Batter/Pitcher Interactions!

[alexfall862]

Hi! 

Trying a new composition technique I borrowed from @Rocketcan after simFBA.com ate my Nebraska team preview article without saving a draft. This has been drafted in google docs. It should display correctly, but let me know if it doesn't.

I started writing Part 2 of the roster/ability/strategy breakdown, and then realized that the way the game handles baseball actions needed to give context to all the different tendencies/strategies in Part 2 of the player creation and strategy walkthrough - and then that aside grew until it was definitely too big to make part of another dev diary, as I tend to write a lot anyways.

So today we’re taking a break from the macro to talk about the micro and dive into the Batter/Pitcher Interaction!.

Batter/Pitch Interaction

The very nuts and bolts of the sim (just like irl baseball) is built on the interaction between the pitcher and batter. There appear to be two schools of thought on how to simulate this among baseball games. 

The first is to take the outcome and work backwards - so measuring the singles/doubles/triples/HR/strikes/walks etc. and assigning a weighted average from there. Most online simulations appear to take this approach.

The other approach is to break down the plate appearance into a series of pitches, with the likelihood of each particular outcome being derived from the same statistics as above, but broken down pitch by pitch instead of plate appearance by plate appearance. 

These levels of abstraction work decently well, but have a couple drawbacks - namely that digging into cause/effect relationships in baseball beyond the outcome of the macro batter/pitcher matchup means that it becomes difficult to meaningfully add consequences or user input to pitcher or pitch choices beyond good pitcher = better outcome and bad pitcher = bad outcome.

simMLB attempts to capture a lever deeper than that. What I mean by going deeper than the batter/pitcher matchup is that simMLB doesn’t abstract the pitch itself as an aspect of the pitcher’s ability, but it is itself it’s own game ‘event’ even after it leaves the pitcher’s hand. There is a step in the sim where the batter is evaluated on their ability to identify and take action on the placement and type of pitch itself.

This should ultimately allow for things like pitch location charts and batter spray charts - which other sims have, I think - and the ability to have influence over them - I don’t think that’s as common - as the meddling owner/manager/pitching coach that you are!

Right now, what is operational is the minimum viable product of that system (bare minimum to get functional and tuneable outcomes) where when a pitch is simulated, it takes contextual inputs like pitch choice, pitch ratings, the frequency of that pitch so far, and the number of times it’s been thrown consecutively and sends that to a batting function that takes those inputs and outputs what the batter is going to try to do. 

From there, the system simulates the batter acting on their intention, where they can access the game’s context in the same way as the pitcher, looking at strike/ball count, perceived placement of the pitch, perceived rating of the pitch (players), and their ability to then hit said pitch. This is tied largely to the eye rating (can they ID the pitch and placement) and discipline rating (can they make the ‘right’ choice given all the inputs).

Instead of outcomes being ascribed to success, like singles or home runs, the bat swing is then evaluated on its ability to make contact with the ball. 

The is broken down into: 

• Barrel
• Solid Contact
• Flare
• Burner
• Topped
• Under
• Poor

Which are shown in the graphic below, where distance from the player icon is the ‘power’ or exit velocity of the pitch, while the launch angle or ‘contact point’ of the ball :

And credit to statcast, because without their advanced metrics, this would have been impossible to derive and tune to irl baseball. 

Players with a higher contact rating can expect to live closer to the 0-45 degree range, while players with a higher power rating can expect to get outcomes closer to the outside of the ring. Players with a decent eye are more likely to have more events in this spread, rather than whiffing completely. This is a subtle difference in contact and eye between other sims like MLB The Show which I think it important to note. A player’s eye helps them to make contact with pitches, contact is the ability to make meaningful contact between 0-45 degrees - so a high contact low eye player is likely to make more hits, while a low contact high eye player is more likely to make contact with pitches (just not well). 

The way simMLB interactions or ‘actions’ will work is a little more granular than either, in an attempt to open up some matchup strategy and give more agency to the value of pitcher and pitch management.

So that’s the in/out and vertical axis of the ball, but we haven’t covered the horizontal axis! 

