Why you should consider using defensive shifts more in 14

[JMDurron]

I'm never entirely certain whether OOTP models the apparently powerful influence of defensive shifting accurately, since so much of the data indicating the importance of it is relatively new (particularly since the Joe Maddon Rays Era has begun), so I don't shift as much in OOTP as I would if I were somehow a real-life manager.

Against pull hitters (ANY pull hitters), I don't shift until they get 1 hit, then shift until they have 2 (clearly, if they get another one, then my shift didn't work, right?), then shift if they have 3, and so on. If there was an indication from the devs that OOTP was weighting IF shifts as much as the real data seems to indicate they should be weighted, I'd shift all the time against all pull hitters.

[The Wolf]

Why not ask Markus?

[JMDurron]

Originally Posted by The Wolf

Why not ask Markus?

Well, two reasons

1) My "if there was an indication" language was an opportunity for Markus to ride in on his white horse of awesomeness to give such an indication

2) It's opening Pandora's box in one regard when it comes to another set of House Rules being needed.

To expand on #2, if IF shifting really is as powerful as it appears to be IRL, then OOTP should model that impact. If OOTP models that impact, then for the sake of realism, it would also have to model the fact that current and historical managers have not been using IF shifting as often as is optimal, and have appropriate strategy settings for the AI that reflect that. This means that in current day games, the human player would have a slight advantage over the AI teams by aggressively shifting, Maddon-style. It also means that in historical games, since there is no reason to expect shifting to be less useful in different eras once you get outside of the deadball era maybe, given that historical managers hardly used it at all, the human player would have a very significant strategic advantage over the neanderthal AI managers when it comes to defensive shifts.

There are two potential ways to address this.

1) Markus waters down the impact of IF shifting in historical contexts. I don't like that solution, because I don't believe there would be any realistic validity to it, as a pull hitter is a pull hitter regardless of era (again, once you get outside of the bunt-crazy deadball period), and a groundball is a groundball, so I'd expect the impact of shifting to be consistent from roughly 1920 to present.

2) House rules for players, to not shift more than was realistically done historically, AKA not at all. This is my preferred solution, but it's a case of more knowledge and transparency into the combination of real world effects, the OOTP simulation model, and AI manager tendencies giving the human player yet another edge over the AI. I happen to think that's appropriate for this particular issue, since it's reasonable to me that a modern day manager, if transported into the past to manage, really would use shifting more than his historical counterparts. I also happen to think that there's a not insignificant subset of players/posters who are constantly on high alert for any additional advantages conferred upon the human players against the AI, and who might therefore lobby for solution #1.

[The Wolf]

All good points there.

[VanillaGorilla]

Originally Posted by JMDurron

Well, two reasons

1) My "if there was an indication" language was an opportunity for Markus to ride in on his white horse of awesomeness to give such an indication

2) It's opening Pandora's box in one regard when it comes to another set of House Rules being needed.

To expand on #2, if IF shifting really is as powerful as it appears to be IRL, then OOTP should model that impact. If OOTP models that impact, then for the sake of realism, it would also have to model the fact that current and historical managers have not been using IF shifting as often as is optimal, and have appropriate strategy settings for the AI that reflect that. This means that in current day games, the human player would have a slight advantage over the AI teams by aggressively shifting, Maddon-style. It also means that in historical games, since there is no reason to expect shifting to be less useful in different eras once you get outside of the deadball era maybe, given that historical managers hardly used it at all, the human player would have a very significant strategic advantage over the neanderthal AI managers when it comes to defensive shifts.

There are two potential ways to address this.

1) Markus waters down the impact of IF shifting in historical contexts. I don't like that solution, because I don't believe there would be any realistic validity to it, as a pull hitter is a pull hitter regardless of era (again, once you get outside of the bunt-crazy deadball period), and a groundball is a groundball, so I'd expect the impact of shifting to be consistent from roughly 1920 to present.

2) House rules for players, to not shift more than was realistically done historically, AKA not at all. This is my preferred solution, but it's a case of more knowledge and transparency into the combination of real world effects, the OOTP simulation model, and AI manager tendencies giving the human player yet another edge over the AI. I happen to think that's appropriate for this particular issue, since it's reasonable to me that a modern day manager, if transported into the past to manage, really would use shifting more than his historical counterparts. I also happen to think that there's a not insignificant subset of players/posters who are constantly on high alert for any additional advantages conferred upon the human players against the AI, and who might therefore lobby for solution #1.

