Changing Dynamics of HR and SB Distribution in MLB

and their implications for fantasy

Last time, we looked at how broad level statistical changes might be affecting fantasy values in the pitching pool. This post will examine some trends on the hitting side. As it relates to fantasy baseball, the significant recent changes relate to the Home Run and Stolen Base categories. Let’s look at some data on each.

If you have been following MLB news over the offseason, one of the major storylines has been how poorly the big sluggers have done in free agency. Last year’s AL home run leader, Mark Trumbo, twittled his thumbs on the free agent market for months before eventually resigning with the Orioles for much less money than he was hoping for. 2016 NL home run co-leader, Chris Carter, lingered even longer, nearly having to take his 3-true-outcomes game to Japan before finally finding a paltry (it’s all relative, man) $3 million deal with the Yankees. Edwin Encarnacion, Jose Bautista, and Brandon Moss all had to settle for much less money than they were seeking. The implication is clear: major league teams are not digging the long ball anymore. The reason for this also seems clear: if an undersized middle infielder like Freddy Galvis can knock 20 balls out of the park, why shell out the big bucks for a hulking slugger? So, does the data support this change in philosophy among major league front offices? The data indicates, yes, at least to some extent.

historical-hr

The above table puts hard numbers on the chatter we’ve been hearing about the home run spike in baseball. Whereas in 2014, when pitchers seemed to rule the world by limiting home runs to a generational low (4186 HR to be exact) the past two seasons have seen an explosion in HRs to levels not seen since the steroids heyday (5610 in 2016, second only to the 5693 dingers in 2000). That’s a 34% increase in HR in just 2 seasons for those of you keeping score at home.

That kind of change in and of itself is something all fantasy players should be aware of, but what really matters in the fantasy game is how those extra HRs are distributed. If all hitters up and down the value ladder are increasing their homers at the same rate, then fantasy values are not too strongly impacted. But what we saw in 2016 was a sort of democratization of home runs; they were spread out somewhat more widely across the player pool. Whereas the top 30 home run leaders in MLB accounted for 21% of all HR hit in 2014 and 2015, the top 30 only accounted for 19.5% of all HR hit in 2016. That may not be an earth-shattering change, but it represents the lowest share for the top 30 in the 17-season time period examined here. With home runs spread out more widely throughout the player pool, the top sluggers, especially the one-dimensional guys who offer no speed and who act as a batting average anchor, are becoming less valuable in fantasy baseball, just as they are in the real-life game.

Turning our attention now from the power game to the speed game, let’s see how stolen base trends may be making an impact on fantasy baseball. As you can see in the chart below, stolen bases have declined significantly over the past four seasons from their peak in 2011-2012. Whereas there were 3279 SBs in major league baseball in 2011, by 2015 that number had dipped down to a generational low of 2505, a decrease of nearly one-quarter (SBs increased ever so slightly to 2537 in the 2016 season). Recent seasons have also seen a modest upward distribution in stolen bases, as the top 30 basestealers in MLB accounted for only 30.2% of all SBs in 2012, but by 2016 had increased their share to 35.2%.

historical-sb

To summarize: home runs are up significantly the past couple seasons and have been spread more evenly across the player pool. Meanwhile, stolen bases are sharply down and the top basestealers are accouting for a greater share of all stolen bases. The distributional changes are what’s really key here, and although they may not be dramatic (21% to 19.5% for HR leaders; 30.2% to 35.2% for SB leaders) they are certainly noticeable and significant. When combined, these trends clearly indicate that top speed guys are gaining value in fantasy baseball relative to power hitters. Just as I argued in the last post that workhorse starting pitchers are now worth a buck or two more than they would have been in seasons past, it seems evident to me that the top speed burners are also worth an extra buck or two heading into the 2017 season than they would have been before the recent changes to the MLB hitting environment. Historically, I have shied away from the speed-only guys, preferring to fill my roster with players who can contribute across multiple rotisserie categories. But this might be the season where I take the plunge and spend what is necessary to roster the likes of Billy Hamilton, Dee Gordon and Rajai Davis.

Some Data on the Changing MLB Starting Pitching Environment

and how they can subtly impact your fantasy strategy

In my early preseason draft preparations, I have been reading quite a bit about last year’s home run explosion in major league baseball. This got me curious to look in greater depth at changes in the game and how much last year’s stats were an outlier (or not). In this post, I will explore trends from the starting pitching angle, with a future post to be devoted to hitting.

