Monday, August 19, 2019

Evolution of League-Wide Pitch Sequencing

As the Statcast era continues to age, we (i.e. us baseball obsessives) are collecting more and more pitches to analyze in countless different ways. BAM releases 90 different metrics for every pitch thrown, including a pitch's classification, where all the defenders are standing upon the pitch being thrown, exit speed, launch and spray angle, etc. Analysts across the web, armed with this exhaustive data set, have been able to unearth previously unknowable mysteries regarding team and player performance and league-wide trends.

One area of pitching analysis that has been largely untouched by public analysis is pitch sequencing. Baseball Savant has done some work with visualizing a how a pitcher sequences his pitches, but to my knowledge there is no way to look at pitch sequencing for the league as a whole and see which sequences are most used and most effective. Curious at how pitchers have attacked hitters since 2015 (the beginning of the aforementioned Statcast era), I parsed through every pitch thrown in the regular season starting from the beginning of the 2015 season up until August 11th of this year. I looked at how pitchers have paired pitches during every plate appearance. I discarded pitches that were not thrown to the same batter or the same inning; a pitch that is thrown to end an inning followed by a pitch to start an inning should not be considered a sequence (same can be said for two different plate appearances). The sequences should be read as the pitch on the right precedes the pitch on the left. Now, let us look at the trends:


This chart includes all sequences the represent at least 2.5% of all sequences used in a given season. Every year, the most used sequence is a four seamer preceded by a four seamer. Sequences involving a slider and a four seamer have been used more every year in the Statcast era. In response to the league-wide trend of increasing launch angle, two seamers and sinkers have been going out of style; we can see sequences involving these two fastball variants are also on the decline.

The following chart displays the best pitch sequences in the Statcast era (those that yield a wOBA value of less than .300):
And here are the worst: 

Sequences involving fastball variants seem to yield the worst results, while sequences with with fastballs and breaking balls or just breaking balls have been the most successful. Based on the usage patterns across the last four years, these results are not surprising in the slightest. Teams are ever aware of this inefficiency and are slowly programming their pitchers to change their mixes. Look for this trend of leaning on breaking balls to continue into the future, with hitters being forced to adjust. Many of the troubling trends in the game today (strikeouts, walks, lack of balls in play) are often attributed to this rise in breaking pitches. Teams are behaving in such a way that these trends should be expected to continue. Unless the league can figure out a way to address the aesthetics of the game, baseball fans will be seeing more of the same from major league plate appearances. 

Tuesday, August 6, 2019

Leveraging WPA To Make 4th Down Decisions




Anybody who pays attention to statistics and data-driven decision making in the NFL knows that coaches are woefully conservative. We all bemoan the fact that coaches make decisions out of fear of losing, as opposed to making to appropriately aggressive decisions to win football games. Each decision in a vacuum might not seem like a big deal; individually these decisions often swing a team's win probability by no more than 5%. Now, the leverage increases as the scoring margin and time remaining in the game decrease, but nevertheless the bulk of the decisions seem inconsequential. However, when you look at the 4th down behavior of a team over the course of a season, those small, 5% decisions start to add up. Bundling these decisions together can add or subtract 1 or 2 wins to or from your total, the difference between making and missing the playoffs and getting a chance to make a run at the Super Bowl. So, let us look at 4th down plays and how they tend to add or subtract from a team's win probability. 

The following plot shows the win probability added (WPA) on each fourth down play in the 2018 NFL season, sorted by whether or not the team with the ball went for it: 

Based on this plot, there seems to be a break even point at about the 50 yard line. Once the offensive team enters its opponents side of the field, it should almost automatically got for it on fourth down. The problem with just looking at whether or not the offensive team went for it without any sort of context is that when the offensive team is on its own side of the field and its late in the game, unless it is close to the first down marker, the offense is forced to pass the ball on 4th and long. By not converting, your win probability is going to either stagnate or decrease in these late-game situations. Breaking it down by play type can give us an idea of how conceivable the fourth down conversion really is; if the offense is running the ball on fourth down, chances are the offense is close to the first down marker. So now, lets see that data based on play type: 
Now this is more interesting. At about the offense's own 25 yard line (75 yards from the end zone) running the ball on 4th down is almost always more lucrative than punting. And at about the opponent's thirty yard line, kicking a field goal offers almost no WPA in general and either passing or running the ball becomes a positive expected value play. 

So what conclusions can we draw? First, if you are close to converting a first down, teams should be much more aggressive on 4th down at most points on the field, more specifically once they get past their own 25 yard line. Second, once inside the opponents 35 to 30 yard line, teams should automatically be going for it. Kicking a field goal and taking the 3 points does not increase the chance that the offensive team wins the game. In today's NFL, with its emphasis on calling defensive penalties, teams are scoring and moving the ball at historic rates. The best teams convert drives in the form of touchdowns. Getting close to the end zone and kicking a field goal does almost nothing. If coaches are looking to add wins on the margins, and for their own sake, keep their jobs by maximizing their teams potential, they need to start being more aggressive on fourth down. Coaches may think that by being conservative, they are minimizing their chance of losing. But looking at the data, they are doing the opposite. If coaches want to be successful and ultimately keep their jobs for as long as possible, they have to trust themselves and their staffs to draw up the right plays in these higher leverage situations to squeeze the most value out of their teams as possible. 

Data via nflscrapr


Monday, August 5, 2019

How the Astros Can Fix Noah Syndergaard

With Noah Syndergaard trade rumors whirling around the baseball world, it is fair to ponder how a team who would acquire him could improve upon his 2019 results. To be clear, this incarnation of Syndergaard is still an excellent pitcher, but he still leaves us wanting more; Syndergaard has the physical profile and repertoire of a pitcher that teams dream about in their search for their next front-line starter. So how can he improve upon his already great season?

