Guide to Stats
Over the course of this season, and hopefully many more to come, we at River Ave. Blues (and by “we” I mean Ben and me; Mike doesn’t believe in this VooDoo bullshit) will be using various statistics to back up our arguments. While stats don’t always tell the whole story, they’re very useful in helping illustrate a point. If I read a mediot extolling the virtues of, say, Miguel Cairo, I like to look at his statistics and say, “uh, dude, this guy makes a ton of outs.” Or something along those lines.
However, the traditional statistics — Batting Average, Home Runs, Runs Batted In — don’t tell the whole story. In fact, using only those three stats can be very misleading when talking about a player’s value. Yes, it would be nice if Player A drove in 110 runs this season, but his ability to do so hinges on the work of the hitters around him — particularly those ahead of him. Who knows: they could have plugged in a better player at that lineup spot and he might have had 130 RBI.
What those traditional stats lack is context. For a simple and more full explanation of traditional stats, I urge you to read Dayn Perry’s explanation at Baseball Prospectus (this article, along with the one linked later on, is free).
Okay, done? Good. Now let’s take a look at some of the stats you will see here on River Ave. Blues this season:
Batting Stats
WPA
Win Probability Added (also: Win Expectancy). You see this one after every game. There have been intricate explanations written about it, but I prefer the stripped-down version.
WPA tells you what a team’s odds are of winning a game in a particular situation — any situation. This is based on inning, base situation, outs, and score differential. Enter all of those circumstances into the WPA calculator, and it will spit out your team’s odds of winning at that moment.
For instance: Yankees at home and up by one in the bottom of the seventh. Damon grounded out, Jeter singled, and Damon walked, so we have first and second with one out for Alex. Punch those numbers in, and you’ll see that this situation has occurred 843 times since 1977, and the home team won 720 times, meaning the WPA for that situation is .854.
If Alex hits into a double play (worst possible scenario), the situation changes to top of the eighth, none on, none out, visiting team down by a run. That has happened quite often, 7,056 times since 1977, and the home team won 5,441 times, meaning the WPA right then is .771 (Yankees have a 77 percent chance of winning). For Alex, that means that he lessened his team’s chances of winning by .083 (8.3%), so that is debited to his running WPA total. Conversely, if he hits a home run in that situation, the WPA becomes .977 (2,356 games played, 2,302 won by the home team), meaning that Alex added .123 (12.3%) to his team’s chances of winning, which is credited to his WPA total.
So when you look at the graph, that shows you the progression of WPA throughout the game. The table below the graph just tallies each player’s total contribution.
Why it’s good: It measures the contribution each player made to each game. It uses historical data, so it’s not like these probabilities are just made up.
Why it’s not so good: It values a home run in the bottom of the ninth of a close game much more than a home run in the bottom of the first, even though the home run in the bottom of the first might be what kept the game so close in the first place. It also places value on things over which the batter has no control, like the base situation and number of outs. It also doesn’t take into account other factors that affect a team’s chances of winning a game, like, for instance, the opposing pitcher (your team will tend to fare much better against Sidney Ponson that Johan Santana).
Why I use it: It’s fun. It has neat graphs that I can insert images into. It tells a game story well. I think it will evolve over time to become a more telling stat.
LI
Leverage Index. This measures the gravity of each game situation, based on a scale from zero to five, with five being the most critical situations and one being an average situation.
I’m not quite sure of the calculation (mainly because it’s based on the scale), but I know that the root of it is the current WPA situation vs. the worst case scenario. The greater the difference between the current situation and worst case scenario, the higher the Leverage Index.
Why it’s good: It allows us to more objectively define the most high pressure situations (I won’t say “clutch,” I won’t say “clutch”).
Why it’s bad: In itself, its not bad, but I’m becoming less and less a fan of the average LI for each player, pLI. If a player faces a decently high-pressured situation in the third inning and fails, but then bats in much lower pressure situations later in the game, his pLI may average out to around 1.00. And that’s not accurately judging the situations the player hit in.
Why I use it: Because I hate people citing performance in “close and late” situations as a measure of clutch.
EqA
Now we get to the mathy stats. EqA, or Equivalency Average, is much like batting average, except it is just a bit more complex than H/AB. However, it is based on a scale similar to batting average, where .260 is average, .300 is very good, and .320 and above is freakin’ excellent.
The great thing about EqA is that it takes into consideration all that a player does on offense: hits, total bases, walks, and stolen bases. To quote BP’s Clay Davenport on EqA:
Simply put, it’s more accurate, it’s unbiased, and it models the scale of batting average, so it’s easy for a new fan to understand.
