Well ... this was predictable: More on Analytics, Boredom, and Baseball

I had set out to write post on the way analytics had made baseball boring.  And, I wanted to do that without sounding like an old timer railing against change for not other reason that he railed against change. Instead, it is boring because each team follows nearly exactly the same strategies. MLB is fighting against those strategies (expanding the bases, introducing pitch clocks and rules about pitcher disengagement). This is not because analytics failed. It is because it succeeded. Analytics worked, or at least seemed to. One could use massive amounts of data to position defenders, organize the pitching staff, and change the "launch angle" on swings to gain an edge on other teams. Tampa Bay might be the new poster child for this approach to baseball and it is hard to argue with success. The Rays are almost always in contention and almost always with a team that is among the least expensive (in terms of overall salary) in the majors. They have worked ideas such as the now outlawed shift, the starter+bulk pitcher, multiple substitutions, and pitch limits harder than just about any other team. 

The very problem is their success. Every team now has an analytics department and every team is using the same data. The result is that there is precious little variation in baseball strategy between teams. For instance, no one lets starters pitch deep into games any more. In fact, deep into games has been virtually redefined as getting through the fifth inning, the minimum number of innings a pitcher can pitch to record a win. The Jays have four starters who have been basically uninjured this year (I know Gausman has missed a couple of starts) and it is not clear any of them will get to 200 IP.  The point is not that the Jays are doing anything different or unusual. They are doing exactly what everyone else is doing. The only starter in the AL who looks like he will make it to 200 IP is Cole (as always data from Baseball Reference).  In the NL, three or four might make it. 

The addition of the universal DH has also created a a higher degree of homogeneity in baseball strategy across MLB. The absence of a DH forced NL teams in the past to adopt different strategies than AL teams (greater pinch hitting and using more small-ball, one base strategies).  I know we need to guard against nostalgia, but there were appreciable divisions of strategy between the baseball teams of my youth. Some were based on speed (KC, Oakland, the Expos, St. L). Some were based on power (Milwaukee). Some where based on platooning (the Jays, Orioles, and Tigers under Sparky). Some managers were quick hooks (Alou). Others expected their starters to work for their living (Vernon).  Said differently, when different teams played each other it was often a conflict of strategies as it was a conflict of players and teams. 

This is no longer the case. I was watching a Jays/Reds game a week or so ago and each team managed its pitching staff in exactly the same way. The only difference was that the Jays starter (Berrios) pitched 5.2 innings. The Reds (Kennedy), 5. There then followed a stream of pitchers pitching one inning (or, in the case of the Jays, the final out of the 6th).  The next game was as close to a repeat as you could get. The Jays needed to yank Richards faster than they wanted to because he was a bit off his mark, but that was it.  What minor differences there are, are a product of talent (Cleveland has to run because it lacks home run power) or taking advantage of particular situations (the Rockies were running on us because our catching is weak at throwing out runnings). Other coaches take advantage of situation but the overall strategies aren't varying. The result is that baseball fans basically know what is going to happen before it does and that does not make for a particularly exciting game, at least from a strategy perspective. 

Analytics have created two effects on the way baseball teams are organized. First, there has been an expansion of the bullpen. Most teams carry eight bullpen pitchers and five starters, taking up 13 roster spots, or half the team. Because there is almost no such thing as a complete game anymore (Lyles looks like he will lead the AL with 3; the Jays have a total of one complete game -- Bassitt -- this year), and because no one uses a reliever for more than an inning, more pitchers are needed. This leads to point two: because roster sports are needed for pitchers,  no one carries a dedicated DH any more (there are no Edgar Martinez, of Big Papi's, or Hal McRae's ... with the notable exceptions of the Dodgers but is anyone surprised Roberts is doing things differently? And, Phil as they try to get Harper into games). And, players are prized by the degree to which they can play more than one defensive position. 