First up, I got to wondering about starting pitching usage and the implications it might have for Rotisserie baseball. As I have been scanning early projections on pitchers I have noticed that most starters with any kind of major league track record are being penciled in for 180+ innings. That seemed fishy to me since there appear to be fewer workhorses with each passing year. So I headed on over to fangraphs.com and baseball-reference.com to crunch some numbers and see if the data supported my intuition.

One data point that I believe is given insufficient attention by fantasy baseball players is the number of starting pitchers who pitch a complete season’s worth of innings at the major league level each year. First and second year pitchers tend to be on innings limits, as are players coming back from injury, and even healthy veterans are getting removed from games earlier, reducing their ability to rack up innings pitched. Indeed, the number of pitchers who threw enough innings to qualify for the ERA title fell to a modern low last year.

qsimage001

The data is a bit noisy, but it demonstrates that 2016 produced a 17-year low in the number of pitchers who reached the 162 innings threshold, furthering a decline that began in 2015. It should be noted that 162 innings is not a particularly onerous threshold; if a pitcher only averages 5 2/3 IP per start he can still qualify with 29 games started, allowing him to miss 3 turns through the rotation.

What if we beefed up the threshold a bit and started looking for true workhorses? A 190 IP threshold is tougher to reach but can still be accomplished by averaging 6 IP per game while making 32 starts, answering the bell every 5th game of the season. The results are quite dramatic.

190ipstarters2

For most of the past two decades, the number of workhorse starters per season has ranged from the 50s to low 60s. Yet 2015 saw this number plummet to 36 and 2016 witnessed an even further decline to 30. Basically, the number of workhorse starters has been cut nearly in half in recent years. While I think most fantasy players intuitively understand the decline in starting pitcher innings totals, I am not sure we have fully adjusted to the extent of this change. Pitching durability is a skill and an increasingly rare one at that. As we prepare our player projections we should be as methodical about producing a realistic IP total as we are about forecasting rate stats like ERA and WHIP.

Diving deeper into the pitching data, we can start to understand why big innings totals by starting pitchers have declined. Let’s take a look at the length of an average MLB start over the years.

ippergs

Here we find a statistic that was remarkably stable from 1999-2015, always landing between 5.8 and 6.0 innings per start. However, 2016 witnessed a new low, with the average MLB start lasting just 5.6 innings. This may not appear to be a huge change, but if the trend continues we might begin to see more dramatic declines in Win totals by starting pitchers. As the next chart shows, this decline may already be underway.

starterwinpercent

In 2016, starting pitcher wins (as a percentage of games started) reached a modern low at 33.5%. This is down slightly from the 35% or so figure that we have generally witnessed in recent times. This is not a large enough difference to substantially impact the fantasy game, as it reduces the average pitcher’s win total by only half a win over a 32 start season, but it bears watching in the coming years.

While starting pitcher wins have declined somewhat the past few years, the percentage of wins accumulated by the top starting pitchers has inched up slightly:

top20winpercent

In 2016, the top 20 win leaders in MLB accounted for 21.0% of all starting pitcher wins. This results from a steady, modest increase in the percentage of wins accumulated by the top starting pitchers since the 2013 season, which saw the top 20 racking up 19.1% of all starter wins. In short, wins by starting pitchers are declining slightly and they are being modestly distributed upwards toward the top pitchers.

So, what are the takeaways of these trends for the fantasy game? I would argue that they demonstrate top-tier starting pitchers, especially those who have demonstrated proven durability, have gained value relative to the average starting pitcher. If the changes in the starting pitcher environment continue or even just plateau in 2017, you might gain an edge by going an extra dollar or two for highly skilled workhorses. Also, if you play in deeper leagues (e.g., AL- or NL-only), when it comes to rostering your last pitcher you may want to consider selecting a talented middle relief or set-up man over a run-of-the-mill starter (the inverse of declining starter wins is an increase in relief wins; as the win total gap between a mediocre starting pitcher and a quality reliever narrows, it becomes more enticing to select the reliever who is likelier to contribute positive value to your team’s ERA and WHIP).

