Talking Baseball: Starting Lineup Win Pct and Plus/Minus

There’s a lot of statistics that show how well a certain baseball player performs as an individual, but not a lot about how they perform as a member of a team. A few weeks ago, I created some initial Plus/Minus numbers for members of the 2018 Red Sox as a way to measure this.

Now, you may ask yourself, if we know how well an individual player performs; then why do we need to measure how they affect their team? After all, if a player is successful, then the team should perform better. It would seem to be common logic, but is it true, or are we taking things for granted. I’m recalling the 2017-2018 Celtics season where the Celtics made it all the way to Game 7 of the Eastern Conference Championship without, arguably, their two best players (Kyrie Irving and Gordon Hayward) nor did they compete with their best player from the previous season (Isiah Thomas who was traded for Irving). Now, you could say that the Celtics overperformed in the playoffs; or you could say that maybe a team can perform better without the best individual players.

Before you say, well, that was basketball and this is baseball; another example can be seen in the World Baseball Classic. The American teams for 2006, 2009 and 2013 were all compromised of Major League players, but the best that any of those three teams performed was a 4th place finish in 2013 despite having the “best” roster in terms of Major League numbers. Granted the 2017 team won the gold, but after three straight disappointing finishes, you could hardly argue that having the best players means winning the most games.

I’m going to spend the year tracking my hypothesis, that team performance can be segregated from simply having the best roster. I’ve decided to measure this in two ways:

1) How often does a team win when a player is in the starting lineup.

2) How well does a team performs (as measured by runs scored vs. runs allowed) when a certain player is in the lineup.

I’ve compiled the numbers for all teams, but since my site’s demographic universe seems limited to my Boston-area friends and the season has only just started, I’m initially going to publish only the results for Red Sox.

Measurement #1: How often does a team win when a player is in the starting lineup

So, here’s my thought on this one. A team can only have 9 or 10 players in a starting lineup (based upon the league the home team is in) and the presence of any one player on the roster means that another player is not able to start. Furthermore, a starting lineup dictates the offensive order of a team and having the “wrong” player in the “wrong” slot could cause problems for a team. Now, one player may replace another in a later inning, but even then, the “ecosphere” of the team (who is playing what position and batting in what spot in the order) has already been set.

Since we’re early in the season, I’ve set a minimum number of Games Started to 3; which means that we won’t see any Starting Pitchers yet, and as the season progresses, I’ll up this number to eliminate numerical “glitches” from having players with too few appearances, but the logic of this metric is pretty simple: (1) Find out the Win Percentage for the team – TeamWinPct. (2) Find out the Win Percentage for all games in which a particular play starts – SLWinPct. (3) Subtract the first number from the second – PlaySLWinDifFPCT. (NOTE) If a certain player starts every game for the team, the SLWinPct will equal the TeamWinPct and therefore the PlaySLWinDifFPCT will be zero.

Now, it is far too early in the season to establish any proof with the limited data sets that we have, but as the season progresses, we should see trends. With this initial dataset, the one trend that I see is that the Red Sox perform better with Blake Swihart starting the game. While overall they are 3-11 (a 27.27% win rate) with Swihart starting, they’re 2-2 (a 50% win rate). Subtract the team score from Swihart’s, and you can say that the team wins 27.73% games more than normal when Swihart plays. Compare that with the player that Swihart would most likely replace (Christian Vazques, the team’s other catcher). The team has only won 1 out of the 7 games that he started for a SLWinPct of 14.28%; which gives him a -12.98 PlaySLWinDifFPCT value.

Now there are a lot of numerical inferalls you could make, such as replacing Vazquez’s -12.98 with Swihart’s +22.72 PlaySLWinDifFPCT value could mean a 35.70% jump in team performance. Personally, I’d be curious to see what happens if Vazquez caught in a game where Swihart was DH or playing 1B like he did last year. Would we see the team win more or less? I’m curious to see how this number develops as the season progresses.

Measurement #2: How well does a team performs when a certain player is in the lineup.

Where Measurement #1 is only concerned with how a team does when a player is in the starting lineup, this measuremen looks at the inning-by-inning perfromance of a player, with the idea being that once a player is in the game he brings with him a unique skillset of offensive, defensive and baserunning abilities that can indirectly affect the game. It seeks to address the issue of whether or not a player who is good at defense but bad at offense gives a better result (in terms of Net Runs for the team) than an offensive prodigy/defensive liability.

Like the PlaySLWinDifFPCT measurement above, the values shown here exclude pitchers; since I put in a minimum 25 inning threshold; but we should start seeing some pitchers in those numbers soon as starting pitches hit their 4th time in the rotation. The goal of this measurement will allow us to see the effectiveness of bench players who serve more as late-inning replacements and relief pitchers. The rule of thumb is that if the player is in the starting lineup for the inning, or if he replaces someone who was in the starting lineup at the end of the inning; then that counts as the player being “in” the inning.