For the Jays, the effect is periodically shaky outfield defence because we play infielders in the outfield (our backup or platoon outfielders are Biggio and Merrifeld, both infielders).  Throughout much of the year, our backup third basement and shortstop was Santiago Espinal, a player who was so not reliable as a short stop that the Jays traded (we really traded nothing so I am not all that concerned) for Paul DeJong.  I actually think Merrifield has done the job the Jays asked him to do this year and it has not an easy job. He makes the odd really good catch as an outfielder if he is moving forward but he has problems with balls over his head and tends -- as we would expect of an infielder -- to misjudge them. There were plays I've watched where Biggio, likewise, seemed completely miscast for the role he was asked to play. 

Is this a management issue? I'm argue that the heavy use of analytics creates this problem. Because the Jays are using up so many roster sports on pitchers, they can't carry a real back up outfielder. 

Likewise, the Jays have to suffer with Vladdy at first. I'll grant him Vladdy due. He did not come up playing first base, defence is not his primary value, and he does make some good plays. But, overall, his defence is hurting the team (it even hurt the team last year in his Gold Glove winning year, its just that all other first basement were worse).  The Jay would improve if they had even a break-even defensive first baseman and moved Vladdy F/T to DH. But, they can't because no one -- exceptions noted above -- carries an F/T DH anymore. Instead, it is used as place holder to give a player a rest. Because benches are so thin, there are just not a lot of backup players to fill in and so the DH spot is used to spell a player from fielding for a game. Or, for the same reason, to keep a player with a minor injury in the game. 

Not only does this homogenize strategies but, with regard to the Jays, it has made them a worse team. I am not saying Vladdy could be the Jays Big Papi, but I actually don't see any reason why he could't be. 

And, I have not even mentioned what "max effort" is doing to pitchers arms. 

In short, it is hard to complain about the Jays. I have and will. But, we also need to see that the problems with which the Jays grapple are inherent to this era of baseball (for instance, I know Hernandez had horrible defence. I said it. But, the response to that does not need to be to trade him, even while I think the trade worked out. Another alternative, would have been to teach him how to be an outfielder but, for whatever reason, that die not happen). 

I find myself drifting from baseball and have been trying to figure out why. This post, I guess, is in that spirit. 

Analytics, Boredom, and Mis-assessing MLB Player Value

The analytics revolution in baseball was about uncovering player values. It was about moving away from relying *just* on traditional stats (RBIs, runs, wins, etc.) to try to find ways to assess players independent of their context, say a manager who misplayed them or a pitcher who won a lot of games because their team scored a lot of runs.  Since Bill James and others first starting making the case for a better approach to baseball math over a generation a go, we now have a raft of new statistical categories that sub in for traditional numbers. This includes OPS, OPS+, wins above replacement (WAR), etc.  There are mathematical variations in some of formulae used to calculate these stats (not every determines WAR in exactly the same way), but by and large I think these new stats are remarkably useful for determining player value. But, they also have their problems particularly when combined with an MLB penchant to measure just about everything that can be measured, at least this seems to be Sportsnet's approach with the Jays, where we are treated to a daily discussion of "exit velocity," distance home runs are travel, etc. I find this fascination less useful and I also want to explain why. 

What is now called analytics was about finding more effective ways to evaluate players. The original analytics writings did not dismiss older stats, but argued that they needed to be understood better and in context. For instance, imagine a player who drives in a lot of runs and hence has a lot of RBIs. Was that a product of their ability to hit in the clutch (when, say, their team had a runner scoring position) or was it a product of the fact that the players who batted before him were on base a lot. If that were the case, a player would have more RBIs and not necessarily be any better at driving in runs than a player who have fewer. 

I'm going to come back to this but the two most important offensive skills in baseball are the ability to get on base (hard to score without people on base) and the ability to advance runners (that is move them from one base to the next). Some players are really good at that.  The batter in front them gets a double and is on second with none out. The next batter, hits the ball behind the runner, allowing him to advance to third. It doesn't show up on standard score sheets but that is a productive out, moving a teammate to third with one out where he can now score with a sac fly. Analytics attempted to find ways to enumerate these skills so that a player could be assessed on their merits. 