The key unknown is whether 2016 represents the acceleration of a trend that will continue into 2017 and beyond or whether it represents something of an outlier. Will MLB front offices and managers continue to give earlier hooks to their starting pitchers? Will MLB teams continue to utilize more young pitchers on cheap contracts (and on strict innings limits) in their rotations? Will MLB get better at limiting major arm injuries to pitchers, thus allowing more pitchers to throw a complete season’s worth of innings? No one can answer this definitively ahead of time, but most of what I read about how MLB teams view pitcher usage leads me to believe that it is more likely than not that 2016 is not an outlier. I think the scales are tilting such that top-tier starters and non-closing relievers are becoming more valuable in our game, relative to starting pitchers with either shaky skills or uncertain durability. If my leaguemates do not recognize these trends I will likely be owning more shares of the Max Scherzers, Jeff Samardzijas, and David Phelpses of the world and fewer shares from the likes of Julio Urias, Kenta Maeda, or Alex Reyes than I might have in seasons past.

Valuing Players for Rotisserie Baseball: Are You a Schechterian or a Shandlerite?

Everybody’s shouting, “Which side are you on?”

Spring Training is right around the corner and soon it will be time to get into the nitty-gritty of preparing for our fantasy baseball drafts and auctions. But before we dive headlong into the new season, now is a great time to take a 30,000 foot view and consider some broad strategic principles. Specifically, what are our overarching goals as we construct our fantasy rosters? There are many ways to answer this question and the conclusions we reach structure how we go about preparing for the season. In this post, I will evaluate the approaches of two of the leading experts in the fantasy baseball industry, Larry Schechter and Ron Shandler, to show how successful players can have very different understandings of what they are trying to achieve when they take their seats in the draft room.

Larry Schechter is perhaps the most successful fantasy baseball player since the game has been invented. He has won more “experts” league titles than anyone else going. He laid out his approach to success in the book Winning Fantasy Baseball: Secret Strategies of a Nine-Time National Champion. If you are serious about becoming a better fantasy baseball player you should read this book. I don’t intend to provide a full review here, but I will flesh out his general approach to the game and how it contrasts with Shandler’s. Schechter succinctly states his general strategy: “Get the most total value possible at your auction or draft.” That’s it, very straightforward, nothing gimmicky. With that grand vision in mind, Schechter executes his strategy by constructing very precise player value projections and purchasing players at auction for a discount (i.e., buying players that sell at prices lower than their projected values). There are lots and lots of details to fill in about how, specifically, to pull this off, but for now let’s just call this the scaffolding of the Value Maximization approach to fantasy baseball.

Ron Shandler has probably done more to promote the game of fantasy baseball and encourage analytical thinking about the game than anyone else. His annual Baseball Forecasters are a treasure trove of information for the fantasy player. The bulk of each Forecaster consists of the player statistical projections section. Much as with Schechter’s approach, the Forecaster provides a detailed projected valuation for each player. However, the Forecaster includes a “Consumer Advisory” which emphasizes that the statistical projections are only general guidelines and that the true value of the book lies in its analytic essays and underlying concepts. In recent years, Shandler has gone one step further and backed away from detailed statistical projections entirely. He has developed a new approach, the Broad Assessment Balance Sheet, detailed in The BABS Project: Uncovering the Truth About Winning at Fantasy Baseball. Rather than attempting to maximize projected statistical output, BABS emphasizes risk management by maximizing assets (e.g., skills such as power, speed; playing time) while minimizing liabilities (e.g. injury risk, lack of big league experience). Rather than assigning players a statistical projection, they are instead given broad scores in different categories of skills and liabilities (extreme impact, significant impact, moderate impact) and then grouped into “asset classes” with players who share similar scores. Players within asset classes are treated as essentially interchangeable. Fantasy players utilizing the BABS approach enter the auction with a balance sheet containing columns for each of the different asset and liability classes and seek to meet acquisition targets for players with impact scores in each of the different asset categories while not exceeding preset limits in each of the liability categories. While there is a great deal of statistics underlying the assignment of players into asset classes, at its core BABS is a sophisticated Qualtiative approach to fantasy baseball.

Now that we have a general understanding of what Schechter’s and Shandler’s strategies are trying to achieve, let’s take a moment to evaluate the fundamentals of each approach. Schechter’s Value Maximization strategy is elegant in its simplicity and comprehensive in its incorporation of all the various factors that impact player performance. What I mean by this is that, in theory at least, all the gimmicky fantasy baseball rules of thumb that you read about on the Internet (target Age 27 players, avoid pitchers who had a big IP jump the previous season, make sure you get power at 1B, etc., etc.) are subsumed by the Value Maximization framework. A player’s statistical projection will take into account factors such as where he is on the aging curve, his injury history, his risk of losing playing time, etc. The projection represents something approximating an over/under line for a player in each of the Rotisserie categories (it’s not exactly the same as an O/U, but we can geek out on those details in a later post). Once you have projections for each player, it is relatively easy to model the player pool and convert statistical output into dollar valuations. With those dollar valuations in hand, your marching orders for the auction are straightforward: buy players who sell at bargain prices. This is, so far as I can tell, the dominant strategy among people who consider themselves fantasy baseball experts. The very wide differences from one fantasy player to the next lie in the details of how they execute this strategy.

