Archive for Meta Analysis

The longer we have Statcast data at our disposal, the more ways we find novel uses for them. What follows is my proposal to use the difference between expected and actual value segmented by batted ball type and venue to determine park factors (and potentially evaluate defensive value, as described in the footnote). Unfortunately, someone smarter than me was already way ahead of me. I’ll get to that in a second.

A typical park factors grid, such as those produced by ESPN or FanGraphs, commonly relies on outcomes — outcomes of plate appearances (ESPN), batted ball categories, or both (FanGraphs). They describe what actually occurred, the way wOBA describes a hitter’s actual production. Conversely, expected wOBA (xwOBA) describes what should have occurred based on a batted ball’s exit velocity (EV) and launch angle (LA). It strips away everything else, holding constant all other environmental factors in order to deliver an otherwise-context-neutral EV/LA-based value.

The difference between wOBA and xwOBA (“wOBA minus xwOBA,” or wOBA—xwOBA for short), therefore, effectively captures all value amassed or lost by other variables. In other words, if wOBA explains what actually happened in a non-neutral environment, and xwOBA explains what should’ve happened in a neutral environment, then the difference between them characterizes the effect of the environment — the ballpark itself.

Unfortunately for me (but fortunately for everyone else), Tony Blengino already did this (which is why he’s a former MLB executive and I’m not). In 2017, he used Statcast data to calculate park factors on the basis of expected outcomes relative to actual league-average production. For all intents and purposes, it’s the same idea.

Consider this post a refresher on the topic.

Let me call your attention back to a simpler time. If you search “Miguel Cabrera xwOBA” on Twitter, you’ll find, well, not a multitude, but at least a sampling, of Tweets from the summer of 2017 lamenting Cabrera’s (and his teammate’s) bad luck by measure of wOBA—xwOBA:

Read the rest of this entry »

After spending his first nine seasons in Pittsburgh, Andrew McCutchen will now be joining his fourth team in two years. McCutchen agreed to a three-year deal with the Phillies, likely spoiling Roman Quinn’s sleeper hype. Having played the vast majority of his career in home parks that seriously favored pitchers, will the move to Citizens Bank Park boost his offense? Let’s check the park factors.

Read the rest of this entry »

Last Thursday, Jonathan Schoop signed a one-year deal with the Twins, bringing him back to the American League after a short stint with the Brewers in the National League. After posting the highest wOBA of his career in 2017, his offense sunk this season, as his wOBA dipped below .300, thanks primarily to a dramatic decline in BABIP. Will the move to Minnesota help him rebound? Let’s see what the park factors have to say. Since h played nearly his entire career in Baltimore, amassing just 134 plate appearances in Milwaukee, we’ll keep the comparison between Baltimore and Minnesota.

Read the rest of this entry »

For fantasy baseball writers, this time of year means only one thing, player profiles. Dozens and dozens of player profiles to slog through. While writing up some hitters, I can to the realization I didn’t have a quick method to flag hitters who may be shifted. While I could navigate to the splits page, scroll down, and calculate each player’s shift rate, I wanted a quick and dirty method to flag them. While I went over several methods, I settled on one stat and two values.

I did some shift work for an older edition of a The Hardball Times Annual but I have stayed away from the topic over the last few years. My initial conclusions stand which are the shift really affected some hitters and the rate of effectiveness slowly declines more players are shifted. The shift isn’t going away though. It’s still effective against slow pull hitters.

I initially thought a hitter’s pulled groundball rate would be a major factor, but it was not. Just pull rate was enough.

Read the rest of this entry »

Yesterday, I wrote about how sprint speed relates to wOBA minus expected wOBA (wOBA–xwOBA). Today, I summarize my investigation into what factors most readily affect a player’s stolen base success rate (SB%).

This invitation from BatFlip Crazy, embedded in this lengthy Twitter exchange, served as the catalyst for the research. In hindsight, I’m not sure I totally answered the question. Manipulating data from multiple different sources (in this case, Baseball Reference and Baseball Savant) can be exhausting.

I used my final Frankenstein data set, which contained statistics for all players from 2016-18 with at least 100 stolen base opportunities (SBOs) in a given season, to investigate relationships among the following various stolen base metrics:

Read the rest of this entry »

One of the keys to fantasy success involves finding when projections can be systematically off. The hard part for fantasy owners is that most of these findings, like pitch velocity, get quickly incorporated into projections. Since it’s difficult to find these discrepancies, I was intrigued when I saw this quote by Mitchel Lichtman (MGL) in an article he wrote:

So, the substantial under-projections seem to occur when a player gains speed but his wOBA remains about the same.