I agreed with analytics. I think its advances were understood long before baseball had analytic departments or anyone had seen Moneyball. It makes sense that the ability to get on base is important and that it does not matter exactly how one gets on base. Thus, on base percentage is a more important statistical measure than batting average. I get it.

I don't think "we have gone too far" with analytics but I worry that analytic stats are disguising other stats. I'll say it again, Bill James never rejected traditional stats. Instead, he worked with them and interpreted them. The substitution of one stats for another taken out of its context -- I'll say it again, taken out of its context -- disguises a player's value and that, it seems to me, is one of the problems that haunts major league baseball today. The Jays are a case in point and I'll give you an example: Brandon Belt. 

The Jays brought Belt in to provide veteran leadership and good defence at first. He was never going to play everyday but rather was a role player, doing time at DH and backing up Vladdy at first. He could defensive sub in tight games. It is a valuable role on the team. Belt got off to a bad start, as happens but the Jays kept him in and he's played more and more over time. Right now, he's batting third and seems -- barring something unforeseen -- ensconced their for the rest of the year. 

When you watch Jays games, the announcers make a lot of the fact that some -- often nameless -- people were down on Belt because of his slow start. I might say that anyone who knows much about baseball was not down him and recognizes that not everyone gets out of the gates at the same pace. It is annoying (and likely more so to the coaching staff) but nothing to write home about. It is a regular part of the game. After this, however, the announcers go on to make a comment the goes something like "Belt has the second highest OPS, OPS+, etc., since date X."

This is less useful. It is good to know but what are more traditional stats telling us about Belt: that he's not driving in runs at the pace one needs from a top of the order guy. Right now Belt has 350 plate appearances (the lowest of a non-catcher starter or a player who has not missed time for injury, but not a bad number). The problem is that he has only 37 RBIs and 15 of those RBIs come from homers. IOW, aside from himself, Belt has driven in only 22 other runs this year. That is a bad number no matter how you cut it. My point, of course, is not that Belt is secretly a bad player. I don't think he is. My point is that the focus on new analytic stats -- where he looks really good -- disguises the fact that the Jays have put a guy at the top of their lineup who is not doing what a guy at the top of the lineup is supposed to do. 

There are a lot of qualifications, to be sure, and it would not be fair to lay the blame for the Jays inconsistency at the feet of the Belt. But, it might also be fair to say that he's not well cast to the role in which the Jays have him. 

Likewise, the fascination with exit velocity -- the speed of the ball coming off the bat -- and distance for HRs create other problems.  It does not matter how hard a ball is hit if that speed is not producing actual hits. Vladdy hits the ball harder than just about anyone. But, that velocity is not translating itself into either hits or RBIs. Again, don't get me wrong. Vladdy is an exceptionally good younger player who I'd want on my team. He's also the opposite of Belt. He got off to a blazing start.  Hitting .309 by at the end of April (I get my data from Baseball Reference) with an .885 OPS. We don't need to worry too much about precisely what that stats means but it all-star level. He's been up and down with a lot of down since.  Right now, he is not on pace to drive in 100 runs, something you need from your best player. In this case it is not an analytic stat that is disguising a problem but the fascination new measurable stats. 

Let me make this point clear: if a ball is a HR, what difference does it matter how far it traveled? It might be interesting for fans, but not relevant to the score. If you need to hit the ball 400 ft to get a home run, what difference does it make if the ball is hit 401 feet or 456 feet? If you hit a home run, what difference does it matter that its exit velocity was 106 mph or 90 mph. If you said "none," you were right. 

