Excellent stuff again Archi! It's a great follow on to your conversation with John Gennaro on the Section 1904 podcast.
Maybe you could mention somewhere here when you appear on other podcasts?
Also you got good mentions last week on both Ben & Woods and the 97.3 Padres Roundtable about your Compounding post. Plus 17 comments on the Reddit post for that, whoa! (just ignore the fact that most of those comments were 2 people sniping at each other about Xander...)
Interesting concept but I see a need for a little more nuance. Using the 30% of bips going for hits is a bit too simplistic.
The kind of contact matters significantly. Bunts don't go for hits 30% of the time. Nor do weak ground balls. This is why people don't bunt with 2 outs and RISPs.
A better way of thinking about it might be to maximize your xba, not just minimizing your strikeout rate.
A deeper analysis would include walks. With the bases loaded down a run or in a tie game in the bottom of the 9th in a 3-0 count, swinging and making contact with a 30% chance of getting a hit is suboptimal to taking and potentially drawing a walk.
The REK stat is interesting, but teams are very aware of the value of a K in a high leverage spot. That's precisely why high K% relievers pitch the late innings of a game and why low K%, high GB% pitchers (like Graterol, Hill, etc) are reserved for situations where you need a DP and not a K.
Seen another way, last season there were 2,978 PAs against a reliever in a 2 out, w/RISP, high leverage situation. This should capture whether teams are trying to optimize for Ks or BIP. In that situation, batters had a 24.4% K%. If instead we look at the 4,549 high leverage, w/RISP PAs with zero or 1 outs, the K% drops to 21.6%. Are batters more willing to K with 2 outs than with no outs? That seems unlikely. What is happening is that pitchers (and managers through their choice of RPs) are clearly prioritizing getting a strikeout in that spot. The large increase in BB% with 2 outs supports that pitchers are going all in on the K and that batters aren't expanding the zone on their own. Now, our low K% batting approach and fire bullpen makes us very well suited to benefit in these spots where a ball in play is more valuable, but I think that's generally true and not a result of a change of approach in these situations.
Finally, on the WPA calc, one thing BABIP doesn't take into account is HRs. In a tie game, one run is very valuable, but the second run is quite valuable as well and decision to totally wipe out the possibility of the HR should be captured in the WPA calc.
I don't have a perfect way to measure that, but while we currently have the 2nd best K% and 5th highest BA with 2 outs and RISP, we also have the 2nd lowest ISO, no HRs, 11th best wRC+ and 10th best runs per PA w/RISP. In certain situations playing for 1 run makes sense, but doing it early in a game or with the wrong hitter (see all of Arraez's stupid bunt attempts) is one way of ensuring you are in a close game later on.
I would think there is a small but non-trivial benefit to DELAYING getting REKd. That is, the more pitches the batter sees, the more WPA, even if the batter eventually strikes out. Every pitch that doesn't end in a K carries with it a potential passed ball, wild pitch, or even a balk.
Love this. Hope the team has used analytics like this consciously to strategize and coach the players.
One nuance to this is the following:
While the successful (batter gets on base) BIP is approx 30%, I wonder how “make contact / don’t strikeout swings” do. Impossible to measure I would guess. (Maybe full swing speed before two strikes and with two strikes would be the tell but I’d have to think about it more.)
I would hypothesize that the successful BIP percentage would be marginally lower with two strikes vs an aggregated BIP number across all non-two strike counts.
However, I would also hypothesize that the lower BIP figure would still result in a WPA that is still largely positive. Effectively, the two strike swing only has a marginal impact on the WPA figure.
Certainly if a swing becomes too defensive the BABIP is likely to suffer, but a competitive swing with a quieter lower body can still produce damage while slashing the whiff rate. Juan Soto and Bo Bichette are really good at that.
Excellent stuff again Archi! It's a great follow on to your conversation with John Gennaro on the Section 1904 podcast.
Maybe you could mention somewhere here when you appear on other podcasts?
Also you got good mentions last week on both Ben & Woods and the 97.3 Padres Roundtable about your Compounding post. Plus 17 comments on the Reddit post for that, whoa! (just ignore the fact that most of those comments were 2 people sniping at each other about Xander...)
Interesting concept but I see a need for a little more nuance. Using the 30% of bips going for hits is a bit too simplistic.
The kind of contact matters significantly. Bunts don't go for hits 30% of the time. Nor do weak ground balls. This is why people don't bunt with 2 outs and RISPs.
A better way of thinking about it might be to maximize your xba, not just minimizing your strikeout rate.
A deeper analysis would include walks. With the bases loaded down a run or in a tie game in the bottom of the 9th in a 3-0 count, swinging and making contact with a 30% chance of getting a hit is suboptimal to taking and potentially drawing a walk.
The REK stat is interesting, but teams are very aware of the value of a K in a high leverage spot. That's precisely why high K% relievers pitch the late innings of a game and why low K%, high GB% pitchers (like Graterol, Hill, etc) are reserved for situations where you need a DP and not a K.
Seen another way, last season there were 2,978 PAs against a reliever in a 2 out, w/RISP, high leverage situation. This should capture whether teams are trying to optimize for Ks or BIP. In that situation, batters had a 24.4% K%. If instead we look at the 4,549 high leverage, w/RISP PAs with zero or 1 outs, the K% drops to 21.6%. Are batters more willing to K with 2 outs than with no outs? That seems unlikely. What is happening is that pitchers (and managers through their choice of RPs) are clearly prioritizing getting a strikeout in that spot. The large increase in BB% with 2 outs supports that pitchers are going all in on the K and that batters aren't expanding the zone on their own. Now, our low K% batting approach and fire bullpen makes us very well suited to benefit in these spots where a ball in play is more valuable, but I think that's generally true and not a result of a change of approach in these situations.
Finally, on the WPA calc, one thing BABIP doesn't take into account is HRs. In a tie game, one run is very valuable, but the second run is quite valuable as well and decision to totally wipe out the possibility of the HR should be captured in the WPA calc.
Hit post too soon...
I don't have a perfect way to measure that, but while we currently have the 2nd best K% and 5th highest BA with 2 outs and RISP, we also have the 2nd lowest ISO, no HRs, 11th best wRC+ and 10th best runs per PA w/RISP. In certain situations playing for 1 run makes sense, but doing it early in a game or with the wrong hitter (see all of Arraez's stupid bunt attempts) is one way of ensuring you are in a close game later on.
I would think there is a small but non-trivial benefit to DELAYING getting REKd. That is, the more pitches the batter sees, the more WPA, even if the batter eventually strikes out. Every pitch that doesn't end in a K carries with it a potential passed ball, wild pitch, or even a balk.
100%
Walk him.
Love this. Hope the team has used analytics like this consciously to strategize and coach the players.
One nuance to this is the following:
While the successful (batter gets on base) BIP is approx 30%, I wonder how “make contact / don’t strikeout swings” do. Impossible to measure I would guess. (Maybe full swing speed before two strikes and with two strikes would be the tell but I’d have to think about it more.)
I would hypothesize that the successful BIP percentage would be marginally lower with two strikes vs an aggregated BIP number across all non-two strike counts.
However, I would also hypothesize that the lower BIP figure would still result in a WPA that is still largely positive. Effectively, the two strike swing only has a marginal impact on the WPA figure.
Thoughts?
Certainly if a swing becomes too defensive the BABIP is likely to suffer, but a competitive swing with a quieter lower body can still produce damage while slashing the whiff rate. Juan Soto and Bo Bichette are really good at that.
Fair and that’s when judgement and context comes into play. A “defensive swing” from Manny is very different than from Maldonado.