And by substantial, it was a 22 point difference is wOBA. This is a major difference and could point owners to some nice upside plays. I decided to go ahead and dive in.

Read the rest of this entry »

Fantasy analysts and enthusiasts alike are still searching for ways to use Statcast’s expected wOBA (xwOBA) metric meaningfully to gain an edge. Unfortunately, beyond leveraging the difference between xwOBA and actual wOBA (what I, and probably countless others, refer to as the “wOBA minus xwOBA differential”), I don’t know yet how else you can use xwOBA effectively. Given already-widespread use of the metric, the minimal edge you can glean will come from interpretation.

I discussed the interpretation of xwOBA multiple times in 2018. In May, I highlighted hitters on whom to buy low because of their extreme/outlier wOBA–xwOBA differentials. In July, I called out xwOBA’s inability to account for what appeared to be the ball becoming un-juiced, thereby overestimating xwOBA across the league. In September, I investigated the predictiveness of xwOBA in-season (that is, the predictiveness of first-half xwOBA on second-half wOBA).

I discussed all of these topics during my presentation at BaseballHQ’s annual First Pitch Arizona forum, especially the former-most. Basically all of the hitters I tabbed as buy-lows outgained their prior performance by substantial margins — all of them, that is, except for Victor Martinez. Could he be considered a miss? Sure, except he was different from the rest of his fellow underachievers: he perennially underperforms his xwOBA. Perhaps the better question, then, is: Why was he a miss?

Read the rest of this entry »

Every pre-season, I compare my rankings to the RotoGraphs consensus (which excludes my ranking of course) to discover which players I’m apparently more bullish or bearish on than the others. I was strapped for time this year and so only published a bullish piece for hitters, dubbed Pod’s Picks. Let’s see how these hitters ended up performing and ultimately ranking. Remember that the rankings are positional and not overall, which makes this exercise a bit more difficult. I’ll be using the FanGraphs auction calculator to compare values.

Read the rest of this entry »

The Process is done. After months of writing and disagreements, the fantasy baseball guide I wrote with Tanner Bell is for sale. Simply, we researched and created the guide we would want. It’s not going to contain player capsules or positional rankings. Dozens of sources provide them. Instead, it’s going to give the reader the ability to create their own and the 2017 NFBC Main Event Champion and my podcast mate, Rob Silver wasn’t happy about the end results as he states in his intro.

Jeff Zimmerman and Tanner Bell are both awful people.

Thanks Rob. Before Rob gets to explain himself, here is the general flow of he book.

Read the rest of this entry »

About a month ago, I published a post that predicted 2019 ADP (“xADP”) values using eight years’ worth of average draft position (ADP) data from the National Fantasy Baseball Championship (NFBC) and end-of-season (EOS) values from Razzball. The model was pretty good — it explained nearly 60 percent of the data’s variance (adjusted r² = 0.59), which is pretty dang good. It felt unfulfilled, though; it accounted for some players but not others — namely, breakout rookies who were completely off the radar the previous season and top prospects who had yet to debut.

I took some time (really, a lot of time) to clean up my data to see how much it would improve my model, if at all:

1. Originally, my data set did not account for players who were not drafted (aka had no ADP value) but made an impact in 2018 (think Juan Soto). Conversely, my data did account for players who were drafted but made no impact in 2018 (think, uh, Troy Tulowitzki, I guess). It was kind of like addressing a Type I error but ignoring a Type II error (or the other way around? I don’t know). I took painstaking care to fill in these holes.
2. I took equally painstaking care to ensure all player names were consistent — no “Nick Castellanos”/”Nicholas Castellanos” mismatches that might pollute the analysis. Odds are, there are a couple of players I missed, but having spent hours poring over the data, I feel confident that the issue is no longer pervasive.
3. I added ages! They make a small impact, most meaningful to players at the extremes, such as the very young (think Ronald Acuna) and the very old (think Nelson Cruz).
4. Lastly, a theoretical and methodological adjustment: I forced negative ADP values to $0. I wanted the model to reflect an actual draft, in which players are never bought at auction for negative dollars — rather, their values converge on zero. It’s important to note here that a player can still end the season with negative value based on the concept of replacement level. Accordingly, only negative ADP values, and not negative EOS values, were forced zero.

Fortunately, the extra work was worth it: the model boasts an adjusted r² of 0.75 (with ages; 0.73 without). That’s a massive improvement, and it can be attributed almost entirely to the slight (but profound) change in the model specification.

Read the rest of this entry »