelong Mariners fan and reader of multiple excellent M’s blogs over the years, I, like many of you, was fortunate enough to be raised with saber-joy, and not saber-fear.
With that said, being raised with sabermetrics has still left some holes in my own statistical understanding. This is a photo of me in my AP Statistics class my senior year of high school, in which you can see the numbers of days in which different individuals were late, listed by their initials on the whiteboard (I’m so sorry Ms. Farmer). [ed note: he is for real.]
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/7601063/AP_Stats_Class.jpg)
I don’t really have a great answer as to why some of the more complex concepts still cause me some difficulty.
I suppose I may never know.
With this in mind, today we are launching this series, wherein I will attempt to give a wieldy, basic explanation of the metrics that we rely on most here at LL. If you are someone who is very comfortable with your sabermetric chops, one, awesome, and two, this may not be for you. It may, however, be an opportunity for you to take folks under your wing and help bring more of our community into our wonderful stat-ernity, no hazing necessary.
Today we are going to focus on two very core concepts: WPA (Win Probability Added) and WAR (Wins Above Replacement).
WPA
After every game we share a chart like this one, which measures the win expectancy as a percentage for the Mariners and their opponent.
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/7601225/Massive_Comeback_Chart.png)
Y’all may recognize this game as one of the most improbable comebacks in MLB history. We are going to go through a few of the key moments to try and highlight the way WPA and our postgame charts can be useful.
What this Win Expectancy chart attempts to show is, simply, the likelihood of one team or another to win. These charts always begin with the odds at 50/50, which is an oversimplification, of course, since a game where Jake Arrieta and the Cubs face Jered Weaver and the Angels isn’t likely to be an even fight, but that will more likely than not quickly bear out on the chart as the Cubs pounce on Weaver.
Where WPA comes in is at the individual level. WPA does not measure defensive contributions, which is one of many reasons it is useful for creating a story, but less so for predicting future production. Each play the hitter and the pitcher have the chance to help their team towards winning or not. The closer the game is to its conclusion, the more weight each at-bat has, because there simply are fewer opportunities remaining to produce runs. Equally intuitively, when the score is close, each at-bat is more important to the outcome of the game, and therefore the WPA at stake is greater.
Look, for instance, at that first little dip in the line towards the Mariners side in the first inning. That’s a Kyle Seager single that drove in a run to give the M’s a 1-0 lead. Dope. According to the chart, that increased the Mariners chances of winning from 49.7% to 57%, so Kyle gets awarded .073 WPA (7.3% more likely the M’s win). Good work Kyle, we are very proud of you. Unfortunately, the game gets a bit out of hand as Wade Miley gets utterly shelled and gives up nine earned runs on 12 hits in 4.2 IP, which is, incidentally, good for -.457 WPA. In this game Wade Miley made it 45.7% less likely the Mariners would win.
Well screw you too, then.
Midway through the 5th inning, this game is over, sensically and mathematically. The Padres reach 99.9% win expectancy as their lead reaches 12-2. The line hugs the top of the chart, and is barely impacted by two big plays. Kyle Seager hits a double that drives in two runs in the 6th, and Dae-Ho Lee follows with a three-run home run. Five runs, a massive contribution, yet the line barely inches south. It certainly didn’t look insignificant.
Seager’s single in the 1st gave him .073 WPA, yet these plays combined for just .022 WPA, because, with the brief time left and the still steep deficit, the Mariners hadn’t done much yet. Of course, we know that wasn’t the end, and the biggest shift of the game came from (this game really had everything) Stefen Romero’s single in the 7th inning, which tied things up and kept the bases loaded, which was worth .228 WPA by itself, improving the Mariners’ odds of winning by 22.8%. That RBI single, of course counts no more than Seager’s first inning RBI, but it came at a pivotal point late in the game.
In Summary
If you missed the game, WPA and Win Expectancy charts are a great way to get a read on the key moments and major contributors. If you saw it, they will often confirm what your eyes told you, but may point to specific moments as notable turning points.
WAR
Wins Above Replacement (WAR) is the metric that likely symbolizes sabermetrics more than any other statistic. A single number, combining several measurements, that represents the value a player has had. WAR is ambitious, and not entirely precise, but is a good enough statistic that it is worth leaning on. There are two different popular equations for WAR. One is done by Baseball-Reference.com, commonly shortened as bWAR or rWAR, while the other is done by the afore-mentioned Fangraphs.com, which is listed as fWAR. The different equations are outlined here and are more similar than they are disparate. It is useful to understand the differences between them, but most important to at least recognize there are multiple sources of evaluation.
I am standing by our minimal math mantra here, so while the equations used to calculate WAR includes several distinct statistics being combined, and different calculations for pitchers than those used for hitters, we will go into some of those stats on their own later in this series, and just focus on how to use and interpret WAR.
Study break.
Alright, I feel better. Just a few things here.
First, “replacement” level does not mean you, the reader, unless you are Stefen Romero or Joe Wieland. I am sorry, but the mathematical powers-that-be decided that “replacement level” refers to a roughly AAAA-level player - someone not quite good enough to be trusted with an MLB job, but talented enough that teams will sign and play them if an established starter gets hurt.
Second, WAR, like hits or RBIs or K’s, is a stat that accrues over the season. Unlike those other statistics, however, you can lose WAR as well. Have a great game defensively and hit a couple home runs? You’ll see your WAR tick up. Strike out four times at the plate or, as a pitcher, have a game like Wade Miley did in the game we discussed earlier? You’re going to see that WAR total slide back down.
Third, what is considered “replacement level” varies from position to position. Shortstop, for instance, is a position that asks a lot of a player defensively, and generally requires less offensive prowess to still be considered a useful player. Conversely, first basemen can look like Prince Fielder, and are there to hit, and hit well. One of my favorite examples of this, as well as the way WAR can help us compare the value of players with vastly different skill sets, is comparing speedy OF Juan Pierre and slugging OF(if you really hated your pitching staff)/1B/DH Adam Dunn, as is done expertly in this piece by Ben Lindbergh. If you want the instant gratification, spoiler alert: in their 14 year careers, both players ended up earning identical value, with 16.6 WAR.
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/7615257/113307864.jpg)
Below is a chart from Fangraphs, showing a general appraisal of the value of a player, based on their WAR over a full season.
Now, lets compare those numbers to the 2016 Mariners’ starting pitchers, shall we?
As you know, having watched the Mariners last year, not great. Not only did just two Mariners starting pitchers grade as “solid starters”, four players were actually below replacement level in their performances, albeit in brief periods of time. A bright spot, shining at the top of the list is James Paxton, whose fiery heater and fiendish cutter made him as valuable as a “good player” despite throwing just roughly two-thirds of a full season. If Paxton could continue that production for an entire season, he could end up in the “Superstar” range, which I have been told is a good place to be.
Position players’ WAR delivers information that is also likely in line with what you’d expect. Robinson Canó? Like sweet summer rain. Ketel Marte? Like being caught in a hail storm in a t-shirt.
In Summary
WAR is a core statistic that attempts to isolate that player’s individual performance from their teammates and gives a good general sense of the productivity of a player. Over the course of a season and a career, players will gain and lose WAR. While useful as a macro-level view, finding other trustworthy statistics to pair with WAR creates more useful analysis. As we break down this series of articles into Batting, Pitching, Fielding, and Baserunning, we will share some of the numbers we trust most.
Hopefully this word-heavy but numbers-light overview has been, and will continue to be helpful. Especially if you are someone looking to strengthen your grasp on these concepts, feel free to let us know what is helpful and what is not.