Let me start off by saying that I am no statistical guru but I think it’s important to understand the value of some of the breakthroughs statisticians have made when it comes to evaluating the game.
This isn’t going to be a detailed study and there are some of you out there that know a lot more than I do, but my aim here is to try give a basic primer on some of the stats we use and why they’re important. I’ve done this before but the blog has grown exponentially since then and I thought it was a good time to revisit the topic. For some, this is new and for others it’s old hat, but I hope it makes for good discussion either way.
We will use Fangraphs rather than Baseball Reference for the purposes of this article because I feel they do better at isolating individual production from luck.
On Repeatability, Predictability, Luck, and Myth
Few things are more important in evaluation than separating what is within a player’s control and what is due to luck and environmental factors. The reason this is important is because you want to be able to find out what is repeatable and thus what can be used to predict future success. In many ways, it’s what we talk about often when we talk about rebuilding in that we are stressing things that tell us about process rather than results. Process is repeatable, results can vary based on luck and environment.
When it comes to hitters we can look at plate discipline factors, contact rates, and the quality of contact. These numbers tend to stay somewhat constant, though they can also improve or regress over time — and they often tell us a lot more about where a player is headed than statistics that more directly incorporate outside factors, such as RBI and even batting average.
We can take a look at Cubs first baseman Anthony Rizzo to illustrate. Where Rizzo has shown impressive improvement is with his plate discipline. If we look at his O-Swing% (percentage of swings on pitches outside the strike zone), we can see that it has dropped to an outstanding 23.1% — this is a remarkable improvement over his 2012 rate of 38.5%. Not coincidentally, his walk rate has increased to 16.5% over his 7.3% mark in 2012.
But that plate discipline goes beyond chasing bad pitches, Rizzo has become more disciplined within the strike zone (Z-Swing%)– and that has manifested itself in a different way — a higher contact rate and more quality contact. We see a big jump here over last season. His K rate has gone down to 14.8% from 18.4% last year. Rizzo’s line drive rate (LD%) has gone up to 23.4% from 19.6% last season, while his pop-ups (IFFB%) have dropped from 9.9% to 3.6%. We can take this to mean that Rizzo is laying off pitcher’s pitches he can’t hit hard and waiting for a specific pitch he can drive — and so we’ve seen his HR rate per fly ball increase, We all know how good Rizzo has been and we can directly trace it back to the improvement in his plate discipline, contact rate, and quality of contact, all of which are intertwined.
We often use BABIP which stands for batting average on balls in play. We use it because it filters out luck (both good and bad) such as bloop hits or line drives hit right at someone. The league average BABIP tends to hover around .300, but we have to be careful not to take that number too literally. BABIP can vary from player to player because of things like speed (more infield hits) and quality of contact (i.e., more line drives, less pop-ups) . Good hitters tend to have higher BABIPs and we see that when we look at the average BABIP of the top 10 NL hitters last year, which was .362. Some of that is luck, some of it is better consistent contact and/or speed. It can often tell us how well a hitter is doing more than batting average. An unusually high BABIP for that player often means we can expect his average to go down, while an unusually low one often means we can expect the hitters average to improve over a larger sample size.
We also like to talk about P/PA, which is pitches per plate appearances. Players like Anthony Rizzo (4.26) and Luis Valbuena (4.48) see a ton of pitches, so it’s not surprising that both players take a lot of walks, Also not surprisingly, Starlin Castro sees almost one less pitch on average per AB (3.51), a significant difference for this particular statistic.
This explains (in part) why we talk about hitters having good ABs vs. bad ones — though again, nothing is set in stone. If you know a pitcher likes to get ahead with his fastball and you jump on it and deposit onto Waveland Ave., then I think we can agree that’s a pretty good AB. But when we talk about stats, we speak in terms of generalities and trends. There will always be shades of gray. The important thing to take is that a good approach is repeatable even if the results are subject to outside factors, though good process should yield good results much more consistently than bad process.