I will agree with The Wolf and say these are all very good points.

Some added thoughts:

Baseball is a game of adjustments, and it has evolved over time, and will continue to evolve. One difference now, than even 20 years ago, is the ability for hitters to pull up video of pitchers vs themselves, similar batters, or any batters in any situation and look for trends and assisting them in "guessing" pitches. Knowing that their history can be compiled and parsed by future opponents to the degree that it is now results in an incentive to consciously not to pitch to form every occasion. If a big league hitter can guess what is coming, he can hit it. Just ask a gimpy, but fist pumping, Kirk Gibson in 1988.

As the idea of shifting catches on, as all successful ideas do, hitters will make adjustments and value will be placed on players with the ability to hit against shifts. As more shifting occurs, more players will hone this ability. 40 years ago, Dominican players had a saying that "You don't walk your way off an island" that explained the free swinging nature of these players. Scouts looked for guys that made contact and those were the guys that got a trip to the continent. Walks were not valued by teams and scouts, as they are now.

In the last 15 years, or so, making contact has been devalued in favor of power and walks. As shifts become more mainstream in their frequency of use and degree, players with the ability to "hit 'em where they aint" will be of greater value.

A simulation must make assumptions in how RL data is translated into the simulation, itself. OOTP takes the historical data*, which is the result of whatever conditions were in existence, and uses a normalization process to translate the base data into a universal output for whatever type of environment we want to create.

If shifting is to be reflected in the game as a way of diminishing a player's offensive chances to a greater degree than it is in use now, the game must also give players an ability to "hit against a shift" that is separate from whether or not they are a pull or spray hitter.

How do you work that into the game? Would a more intelligent hitter be better at recognizing a game situation where it is advantageous to give up 150 slugging points to gain 150 batting average points, for instance? Would a selfish player just not care and want to hit the HR regardless of game situation vs a shift? Would a pull hitter with a high work ethic be more likely to learn to hit against the defense than one without a high rating?

To reflect the more nuanced shifts described in the linked article in the OP, OOTP would need to employ a much more diverse batted ball location grid.

I am in favor of the game being more detailed, but there is more than one part to the equation as far as OOTP working the use of advanced shifts into the game.

One final note: The article mentions the improved fielding% and balls fielded by the Pirates because of the shifts. There have been many discussions, especially in the Suggestions forum, regarding what impact, if any, coaching has on player performance. In the case of the Pirates, it appears that coaching has indeed had a positive impact on the fielding numbers of the Pirates, which has had a positive impact on the pitching numbers.

[The Wolf]

Excellent point about the coaching. Being more scientific clearly has made a difference.

[JMDurron]

Originally Posted by VanillaGorilla

I will agree with The Wolf and say these are all very good points.

Some added thoughts:

Baseball is a game of adjustments, and it has evolved over time, and will continue to evolve. One difference now, than even 20 years ago, is the ability for hitters to pull up video of pitchers vs themselves, similar batters, or any batters in any situation and look for trends and assisting them in "guessing" pitches. Knowing that their history can be compiled and parsed by future opponents to the degree that it is now results in an incentive to consciously not to pitch to form every occasion. If a big league hitter can guess what is coming, he can hit it. Just ask a gimpy, but fist pumping, Kirk Gibson in 1988.

As the idea of shifting catches on, as all successful ideas do, hitters will make adjustments and value will be placed on players with the ability to hit against shifts. As more shifting occurs, more players will hone this ability. 40 years ago, Dominican players had a saying that "You don't walk your way off an island" that explained the free swinging nature of these players. Scouts looked for guys that made contact and those were the guys that got a trip to the continent. Walks were not valued by teams and scouts, as they are now.

In the last 15 years, or so, making contact has been devalued in favor of power and walks. As shifts become more mainstream in their frequency of use and degree, players with the ability to "hit 'em where they aint" will be of greater value.

A simulation must make assumptions in how RL data is translated into the simulation, itself. OOTP takes historical which is the result of whatever conditions were in existence and uses a normalization process to translate the base data into a universal output for whatever type of environment we want to create.