The formula is as follow:
H + TB + 1.5(BB+HBP) + SB -------------------------------------- AB + BB + HBP + CS + (SB/3)
If you’re wondering why hits are valued over walks, you can read the article linked above. For the short, short version, it’s to balance out the hits and walks, since the denominator doesn’t contain all aspects of plate appearance (sacrifices, most notably).
Before the final number is published, it is normalized for league difficulty and park factors. This is what scales it so that we can separate an average hitter (.260 EqA) from an excellent hitter (.320 EqA).
VORP
Famously chastised by Murray Chass, VORP, Value Over Replacement Player, is a counting stat, much like home runs, RBI, stolen bases, etc. What it counts is the number of runs a player adds to his team over the course of a season.
The big question here: what is a Replacement Player? You mean a scab? No, no. What is meant by “Replacement Player” is a substitute for a regular — a AAA call-up or bench player, if you will. This hypothetical player is measured as roughly 80 percent of league average, giving those average players the props they deserve. The number actually fluctuates, starting at .75 percent of league average for the tougher defensive position (catcher, shortstop) and .85 for the easier ones (first base, DH).
Why Replacement Level, though? Because when a player need to be replaced, most of the time a player must do so with a player below league average. VORP, therefore, calculates the value a player adds over the schmuck that would play his position if he were to succumb to injury.
Keith Woolner of Baseball Prospectus came up with the stat, and you can read a simple explanation here. Wikipedia also has a decent explanation.
For our purposes at RAB, VORP will be used only to measure the value of hitters.
OBP, SLG, OPS
These stats are a bit more common, and will likely be used more frequently than their more complex counterparts. If you don’t know the drill:
OBP: On Base Percentage (a.k.a. OBA; On Base Average). In a game of baseball, you have a mere 27 out, and only three before you have to erase all men from base and start over. They are the most valuable commodity in the game, and the team that best conserves them is the team that, in the end, is going to score more runs and win more games (though much of that depends on pitching). OBP measures how many times a player didn’t cost his team an out, therefore bringing them closer to the end.
SLG: Slugging Percentage. Instead of dividing hits by at bats, SLG divides total bases by at bats. Total bases are assigned as 1 for a single, 2 for a double, 3 for a triple, 4 for a home run. Unfortunately, these are rather arbitrary assignments. But it does give us a rough idea of a player’s ability to hit for extra bases. Isolated Power (ISO) is a derivative of SLG. It subtracts AVG from SLG, taking away singles and allowing a measure of pure extra bases.
OPS: On Base Plus Slugging. Just like the acronym suggests, it adds OBP and SLG to form a crude statistic. It’s very arbitrary, but it serves its purpose in allowing us to roughly figure out who is adding big value to the team. The best things you can do in baseball are 1) not make outs and 2) hit for extra bases. Players who can do both are the most valuable to their teams. OPS helps us determine that.
A note on OPS: though its nature is arbitrary, it does have some practical use. As noted in Moneyball, it is the only individual statistic that has a correlation to a team’s success. I’m sure, though, that the author, Michael Lewis, wasn’t familiar with newer stats like the aforementioned EqA and VORP at that point, which may have a greater correlation (though I’m not willing to do that research at this point).
Pitching Stats
Because much of pitching hinges on defense, it’s tough to come up with a truly effective pitching stat. Yes, we have the old, trusty ERA, but that’s flawed, too.
Yes, it measures the earned runs allowed by a pitcher, which lets us judge his output. However, as we mentioned, if defense plays such a role in pitching — or the results of pitching — then ERA may not tell the whole story. If a finesse pitcher is playing with a crappy defense behind him, his ERA is likely going to be higher than if he had a good defense behind him. Therefore, his ERA is not a measure of his ability, but rather the ability of him and his defense together.
Another bone to pick with ERA: why is a pitcher not responsible for runners who reached on an error? He gets credit for making an out when a fielder makes a spectacular play, so why not debit him for the reverse? Runs Allowed Average paints a much fairer, if not clearer, picture of a pitcher’s output.
This all feeds into the argument that a pitcher has little to no control over what happens to a ball once it is put in play. They do, however, have some control over what type of ball is put into play. So let’s start there.
LD%: Line Drive Percentage. These are the killers. The more line a batter hits, the more they’re going to dunk in for hits (which is why hitters who hit more line drives have higher batting averages on balls in play). Keeping line drives down will keep hits down, which in turn will keep runs down. For instance, to go along with his amazing groundball rate, Chien-Ming Wang had a 16.9% line drive rate, ranking him fourth in the AL (Randy Johnson ranked first at 15.8%).
GB%: Keep the ball on the ground, and you’re assured it won’t leave the park. You may be at the mercy of your infield defense, but they’re bound to convert a good majority of those batted balls into outs. A pitcher that combines a high groundball rate with a low line drive rate is likely to do very well, even without striking guys out (once again, see Wang). Only five AL pitchers cracked the 50% barrier in ground balls in 2006: Wang, Jake Westbrook, Felix Hernandez, Roy Halladay, and Kenny Rogers.