With pitchers it used to be common to focus on wins and ERA, but those two statistics depend at least in some part to the quality of the offense and defense behind them. With wins, run support is an obvious factor. You can’t win many games if you’re offense isn’t giving you runs. The idea that pitcher pitch to the score is a myth (see Jack Morris). It’s less intuitive with ERA, but if your defenders get to less batted balls (resulting in a high BABIP for that pitcher), then the pitcher is going to give up more hits, and in general, if he gives up more hits, he’s going to give up more runs. That is because a pitchers LOB% (left on base percentage, also known as strand rate) tends to remain pretty constant (around 70-75%). A pitcher can be unusually lucky or unlucky with men on base and that can affect his ERA, so both BABIP and LOB% have some basis in luck and that extends to the ERA statistic. Obviously the size of the ballpark can affect a pitchers ERA as well. For example, Andrew Cashner’s ERA in spacious Petco is 1.77, while it is a more mundane 4.16 on the road. Cashner is still a good pitcher overall, but if we use xFIP, we can see it is higher than his ERA, in part for this reason (3.39 FIP vs. 2.72 ERA).
So we focus on what pitchers can control by isolating what he does individually from what is affected from his offense, defense, and ballpark. This is why we use FIP (Fielder Independent Pitching) and xFIP. FIP assumes pitchers can control walks, strikeouts, and the number of HRs he gives up while xFIP is the same statistic but feels some luck/environment is involved with HRs, so it normalizes the pitchers rate to that of the league average.
This is why we always talk about pitchers K rates and BB rates (and thus FIP and xFIP) more than the amount of hits they give up. Those rates tend to be better predictors of future success than hits allowed, wins, or ERA because they tend to stay relatively constant throughout a player’s career.
Another note: It doesn’t necessarily account for the quality of contact a pitcher gives up (i.e. a pitcher could be having command issues within the strike zone and thus throwing a lot of hittable pitches), so again…shades of gray — and it’s a reason why we must always look at multiple factors (i.e, a pitcher’s line drive rate LD%) and also combine statistics with scouting.
This is a subject of much debate because some are fans of what is commonly known as “small ball” — and that means things like bunting, stealing bases, and so-called productive outs.
The problem there is that all of those situations create outs when, in general, you only have 27 throughout the game. Outs are a limited commodity and giving them away has been shown to result in fewer runs scored. I should note that stolen bases do not cost runs, of course, but getting caught stealing does. Estimates put a successful percentage of stealing bases at anywhere between 75-80%. Anything lower has been shown to prevent runs.
Take a look at the chart below…
We can see from above that a team is more likely to score with a man on first and nobody out, then they are with a man on 2nd and two outs. Knowing this, why would anyone sacrifice bunt except for certain situations (i.e the pitcher at bat late in a close game)? The extra base tends to be less important than the out squandered.
Of course, the chart also exemplifies how important it is to get men on base in addition to avoiding outs — and this is why we so often talk about OBP (on-base percentage). Obviously, every time a player gets on base, he has avoided making an out. More baserunners and less outs leads to more runs. You can read about that correlation here.
I sometimes get criticized for my support of Luis Valbuena’s presence in the lineup but his .356 OBP — which is above league average, helps the Cubs score runs despite his .211 batting average.
Valbuena doesn’t just help score runs, he helps prevent them with excellent defense. While errors give an immediate reaction and fielding percentage was once the standard measuring stick of a good defender, both are far too limited to give you an accurate gauge of a defender’s overall ability.
On an intuitive level, this makes sense. All things being equal, would you rather have a guy who fields 100 ground balls and makes 3 errors or a guy who fields 90 and makes one error? The first player has a 97% fielding percentage while the second player is at just under 99% — but yet the first fielder has made plays on 97 ground balls while the second fielder has made 89. That’s 8 potential hits saved — and save enough hits and that eventually translates to runs, save enough runs and that eventually translates to more wins.