However, Shandler, the once king of the statheads now turned heretic, has called this framework corrupt at its core. He argues that the Value Maximization approach relies upon an illusion of predictive precision that does not exist. The early chapters of his BABS book detail the statistical volatility and biases of human judgment that make precise forecasting, in his opinion, a fool’s errand. You should read these chapters to get a full sense of how fallible even the best projection systems are. These are strong challenges to the dominant paradigm and should be taken seriously. However, while the Qualitative approach is intriguing, and may be very useful to a certain type of fantasy player, I’m not sure that it escapes the fundamental problem of fantasy baseball – how to predict future changes in skill, luck, health and performance amongst a group of highly talented human beings in a dynamic, competitive enviroment – any better than the Value Maximization approach. Pardon a little basic statistics jargon here, but when you convert continuous variables (e.g., FIP, K%, etc., variables that have values along a numerical continuum) into categorical variables (e.g., lumping players into skill groups of extreme/significant/moderate) you risk losing a lot of useful information that can help you differentiate one player from another in the name of avoiding false precision. A valuation system that treats, say Madison Bumgarner and Felix Hernandez, as interchangeable assets is, I would claim, a bit too modest in its ambitions of assessing player talent. More troubling, with a system based on categorical variables containing very broad ranges in each category, as BABS is, you also run into the problem of boundary issues. A player whose skills are close to the edge differentiating one category from another is likely to be placed into an “asset class” in which his skills are more different than those of players in the adjacent class. For example, a player in the top 25% of a skill is labeled with the “significant impact” tag, whereas those in the top 50% of a skill are labeled as having “moderate impact.” Now, what about a player in the top 26%? This quite talented, well above average player gets dumped into the same bucket as merely average players, even though his skills may more closely resemble those of many players who just sneaked across the line into the “significant” zone. Also, the determination of “top 25%” is made by a statistical computation that attempts to quantify a player’s underlying skills. These computations are subject to imprecision, just like the statistical projections used in the Value Maximization approach.

What the Qualitative approach has going for it is that it emphasizes balance and management of risk. Especially if you play in a league of gung ho owners who bid every prospect to the stars and pay for Strasburg like he’s a lock for 200 IP or guarantee that this is the year Giancarlo Stanton finally tops 150 games played, then BABS might very successfully guide you to a roster filled with reasonably priced, stable veterans who put up solid numbers and keep you in the running in all 10 categories (everybody plays 5×5 now, right?). I have seen this work before, although it’s been my experience that this is more of a recipe for a 3rd or 4th place team.

For now, I am sticking to the Value Maximization approach. Despite the fact that all projections are subject to error, I will still be entering a projected statline for each player this spring and deriving a projected dollar value based on those expected stats. I like entering an auction with the idea that I can purchase any player at the right price, that I am not limited by overly restrictive prohibitions against buying certain types of players. What if my league’s owners are generally squeamish about bidding up injury prone players? What if my NL-only mates haven’t done their homework on new players entering this year’s pool from the AL, Japan, Cuba or the KBO? What if owners in my keeper league are hesitant to pay the 50+ in auction-inflated dollars it takes to acquire a superstar? Nearly all leagues have their market inefficiencies. The trouble is it’s hard to know in advance what those inefficiencies are going to be. The Value Maximization approach is the best framework for being able to pounce when those inefficiencies reveal themselves in the auction room. Nevertheless, it behooves the fantasy player utilizing this approach to keep Shandler’s critique in mind. While we rely on projected stats and dollar values to guide us in the auction, we shouldn’t be overly dogmatic about it. If it makes strategic sense to bid an extra dollar or two on a particular player, by all means, we should do that. Our dollar value cheat sheets are guides, not restrictive edicts that prevent us from exercising judgment on the fly.