If shifting is to be reflected in the game as a way of diminishing a player's offensive chances to a greater degree than it is in use now, the game must also give players an ability to "hit against a shift" that is separate from whether or not they are a pull or spray hitter.

How do you work that into the game? Would a more intelligent hitter be better at recognizing a game situation where it is advantageous to give up 150 slugging points to gain 150 batting average points, for instance? Would a selfish player just not care and want to hit the HR regardless of game situation vs a shift? Would a pull hitter with a high work ethic be more likely to learn to hit against the defense than one without a high rating?

To reflect the more nuanced shifts described in the linked article in the OP, OOTP would need to employ a much more diverse batted ball location grid.

I am in favor of the game being more detailed, but there is more than one part to the equation as far as OOTP working the use of advanced shifts into the game..

Good post, but I think you're mixing some micro vs macro considersations here.

At the high level, OOTP classifies hitters as Pull Hitters, Spray Hitters, or Normal. Normal hitters should not be shifted against. I'm not sure if there's any argument for opposite shifts against spray hitters. Pull hitters should be shifted against.

Hitters who are able to consistently beat IF shifts, in my mind, should not be considered Pull Hitters by OOTP. Since teams are using specific player hit charts to determine whether or not to bother shifting against them IRL, then I feel like OOTP is simulating this process already just by accident via the Pull/Spray/Normal designation. Hitters who can consistently beat shifts aren't shifted against today, so we're only dealing with the subset of "Pull Hitters" to begin with as we're discussing what the impact should be against those hitters.

Since OOTP is simulating the end result statistics of the batter-vs-defense matchups, I think we would be best served to think of that end result as already having considered the shift-beat-change classification-adjust process.

I think you're thinking in terms of the individual batter analysis, while OOTP needs to be treating this at the higher level. Today, a hitter shows a pattern of balls put in play. Defensive teams analyze this data. For hitters who real world terms classify as "pull hitters", good defensive teams shift against them. If the defensive team is wrong about that hitter being a pull hitter, and the hitter is really a "normal" or "spray" hitter, then that hitter will beat the shift. The defensive team will adjust back to not shifting against that hitter. Hitters who are impacted by shifts should be classified by OOTP as "Pull Hitters", and therefore vulnerable to IF shifting mechanics. Hitters who hit the end result of Normal or Spray, even if there was a brief moment in time where a shift increased or decreased their capability in a particular game, or even in a particular week or month, should be classified as such by OOTP.

I don't think that OOTP needs to model the overall changing nature of the batter-vs-pitcher adjustments, because the end results of the capabilities should be reflected in the player classifications and ratings.

As an equivalent example, if a hitter has a month of hot performance with high contact, then teams adjust and he slumps, then he adjust and becomes a medium-contact, medium-power hitter, I don't think that OOTP should have a rating of that player (let's say it's a real player that OOTP is rating for the following season in the MLB roster pack) that changes based on when exactly those streaks happened, OOTP should give that player an overall aggregate rating based on the sum total of that performance. It doesn't mean that adjustments weren't made and that the player's performance didn't change, it just means that for OOTP's purposes, a final classification/rating at the end of that performance window should adequately capture the overall capability, which will then be influenced by the usual mix of hot/cold/ballpark/opposing defense/opposing pitcher factors when playing the game with that player.

I may have been unclear here, if so I apologize. I don't think further rambling will help at the moment, so I'll re-attack this point later today if necessary.

[VanillaGorilla]

Originally Posted by JMDurron

I may have been unclear here, if so I apologize. I don't think further rambling will help at the moment, so I'll re-attack this point later today if necessary.

I know nothing of this "rambling" you speak of.....OK, I was trying to avoid it myself, but I think I left out something in what I worte that I was thinking that you addressed.

The thing about teams shifting against pull hitters in RL, like Dunn and Ortiz, for example, is that the stats that are compiled by Dunn and Ortiz, upon which their ratings are based, come from having these shifts against them. Do you double penalize these hitters by instituting a shift on them in the simulation of RL when the resultant outcome used by the simulation is derived from the player having faced a shift IRL? This would not be an appropriate simulation of outcomes.

We need to be keenly aware of what we adjust and from what base we are adjusting.