The defensive statistic often used at this blog is UZR/150, which breaks the field up into zones and tracks the number of plays made both in and outside that zone (and naturally incorporates errors into the equation). That number is translated into a rating, which in turn is translated to runs saved. Using Valbuena as an example again, we can get a feel for just how much value he adds on defense. Last season, Valbuena was a slightly below average offensive player overall — yet was considered a fringe average starter overall in part because of his defense. An average defender will have a 0.0 rating. Valbuena’s UZR/150 last year was an outstanding 18.6. For frame of reference, NL 3B Gold Glove winner Nolan Arenado’s UZR/150 was 22.4. 2012 winner Chase Headley’s UZR/150 was 8.2 while Valbuena checked in at a much better 27.6 that season. It’s not a stretch to say that Valbuena ranks among the best defensive 3Bs in baseball.
A more clear cut example is Darwin Barney in 2012 and for him we’ll focus on Runs Saved. The Fielding Bible calculated that Darwin Barney saved 28 runs — 10 runs saved translates to a win, making Barney’s defense alone responsible for nearly 3 wins over the course of that season. We won’t get into his offense for the purpose of this discussion, which is on run prevention. It’s well documented that Barney doesn’t create enough runs on offense. But we’ll get to that in the next section.
The bottom line, though, is that run prevention is the mirror to stats like OBP, which help create runs. If getting on base leads to more runs on offense, then it logically follows that any time your defender can prevent opposing runners from getting on base, then you will in turn limit their ability to score runs.
Finding unity in offensive statistics
As mentioned, statistics like RBI tend to give a false gauge of individual offensive value. It is, in many respects, a team statistic because it can be affected by teammates performances and even where an individual bats in the lineup. There is no such thing as a clutch hitter, though intuitively this does makes sense to most people — but at that high level of play, you don’t see as much variance in terms of a player’s ability to stay focused and calm as you might at lower levels of baseball. These guys have climbed to the top of their profession for a reason. Take a large enough sample size and you will consistently find that a major league player’s average with RISP is right in line with his overall average.
You have seen us use two statistics to measure the all-around offensive ability of a player. The first is wOBA, which stands for Weighted On Base Average. What that means is that it is a derivative of OBP but puts greater weight on extra base hits. It’s more accurate than OPS because OPS factors batting average in twice (it is a central component of both OBP and slugging percentage).
The second stat we like to use is RC+, which stands for Runs Created. It is similar but preferable for some because it uses 100 as the MLB average — so anyone over 100 is above average is over 100 and anyone under is below average. So far this year, Valbuena has been roughly average on offense (RC+ of 99) while Anthony Rizzo is well above average at 150. Darwin Barney prevents runs, but he also creates far less than the average player (51 last season and a tragically low 5 so far this year).
We will often use both so that the reader can choose the one he prefers.
You’ve seen us use WAR, which means Wins Above Replacement. What WAR does is take all parts of a player’s game: hitting, running, defense — and assigns it a value in terms of wins.
The base value is 0 wins and that is the value of what is called a “replacement player”. A replacement player is loosely defined as an average AAA player or the sort of player you can readily find on waivers. Because anything can happen in any particular game, a team of replacement level players would not be expected to win 0 games. They would be expected to win about 47-48 games in a 162 game season. As an illustration, Darwin Barney has been around a replacement level player the past 2 seasons.
So Wins Above Replacement measures how many additional wins any given player would have over the Darwin Barneys of the world, though keep in mind that WAR is position specific. The offensive threshold for a 1B to be a replacement level player is much higher than it is at 2B or catcher.
To give you an idea, a fringe average starter is a 2 WAR player. Luis Valbuena’s 2013 season is an example. A good starter is at around 3, Starlin Castro’s 2010-2012 seasons are an example of that, while Jeff Samardzija has been at that level the past two seasons. I’d say 4 WAR is an all-star level player. No current Cubs are good examples (Anthony Rizzo projects close to this level this season) but Aramis Ramirez was at this level during his prime Cub years. A superstar player is in the 6-7 range — Derrek Lee’s 2004 season fits in that category and 10 WAR is Mike Trout level.
So I hope this is helpful for many of our newcomers when it comes to getting a sense of how we evaluate players here at Cubs Den from a statistical standpoint. We obviously put a great weight on scouting too, but that is an article for a different day.
Filed under: Uncategorized