Look no further than his right-handed colleague, Gerrit Cole. When Cole was trade to the Astros, the baseball community expected that the Houston would be able to optimize Cole’s raw stuff to build one of the best pitchers in baseball. We expected the Astros would ditch his mediocre sinker and trade it for more breaking balls, Cole’s bread and butter. Sure enough, that was the case, and Cole has become one of the most dominant pitchers in the league:
Let’s look under the hood at Syndergaard’s 2019 season. Here is a zone plot of all of his pitches this year:
He is throwing his four-seam fastball up in the zone and his breaking pitches and changeup down in the zone. That is all fine and well, but looking at the plot, he is leaving a lot of sinkers in the middle of the plate. Batters have a .381 wOBA on his sinker, per Baseball Savant. It is possible that he needs to scrap it. Now, looking at his pitch mix and results, we see how poor his sinker is performing versus his other offerings and how often he is throwing them:
As I said, he is throwing way too many sinkers. Considering the Astros are presumably one of the clubs vying for Syndergaard’s services, I wanted to see how his results would change if he adopted Cole’s pitch mix. From Baseball Savant, I took his wOBA allowed by all of his pitch types and then multiplied those values by Cole’s 2019 pitch percentages, essentially taking a weighted average of Syndergaard’s average results. This may not be an extremely accurate measurement of how he would perform, but it is worth looking at. The result was a .293 wOBA, a figure that would improve his current .308 wOBA by about 5%.

While this hypothetical change in pitch mix would not yield extremely different results, it is still an improvement nonetheless. For a club such as Houston that is vying for a World Series championship, those incremental gains matter. With a division title all but wrapped up, the Astros can afford to look ahead to the playoffs. While the brevity of the postseason does not often reward the best clubs, the Astros should be confident that the addition of an improved Syndergaard can supply that small edge that might put them over the top. A top three of Justin Verlander, Cole, and Syndergaard would be the most fearsome trio of starting pitchers in recent memory and would solidify the Astros’ status as the favorite in the American League.

This post appeared at FanGraphs before the trade deadline

Pitch level data via Baseball Savant. All other data via FanGraphs


What's Next for Chris Archer

The trade and struggles of Chris Archer have been well documented by the baseball community. As many patrons of this very website know, Archer was traded from the Rays to the Pirates at the 2018 trade deadline for Austin Meadows, Tyler Glasnow, and a player to be named later, who was later revealed as Shane Baz. Both Meadows and Glasnow have been very productive at the major league level for the Rays, while Baz is FanGraph's 63rd rated prospect on THE BOARD carrying a future value of 50. Archer, on the other hand, has been an unmitigated disaster. The once tantalizing strikeout artist has delivered a 5.01 ERA paired with a 5.05 FIP. He still has struck out batters at a 26.0% rate, but his walk rate with the Bucs has surged to 10.1%. The main culprit for Archer's spell of bad performance has been home runs; Archer has always tended to give up more fly balls than the average hurler, but since his move to Pittsburgh 19.6% of his fly balls have left the ball park, about 6% more than league average over that time span. Naturally, this leaves all of us baseball fans wondering, what happened to the pitcher who just four years ago was deemed by Dave Cameron as one of the 10 most valuable assets in MLB?

The Pirates were once thought of as one of the savviest organizations in baseball, a team that could spin washed up pitchers into innings-eating workhorses. Detailed in Travis Sawchik's book Big Data Baseball, the Pirates were one of the first teams to weaponize the infield shift. Pirates pitchers under the tutelage of pitching coach Ray Searage have been taught to rely on sinkers and two seamers and to induce ground balls that can be gobbled up by the shift. As hitters have adjusted to the shift and focused on putting balls in the air, the Pirates approach of tailing fastballs low in the strike zone has become outdated. What was once the fountain of youth for veteran pitchers has become the focus of ridicule from the more analytically inclined. The aforementioned Glasnow, Gerrit Cole, and Charlie Morton have all enjoyed upticks in performances upon their departures from the Steel City. Archer, as a member of the Rays, relied on a heavy diet of four seamers and sliders, the latter of which took the league by storm in 2015. Now as a member of the Pirates, Archer has traded out some of his four seamers and sliders for the two seamer, the pitch of choice for Ray Searage and the Pirates pitching staff.

Looking at the way Archer has sequenced his pitches thus far in 2019, we can see how detrimental this shift in philosophy has been. The following plot shows Archer's wOBA allowed for two pitch sequences he has used at least 50 times this season, where the first pitch is the final pitch in the sequence and the second pitch is the one that preceded it:

Archer's sequences that have resulted in the most damage all include his two seam fastball. His sequences with a wOBA allowed of less than .300 all involve his four seamer, slider, and changeup. Based on these results, its clear that he needs to get the two seamer out of the mix, or at least pull back the reigns on its usage. Looking at his start by start pitch mix, it is clear he has noted his two seamers lack of effectiveness and adjusted accordingly:

Archer has made the necessary adjustment and basically scrapped his worst pitch. Thus far, unfortunately, the corresponding results have not been there. He has a 5.03 ERA since the start of July, a similar output to the rest of his Pittsburgh tenure. If this reversion back to his Tampa Bay repertoire does not bear the fruit that made him once one of the most promising pitchers in the league, it might be time to close the chapter on Chris Archer as an usable Major League starting pitcher.

Pitch level data via Baseball Savant. All other data via FanGraphs and Baseball Reference.