MLB players average roughly a 40/40/20 when it comes to pulling a pitch or pushing something to opposite field. This is something that’s partially based on a player executing the timing of their swing - but it also an aspect of intentionality in player’s batting approach.

The other major influence is direction of a hit pitch has to do with location: players looking to push a pitch to opposite field are looking for outside pitches as opposed to intentionally looking to a pull, which might involve waiting until you get an inside pitch. Discipline and eye play a factor is a player accurately identifying and executing the right move there if users dictate one approach over the other. 

So the pull/straightaway/oppo splits will depend on user input instructing players to behave in a certain way (more on that in another dev diary) or the timing and location identification of the hitter executing their swing lining up with the pitch.

Pitch Sequencing

Among other influences on pitches is that players get better at ‘seeing’ pitches the more often they get repetitions in a game against that pitch - this is known as the time-through-the-order effect, which has influenced some strategies in MLB lately - with the Rays experimenting with ‘openers’ to minimize the amount of matchups their ‘starter’ has against the heart of the order prior to the 6th/7th inning.

The effect is hypothesized to be from players getting to see specific pitches and get used to a pitcher’s spread of pitch choices.

This effect is captured by the game tracking two things: the times a particular pitch is thrown in a game, times a pitch is thrown consecutively.

The more a pitcher throws the same pitch, and particularly consecutively, the more a malus is applied to the ‘deceptive-ness’ of a pitch. In other words, a batter is more likely to succeed their eye rating check.  

This is a very slight malus, but the longer a pitching appearance goes, the more the effect begins to compound. Most pitching distributions should be fine - anywhere up to 40% should be more or less fine (still a very mild malus, but not much) while relievers who rely on 1 or 2 pitches better have some serious gas and break on their pitch in order to keep the dropping ‘deceptiveness’ of their pitch from letting them get thrashed.

This mechanic also creates an incentive for putting players with a wider but perhaps less powerful collection of pitch ratings into games as starters, and using players with 1 or 2 extremely solid pitches in as relievers until their full spread of pitches develops (or doesn’t).

A catcher’s sequencing rating plays into this dynamic not by mitigating the effect, but in helping a pitcher analyze which pitch is better after evaluating the pitch ratings after this malus is applied and choosing the most effective throw. 

User’s strategy plays into this as part of the strategy mechanics (more on this in another dev diary) where they choose and shape the pitcher’s baseline pitch choice, including but not limited to whether a player throws their 2nd through 5th pitch at all. 

What’s Next?

Once the ball is contacted, then it becomes a matter of fair/foul/direction/distance and whether the defense can field it successfully - which is a matter for a future dev diary!

Poll Results

Ok - so we had two votes and both were relatively close - but both are very meaningful for users in the sim.

First, y'all voted in favor of a mixed approach in roster construction, with random assignment and some kind of user input. The way that we'll work this is for the top 5 players on your team to be available through a fantasy draft, and then the rest of the roster randomly assigned.

The second vote was on the monetary structure of the league being created. The vote was close, but a majority of users are voting for that to be applied matched to current financial status of teams in irl MLB. Now, as a reminder, this impact is already impacted by the vote to split the revenue consequences in half - so with all teams at an equal record to begin the sim, I don't think this will generated a single team with revenue as low as 60m or as high as 350m. 

This Week's Poll

This week's poll is a place for you to input different ways you'd like to be influential in your batter/pitcher's approach. This will help me a ton in creating options for users to meaningfully and realistically provide the kinds of choices a manager/batting/pitching coach could influence a player's approach. 

Here's the link to the poll: https://forms.gle/wHqvwWsKtoXBkMhu7

Thanks,

alexfall862

[Sarge]

On 3/23/2024 at 8:02 PM, alexfall862 said:

 

This is a very slight malus, but the longer a pitching appearance goes, the more the effect begins to compound. Most pitching distributions should be fine - anywhere up to 40% should be more or less fine (still a very mild malus, but not much) while relievers who rely on 1 or 2 pitches better have some serious gas and break on their pitch in order to keep the dropping ‘deceptiveness’ of their pitch from letting them get thrashed.

 

Spencer Strider would like a word.