Now we move onto a few defense-independent pitching stats that are a bit more familiar:
K/9: Strikeouts per nine innings. One thing a pitcher has some control over is his ability to miss bats. If the ball isn’t put into play, you’re not at the mercy of your defense. As mentioned in the batted ball types, pitchers can get guys out other ways. It just so happens that the most effective way to do so is the strikeout.
However, K/9 misses a vital aspect: efficiency of strikeouts. Yes, we look at it in conjunction with K/BB ratio (to be discussed momentarily), but we can figure out more about strikeout efficiency by simply changing what we measure it against.
In K/9, we’re measuring strikeouts per inning, and multiplying it by nine to give us a more workable number. That doesn’t take into account how many batters the pitcher faced in a particular inning, nor does it consider the number of pitches thrown. The latter is the most precise, so we can use that: strikeouts per 100 pitches thrown. This is especially nice because of the emphasis on pitch count, and getting a pitcher out at around 100 pitches.
Yes, any kind of strikeout is nice, but the best pitchers put guys away with the least amount of pitches.
BB/9: Walks per nine innings. We all know the drill: walking guys is bad, so the lower, the better.
I’m not sure if it makes much sense to take walks per 100 pitches. At first, I thought that the more pitches to a walk the better, since that means that strikes were thrown. However, a walk is a walk is a walk. In fact, a higher pitch count walk is plenty bad, since that means all those pitches were wasted; the runner got a free pass in the end, anyway.
I’m thinking about looking at a percentage of walks per batter faced. That may be a more accurate way of looking at walks.
HR/9: Home runs allowed per nine innings. A defense can’t do anything with a ball hit out of the park, so this one is all on the pitcher (and the park). This is why flyball pitchers aren’t very attractive: fly balls can leave the park.
Thankfully, though, we’ve got another stat to help us with that: HR/FB%. Simply, it’s the percentage of fly balls allowed by a pitcher that leave the park. Some guys just induce lousy contact. Once again, we have Wang, whose HR/FB% was 8.2%, good for third best in the AL. I
K/BB: Strikeouts to walks ratio. Walking guys is bad, but some guys can make up for that by striking out a ton. They’re the two things a pitcher can do without having the batter making contact, so it makes sense to compare the good in relation to the bad. If a guy is under a 2:1 ratio, it’s a huge warning sign.
Defensive Stats
There aren’t many readily available defensive metrics out there, especially those measured in-season. Hence, we base a lot of our defensive evaluation on anecdotal evidence, which is why Derek Jeter has won three straight Gold Gloves despite his below average fielding at his position (you’ve surely been asked this question before: how many times do you hear Kay or Sterling say, “Past a diving Jeter”?).
One way to look at defense is to look at the team as a whole. That’s where Defensive Efficiency Ratio comes into play (DER). It’s quite simple: number of balls in play divided by the number of them converted into outs. It’s binary, so there’s no real dynamic to it, and it’s actually best for aiding in the evaluation of finesse pitchers (like the Tigers last year, who converted a whole helluva lot of balls in play into outs, thus keeping their finesse pitchers’ ERAs low).
The only readily available fielding statistic that has much credibility is Zone Rating (ZR), which can be found on ESPN.com. With Zone Rating, the field is divided into a bunch of different zones (not sure exactly how many). Each player is assigned a zone in which he is expected to make a play if a ball is hit there. If he misses a play in his assigned zone, his ZR drops.
The biggest drawback to this is that it doesn’t give extra credit for plays made outside of a player’s zone. If Omar Vizquel makes a play outside his zone, it’s counted as any other play made would be, instead of giving him credit for a play made out of his zone. I still don’t understand why this hasn’t been corrected yet.
That’s all…for now
This started as a response to commenter Richard C.’s request for an explanation of WPA and LI. I figured that explaining the other stats we use would be helpful, too. Obviously, I didn’t intend it to be this kind of undertaking. However, I feel comfortable with the information I’ve presented here.
I’m always reading about the evaluation of players and teams, so I may latch onto another statistic I feel is relevant. If that happens, I’ll add it to this post and mention the update on the main page. I definitely suggest bookmarking this entry so you can reference back if something we write is unclear.
If anyone has any questions about anything presented herein — or anything I might have missed — send me an e-mail (RABJosephP (at) gmail (dot) com) or leave something in the comments section. This goes especially for any fellow stat nerds out there who feel that I did not do their favorite stat any justice. I believe that I summed up everything accurately, but if I’m misinformed on some aspect, I’d like to be corrected.