One final note on these grand strategies by two giants of the fantasy game: after reading both Schechter’s and Shandler’s books, I am convinced that what makes each of these guys successful at fantasy baseball is not necessarily their systems as such, but rather the hard work of nuanced player evaluation that is part of each man’s preparation process. Whether you produce a detailed projection or do a qualitative skills and liabilities assessment for each player, if you put in the work of understanding a player’s history, his strengths and weaknesses, and the various soft factors that might impact his performance for good or for ill, then you probably already have a competitive advantage over many of your leaguemates. I may not nail every player projection but in the process of developing my projections I will gain a deeper understanding of each player and where he fits in the larger MLB ecosystem. I may still miss wildly on many players, but I believe the understanding I gain from the process helps me to minimize my errors and make it more likely that I will roster a successful team. Regardless of the approach you choose, if you utilize it to make you a more informed player, then you are already one step ahead.

The Making of a Fantasy Baseball Player

Stories from a Rotisserie Dinosaur

Greetings, dear reader, to the launch of Roto Obsessive! In the days and weeks ahead I look forward to posting a number of articles that will help as you prepare to compete in your 2017 fantasy baseball leagues. But I’d like to start with a bit of an introduction about how I became a fantasy baseball obsessive and why you might want to listen to what I have to say.

I got my feet wet back in the late 1990s. I had always been a huge baseball fan and particularly loved the analytical side of the game. As a kid, I would devour the Bill James Baseball Abstract each year as soon as it came out and I spent an embarassing number of hours replaying an entire MLB season with the APBA Baseball board game. I had heard of this thing called Rotisserie Baseball (weird name!), in which a group of fans would get together at an auction and purchase “fantasy” teams of real-life baseball players, using their MLB statistics to structure a competition over who could assemble the best team. Revolutionary! How had baseball fandom existed before this innovation?!?

So, when a co-worker found an NL-only league that had an open slot for the upcoming season and asked me to manage the team with him, I leapt at the chance. We had no idea how far in over our heads we were. Armed with only a single trade publication containg a list of stale dollar values and the urgent advice not to overpay early in the auction, as the bargains would assuredly come later on, we sat patiently and waited as one top notch player after another sold for more than his projected market price. Then we waited some more…and waited longer yet. By the time the auctioneer called the first break we were the only team yet to make a purchase and the best player left on the board was Derek “Operation Shutdown” Bell. Our trusted trade publication had led us astray! Well, actually, it was our lack of adequate preparation in learning how individual league dynamics – auction inflation resulting from keeper value, the size of the player pool, and the total pool of auction dollars – can result in auction prices that differ dramatically from one-size-fits-all cheat sheets. In other words, as rookies, we made some big rookie mistakes.

We regrouped as best as we could and determined to start buying players, fast. We got Mr. Bell at a buck below the list price, but that was small consolation as we ran up against every Rotisserie player’s worst nightmare – more money left to spend than player value left available to acquire. We did end up acquiring several averagish players at discount prices, but our team was thoroughly lacking in star power and we left waaaaaaaay too much money on the table. We walked out of that auction with the worst Opening Day roster I will ever have for as long as I play fantasy baseball (knock on wood). We got schooled, but we learned quickly, managed our team aggressively throughout the season and somehow managed to claw our way to fifth place.

By the next season my co-manager had dropped out but I had caught the bug. The previous year’s auction disaster motivated me to research Rotisserie strategy more deeply. I got turned on to a publication that would change my understanding of the game: Ron Shandler’s Baseball Forecaster. Ron’s essays about underlying player skills and strategy were a quantum leap beyond anything I had read before. I felt like I had found an informational and analytical edge that most of my competitors didn’t have, and I was right (ah, the pre-Fangraphs era, it was a simpler time)! My second auction experience went much better than the first. I ended up finishing in second place that year. By year 3, I added a second league to my portfolio and won them both (thank you, Javier Vazquez)!

At that point I was off to the races. I started adding more leagues, trying different formats, and continued seeking out new sources of information about strategy and player performance. And I won more and more leagues. The confidence that came from those early wins helped breed further success in the years that followed. Fantasy baseball is fun no matter what, but winning at fantasy baseball, whoo boy, now that is dial-it-up-to-11 fun! The enjoyment I have gotten out of the game has made it easy to invest the time necessary to understand fantasy baseball deeply and keep up with its changing dynamics. Wherever you are at in your development as a fantasy player, whether you’re just beginning and trying to figure this thing out or whether you’re an advanced player looking to hold on to your edge, I hope you will find the posts to come to be informative and helpful as you competed in 2017 and beyond.