I didn't get to boredom and baseball so I'll carry that into another post, but finally, let me say a word about "best in the league since date X" syndrome. Baseball loves this kind of language and the Jays media is riven with it. But, it is also a problem. Again: I get it. Media commentators are trying to show a trend -- a player heating up or someone who has made an effective adjustment. But, it also misses a point and that point is this: it does not matter when you lose games if you lose them. For instance, imagine that a team needs to win, say, 90 games to make the playoffs. That means that they will lose 72 over the pan of the season. Does it matter when they lost those games? Again, the answer is no.  The fact that a player gets off to a bad start is not a reason to throw them away and. there will be ups and downs over a season. But the fact that a player is playing well at a certain moment and not at another moment could be a problem for a team that is looking for consistency, as the Jays are. Said differently, this start -- or this approach to reporting stats -- disguises something important. Playing poorly and losing games at the start of the season can keep. team out of the playoffs as surely as playing poorly and losing games late in the season. 

The Oddities of the Jays Approach to Pitching

Despite what it might look like, pitching decisions in baseball are not made willy-nilly.  And, while the manager is responsible for them (and has the ultimate on-field call), they are no longer made simply by managers or managers alone.  Instead, pitching decisions are charted out. While everyone would like a starter to go as long as they can, baseball coaching staffs chart out numerous possibilities. If our team is ahead in the seventh by X+ runs, and the starter is tired or losing the zone or has hit their pitch limit, they are replaced by A. If we need a platoon advantage, they are replaced by B, if we are behind ... and we needed to replace the starter .... you get the point. The manager makes the decision -- the call on the field -- but he's had the benefit of advise from pitching coaches, his bench coach, there is a guy down in the bullpen, and the team analytics department. The individual decision that we, as fans, see is actually a collective decision made by a baseball team's brain trust. 

With that in mind, the Jays brain trust makes decisions that I often find difficult to figure out. Put differently, that defy figuring out. I'll give you an example. On Tuesday night the Jays got battered by the Orioles 13-3. It was likely a game that they had no chance of winning regardless of individual decisions because the Os were on and the Jays were not. But, the game marked the return of Hyun-Jin Ryu to the staff after over a hear off recovering from surgery for a major injury. Ryu looked good and pitched well. He got off to a bit of a bumpy start, but was efficient. He was throwing strikes, not going deep into counts. He gave up two in the first and another in the second but blanked the Os for the next three innings.  At this point, he had thrown -- I forget the exact number but -- high 70s in terms of pitches, which is not a tonne for a starter.  Why not pull him? He'd look good, the Jays were in the game, and as a team the Jays had accomplished what they set out to accomplish. They got a good game from a starter who they want to re-integrate into their starting rotation. 

For reasons I don't understand, the Jays sent him back out for the 6th, he was promptly touched up for a home run and was then pulled. My question is this: what was the difference between the 5th and the 6th. Why was that one batter important? And, why was in more important than keeping the score as low as possible and getting a pitcher back into the swing of major league games? There may be a logic here but I can't see it: what difference did the handful of pitches Ryu threw in the 6th make other than to put the Jays behind. Because the Jays scored no more runs, they would never have caught up even if the Orioles hadn't battered our bullpen, It is an odd decision.

Odder still: last night. The Jays were short staffed. With both Romano and Richards on the IL, they were clearly hoping for a long start from Gausman and that was a reasonable expectation. He's pitched very well this year. It was not to be. Gausman ran into trouble in the 2nd and while he recovered his pitch count soared. One out into the 5th, he'd thrown 103 pitches and so he needed to come out.  Gausman had given up another run in the 5th as well. The Jays turned to Bowden Francis who had just been recalled and he did the job. Francis has actually pitched rather well for the Jays but in limited action. He's spent most of the year in the minors. After Francis came the newly acquired Cabrera. Fair enough, this is what the Jays acquired him for.  But, at this point odd decisions started. 

The Jays really needed to win this game. It was not critical but losing it dropped the Jays 7.5 games back of the Orioles with 52 games left in the season, an almost insurmountable lead. Losing the game meant that the Jays were, in effect, conceding the division championship. The bullpen, as I said, was thin but the Oriels only had a 3-1 lead. The game was still in range. And, with this important game on the line, the Jays brain trust elected to turn the Jays fate over to Thomas Hatch, a player who they had just recalled from the minors and, previous to last night, had thrown a total of 4 innings in the bigs this season. 