I agree that players who demonstrate the ability to hit against a shift are not shifted on. Their overall hitting numbers are based on facing "normal" defenses, even if their tendency when swinging away is to pull. It is their ability to go the other way when given a golden opportunity to do so that results in more pulled balls being hits than if a shift was on.

A hit chart will still show them as tending to pull, because that is their tendency. Shift on them, and they bat .480. So, you don't shift and they maybe bat .290 where as the same pull hitter that can't hit against a defense efficiently hits .260, let's say.

To take that .260 hitter and bring him down to a .230 hitter by shifting in the sim would be a terribly inaccurate representation of game play. Just as it would be inaccurate to take the hit chart that shows the same ball placement by the RL .290 pull hitter who has dribbled 6 balls over a vacated 3B when he was shifted on in April (and because of that display hasn't seen a shift all summer) to make him a .260 hitter by clicking a "Shift" button in an OOTP simulation.

[JMDurron]

Originally Posted by VanillaGorilla

...

The thing about teams shifting against pull hitters in RL, like Dunn and Ortiz, for example, is that the stats that are compiled by Dunn and Ortiz, upon which their ratings are based, come from having these shifts against them. Do you double penalize these hitters by instituting a shift on them in the simulation of RL when the resultant outcome used by the simulation is derived from the player having faced a shift IRL? This would not be an appropriate simulation of outcomes.

...

To take that .260 hitter and bring him down to a .230 hitter by shifting in the sim would be a terribly inaccurate representation of game play. Just as it would be inaccurate to take the hit chart that shows the same ball placement by the RL .290 pull hitter who has dribbled 6 balls over a vacated 3B when he was shifted on in April (and because of that display hasn't seen a shift all summer) to make him a .260 hitter by clicking a "Shift" button in an OOTP simulation.

Editing down for scrollability for everyone else.

The double-counting is an excellent point. In order to properly apply this effect, you'd almost need PbP data (or perhaps a particular year in time at which you apply a "shift impact", or lack thereof to their underlying ratings) to adjust for which hitters did or did not accumulate their numbers against defensive shifts.

I don't think there's going to be a simple solution to this one.

[Biggio509]

Originally Posted by JMDurron

Well, two reasons

..
To expand on #2, if IF shifting really is as powerful as it appears to be IRL, then OOTP should model that impact. If OOTP models that impact, then for the sake of realism, it would also have to model the fact that current and historical managers have not been using IF shifting as often as is optimal, and have appropriate strategy settings for the AI that reflect that.

There are downsides to the shift. The pitcher often has to pitch differently with a shift. Lucas Harrell was a bit outspoken on this earlier this season as Bo Porter has adopted a more TB style. Hence, the article "Bo don't know shifts". This is actually pretty good writing for the Houston Press. It is usually just spouting off with little research. This seems to be an exception.
Bo Don't Know Shifts: Astros Manager's Defensive Shifts Rankle Pitcher - Houston - News - Hair Balls

Like the article points out a good part of the reason we are seeing more shifts is technology and interest in SABR. A manager in the 1950's didn't have a computer screen showing him the breakdown of where someone hits. It was done by feel rather than analyzing hard statistics. Today you have press box guys pulling up the data in seconds and using all kinds of new stats to suggest a shift.

I don't think the shift is any more powerful than it every was. We just see it more because the data is easy to obtain and clubs hire people to pull data to give to the manager to make decisions. This isn't Aarron is up and I think he usually pulls so lets shift. It is this guy hits the ball here x% of the time. The information age is impacting baseball like everything else. Hence, the "debate" of scouts vs. stats guys raging now.

It is not the shift works better now. It is that managers in previous eras just didn't have quick access to the data to fully exploit the shift. It was more guess work than exploiting probabilities in by gone eras.

[RonP]

That was a remarkable article that Wolf turned us on to and I'm guessing, well not much guessing involved really, that shifting your fielders is the next "big thing" in baseball.
Ron
PS. I know that I am doing it now in my strategy settings. Rather, I am trying to do it. It takes a little thought.

[The Wolf]

Originally Posted by RonP

That was a remarkable article that Wolf turned us on to

Thank you.

[PSUColonel]

What data are you using in OOTP to suggest you play a shift? Other than hitter type: pull? Just a player is labeled a pull hitter does not quantify how often he actually does pull his hits.