As for the numbers themselves, they are generated much like the first metric, only I measure the Win Differential as opposed to if the team won or lost. The new values I’m looking at are: (1) The net run differential for the team divided by the total number of innings the team has played – TeamPlusMinus. (2) I then calculate those numbers only for those innings that the player appears in – PlayPlusMinus. (3) Finally, I take the player specific PlayPlusMinus and subtract the Team-based value TeamPlusMinus and that gives us – TeamAdjPlayPlusMinus. This TeamAdjPlayPlusMinus indicates the number of additional runs gained/lost over the team average per inning.

Now again, it’s early in the season and there’s been a couple of blowouts that have skewed the early season numbers, but as time goes by, we should be able to differentiate between statistical glitches and the true patterns.

As before, we see that Blake Swihart is the highest-ranked player, but Christian Vazquez also shows a smaller net positive and you may ask, how does this work since they play the same position. In this case, Swihart is occassionaly used as a pinch hitter. Take for example the game on March 31st, where Swihart was a pinch-hitter who singled in a run while Vazquez was still in the lineup.

Taking a look at the bottom of the list, it seems odd to see that J.D. Martinez has a net negative greater than the team; but a lot of that is due to the fact that he wasn’t in the field for the entire game in the Arizona series and the run differnetial was a little different in those innings than those in which he played. One little run may not mean much in the grand scheme of things, but this early in the season, we do notice it.

Only time will tell if I’m onto something here; but at worst, we’ll have another set of numbers to measure performance.

Written by Dave Curewitz

Talking Baseball: Plus/Minus

“In the spring,” to paraphrase Alfred Tennyson’s Locksley Hall, “a young man’s fancy lightly turns to thoughts of baseball.”

As the 2019 season is about to launch, I find myself wondering if there’s a better way to analyze which players are better than others. The thing is, there’s a hundred different statistics to show which player is better than another: WAR, OPS, ERA, WHIP, VORP, and so many more that your head can explode trying to contemplate the worth of each of them. Sure, maybe a third-baseman has a Total Zone Total Fielding Runs Above Average of 6 and an OPS+ of 108 and hits .284 on Thursday against left-handed pitchers with at least two E’s in their last name; but it says virtually nothing about how this player affects the outcome of the game.

That’s why I want to propose the incredibly simplistic-yet-elegant new statistic called “Plus/Minus”. Other sports have it. All you do is check when a player participates (pitches/hits/fields/runs) in an inning and see if any runs are scored. If a run is scored while a player is on defense, their run counter is decremented by one; if a run is scored while a player is on offense, then their run counter is incremented by one. Take all the half-innings that a player participated in and divide that number by two, and that gets you the total number of innings for the stat. Then take your run counter and divide that by the number of innings and, voila, you get the Run Differential per Inning, but RDpI hardly rolls of the tongue and isn’t nearly as elegant as Plus/Minus.

What I like about this, is that it covers everything: How well they hit, throw, field, and run. Also, since there’s only nine (or 10 slots in the American League), it also considers who is NOT on the field because this player is on the field. Now some of you might suggest that a subpar player may get credit for the work done by superior teammates, but this is why I set a 50-inning threshold. Now, maybe an inferior player is “lucky”, but then again, isn’t the exact location of a ball lucky, or the fact that a defender made a throwing error lucky? Luck is part of the game.

Now I’d like to take a look at the Plus/Minus values for my favorite team (and reigning World Series Champions) the Boston Red Sox for 2018. I pulled the data for every inning of every game, determined who participated in the inning and how many runs were scored. Here’s what I see:

The first thing that stands out is that Mookie Betts, the man who won the American League MVP is only 5th on the list. Now, if you wanted to factor out players with less than 1,000 innings, then Mookie is indeed the best on the team. Still, the #5 overall seems a little low.

Number 1 is the team’s ace, Chris Sale, who participated in 147 innings and whose team scored 55 more runs than it gave up. Now, you can say that Sale, a purely defensive player, got “lucky” because his teammates hit for him, but isn’t it fair to say that he inspired his teammates with his pitching to try harder on offense? If you look at Rick Porcello (the 3rd ranking player on the chart), you could say similar things. Consider the differences in ERA in Sale and Porcello (with Sale’s 2.11 being less than half of Porcello’s 4.28). You could say that Porcello was luckier than Sales to get that run support.

What really amazes/frustrates me is the #2 man on the list: Sandy Leon. Sandy was one of three catchers who was on the Red Sox last year, and he actually saw more action than either Christian Vazquez or Blake Swihart (647.5 innings vs 588.5 vs. 386), and pitchers raved about his ability to call games. Still, his batting average of .177 was the weakest of the bunch, and the Red Sox just released him earlier this week. But getting on base is only part of the game, and the Plus/Minus numbers show that whenever he played, he generated .1304 more Run Differential per inning, which is an over 1.1736 more runs per game difference than Christian Vazquez and just under 1.8 more runs a game than Swihart. This is a case where the pitchers knew what they had. Let’s hope for Red Sox fans everywhere that the management knew what it was doing when they let him go.