Let's think about the logic of this. Here we have a very important game that the Jays need to keep the division championship in their sights. If they win, they're 5.5 games back instead of 7.5. Still a tall order but close enough that a mini-losing streak for the Os is all the Jays would need to get within a game or two, and we play them again. So, is this the situation in which you would turn to a guy who 24 hours before you did not think was good enough to be in the majors? Quite literally, this is a guy who, barring injuries, would not even have been on the team. The options the Jays had were limited and I'll get to that in a minute. But, there were options. Mayza is the obvious one.  Hicks would have been a stretch but apparently he had said (at least according to Dan and Buck) that he could go, and Swanson was the other option. I suspect Swanson was being saved to be the closer but since Hatch surrendered 3 runs in the next 2+ innings, both Mayza and Swanson ended the game on the bench. 

Making matters even more odd, the Jays designated Hatch today. In baseball terms that means, in effect, they fired him. Any other team can take him for his salary and need to provide the Jays with no compensation. If no one else claims him, I suspect the Jays will keep him in the minors but let's process this. The Jays trusted a critical situation in a critical game to a guy who they themselves did not feel was good enough to be on staff and who, the next day, they didn't feel was good enough to keep on the roster. Does this make sense? 

How did we get here? The Jays will claim that the thin staff -- injuries and another player on leave -- forced them to. That is likely exactly what they think, but it just does not make a lot of sense. There is another reason they are short staffed and that relates to the construction of their pitching staff. 

There is nothing unusual about the way in which the Jays have built their pitching staff. They have it set up  like this: the starter gets through 6 innings (or, at least 5) and the Jays then had it over to the bullpen. A series of pitchers in succession pitch 1 inning (or, less if there is a needed platoon advantage). Just about everyone does the same thing. One of the interesting effects analytics has had on baseball is to flatten out strategy differences. Everyone looks at the same data, everyone has analytics departments that crunch numbers, and, on an aggregate level, the data shows something so everyone does exactly the same thing. 

I won't get into the details of why this is done. I'm not trying to say that there is no a reason for it. It just has not served the Jays well. The Jays carry a tonne of pitchers (which also limits in-game decisions, an issue I'll address another time but also something else that is very common), usually 8 relievers and 5 starters but right now 7 relievers and 6 starters because Ryu and Manoah are both back and the Jays intend to go with a six-pitcher rotation because they have a spate of games without a day off. I don't inherently disagree with this decision but it further cramps the bullpen and what happens if ... if, say, your starter has a bad inning and reaches his pitch count early? What happens is that you turn a critical game over to a guy who was not on the team a day or two before and is not now.  

It is a bit like crying over spilt milk, but can I ask about the logic of one-inning relievers? The only guy on the Jays staff who regularly pitches more than an inning in relief is Richards and even he would not meet the definition of "long man" of my youth. The Jays bullpen lacks depth not in the sense that they don't have arms. They do. I think its a good bullpen if each game unfolds according to the 6-1-1-1 formula. But the absence of longer relief was high-lighted last night. Since just about everyone builds their bullpens in the same way, I wondered as I was thinking about last night's game, has anyone looked back on what managers like Felipe Alou did in the past? Alou worked his pens hard but tended to use a much more limited number of pitchers and kept relievers in longer than people do today. 

The Jays handling of the pitching staff, then, was odd in two ways. They made a really odd in game decision that they did not have to make and they have constructed their bullpen in a way that seems to force odd decisions, or at least make a greater chance for them.  I'll be honest. I'm disappointed in how the Jays have played this year and I am looking for reasons that they've under performed. They are a good team and have not played poorly. But, they've not played to the level I thought they would and part of that, I think, relates to the oddities of their approach to the pitching staff.