Assigning Point Values To Swing Decisions
I wanted to dig into swing decisions and start to look at how swings are quantified to hitters and coaches. Really this comes down to the simple idea of ‘Was this a good swing decisions or not?’ When answering this question I want to place the importance on one area in particular and obviously this can shift depending on the coach, but I want to focus on the idea of ‘Is this pitch a strike?’.
The goal is that we want to simplify swing decisions to a simply understood number. The strike zone is not a cut and dry issue. Strikes are called in the strike zone at very different rates. Some obviously in the heart of the plate we see more strikes called while on the boarders we see these percentages change drastically.
While Zone Swing and Out Of Zone swing from places like Fangraphs are great to look at the percentage of pitches inside/outside of the ‘legal’ zone are swung at by batters. I think that we can do a little more to dig into how players are determined to swing at good pitches. O-Swing and Z-Swing are both good places to start but it makes the assumption that all swings in the ‘legal’ zone are good while all those outside of the ‘legal’ zone are bad (which is pretty much correct). I have linked two primers for those who want to dig deeper into these areas.
Alright so far we see that the percentage of called strikes in different areas of the zone can change especially toward the corners of the strike zone. We are going to call the different Tiles that we see in the graph above the different Strike Probabilities that these different zones are.
I want to try to simplify this as much as possible for players and coaches. My thought is that if we can make this some kind of point system with swings with high strike probabilities ‘crediting’ players for these positive swings, while ‘debiting’ them for swings that are taken in less positive areas.
To do this I am going to make the 50% called strike mark the ‘middle’ or the areas of the strike zone which awards players 0 points for swinging in these areas.
So that means that points will be awarded to players like the image below.
So essentially we can start to award players different points for swings that they take in a game. If we are trying to optimize for strike zone swings I think that we start to create a process around this idea. It becomes sort of a game situation where players/coaches can quickly see different shades of where swings are taken. Obviously there will be different ranges for different points within this range of strike probability. What we end up getting on a daily basis is something like this:
We can start to give players some information that summarizes the different swings that they took during the course of a game. Yes we could put percentage or some other variable inside of these plots to give us more information, but I think potentially building a system that revolves around smaller, more actionable information makes it easier for players and coaches to digest on the whole. We could also present all swing that players took during a game based upon this information and present it in a digestible format for them.
When you bring them both together you get something that simplifies exactly what you are trying to optimize for:
Obviously we can shift these to match whatever want to optimize for. This being the first crack at taking on something like this I figured I would start with strikes, and see what could be done as we go along. Beyond simplifying the information you are presenting player this starts to build some clear ideas as to how a player did for a given day. You can see that the overall ‘ProbPoints’ that Mookie added for this day were +4.1 which starts to point a picture for the player of how he did.
Overall the feedback that we give players needs to be simple, but also in a format that allows them to gain specific feedback quickly. Something like assigning values to swings that a player takes can be a good step toward this goal, and something that should be considered when you start to give players feedback.
Simple, Fast Feedback
