Pennants, Bell Curves, Baseballs, & Sir Isaac Newton

Why Baseball Needs Another Computation Package

By: Jim O'Flanagan

Editor’s Note: This article is part 2 of a multi-part series. It is syndicated from OAPSIE.com. Find the series homepage at http://ballparkview.com

For more than 25 years, mainstream professional baseball teams have applied probability & statistical theory to describe the game of baseball and win championships. Prior to the Sabermetric revolution, this approach represented a competitive advantage for any team that successfully employed it. In the present day, the professional baseball market has been saturated with this same probability & statistics technology, and the competitive advantage it represents has been severely diminished.

Consequently, we believe an opportunity exists for anyone willing and able to try something different. That different opportunity is physics! (read: Newton, apple, tree, gravity).

Further, we believe a physics based, real time computation package has a complementary role to play in any baseball teams’ R&D software package. Recent advancements in HPC hardware and AI/ML software technology provide an opportunity to finally analyze baseball players in real time, using physics quantities like Power (Watts, Horsepower), Force (pounds), Impulse (pounds-second), and Energy (calories) rather than statistics like mean and standard deviation.

First in Quantum Mechanics, and then in derivative finance and other areas, scientists have applied a combination of computation, bell curves, mean, and standard deviation to describe natural phenomena like the rate of a nuclear explosion (Quantum Mechanics) or the price of an interest rate swap (Black-Sholes equation).

Beginning in the 1980s, that same technology was applied to professional baseball to find undervalued players, price player contracts, and predict how many home runs someone might hit in the upcoming season; it is the basis for the WAR statistic and many more terms common in baseball today. All of these models make the central assumption that natural phenomena tend to follow a Gaussian (normal) distribution when a sufficiently large, random sample are measured. In other words, the data forms a Bell Curve.

Statistics are a model that describes past events. In baseball, and elsewhere, engineers then use that model to simulate the future and provide knowledge. As with any analysis model, errors can lie in the model assumptions. Sometimes the dataset is not large enough. Sometimes the phenomenon being described does not exactly follow a Gaussian distribution (tail risk).

You may have an undetected bias in your data. Sometimes you may want to measure something the model won’t allow, like the internal free body forces of a mechanical structure (a pitcher’s Tommy John ligament?). These sources of error are present in every piece of statistics based baseball software available today.

The solution, then, is to have another predictive model and software package at-the-ready. One that uses some model other than statistics to describe the natural world.

Enter physics.

Physics is just another model, and yet it’s not. Natural philosophy (Physics) was first developed by Sir Isaac Newton more than 300 years ago. This “model” in turn helped develop modern science and computers.

It helped put a man on the moon, and allows us to make near-real movie productions like Pixar films.

Its language uses terms like equal and opposite reactions, entropy, and conservation of momentum. It deals in things like pounds, calories, inches, horsepower, and watts as opposed to those used in statistics like discrete random variables, mean, variance, and standard deviation.

This article is not an argument for or against either approach; statistics and physics are both tools with long histories of successful application. The purpose of this article is to highlight the fact that physics has its own, (separate) history of development when compared to statistics. It has its own (separate) long timeline of empirical evidence to validate it. The apple fell to the ground and hit Sir Isaac on the head! The proof is in the pudding. Like statistics, physics is a model that uses assumptions and has its own limitations. Physics and statistics together, however, in a complementary software package, is a force multiplier. They cover for each other’s weaknesses and are able to predict things the other can’t.

By using them in conjunction, we aim to give baseball people more confidence in their judgements. Things will be known about player’s bodies that were not known before. #Ballparkview analyzes things like stadium wind flow and the effect of sunlight on a batter’s ability to locate a pitched baseball. All possible by incorporating a physics model into our analysis.

#Pitcherview is a computation package patented by OAPSIE that applies physics and computer power to analyze the motion of a pitched baseball. Recent developments in computing power and AI/ML software have made it possible to measure a baseball player’s motion and produce a real, game-time physics analysis of the pitching motion event, at the same time recording the event for future analysis.

By measuring the Power (in horsepower, like a Ford Mustang!) a pitcher is using to deliver the ball to the plate, and comparing that to their pitch velocity, you can get an idea of pitching effort efficiency. Is the pitcher using too much energy to bring an 85 mph (average) fastball? In that case, it may be time to remove the pitcher from the game. And this is just one example of the technology’s use!

This type of mechanical analysis is not possible with current statistics based methods. The investigation described above runs a Finite Element Analysis (FEA) model to determine forces on the inside of a pitcher’s shoulder, amongst other things. This shows that there are capabilities available to a physics model that is not currently captured with today’s statistics-based baseball analysis software.

Stadium wind flow is another example, and our #Ballparkview package takes a look at that. On a windy day, does a knuckleball work better than on a calm day? That is common knowledge, perhaps, but to what degree does the phenomenon exist? More importantly, can you quantify that degree to allow for comparison of games, players, and situations? We think we can with #Ballparkview.

Knowing this type of grounds related information allows a baseball person to do some cool things. Like knowing whether to send out your knuckleball catcher that day. How about knowing exactly how much bite a pitcher’s off-speed pitch is going to have that day?

That is done by looking at how the Magnus Effect and the Coriolis Effect work on a pitched baseball, and are affected by local weather conditions. Perhaps the manager should go with the off-speed location specialist rather than a power pitcher? Depends on the weather; more exactly, it depends on the degree of difference the weather affects the pitcher. Ballparkview allows a knowledgeable baseball person to measure that. More importantly, we know that we can quantify the difference between the two more exactly than before and make a sound baseball decision.

OAPSIE has patented a computational methodology to analyze the types of stadium and grounds effects on the game of baseball, along with the baseball pitching motion. Those modules are called #Ballparkview, #Batterseyeview, and #Pitcherview. They are all part of OAPSIE’s omnibus patent filing for #Pennantview. There is a module in Pennantview to cover each phase of the game, plus one support module.

The modules are:

• Pitcherview; Pitching, Motion
• Hurlerview; Pitching, Anatomy
• Batterview; Hitting, Motion
• Hitterview; Hitting, Anatomy
• Runnerview, Baserunning, Motion
• Throwerview; Fielding, Throwing Motion
• Ballparkview; Grounds, Wind Flow
• Batterseyeview; Grounds, Sunlight
• Saberview; Data, Backend, Support

We have written about Ballparkview, Pitcherview, Pennantview, and Batterseyeview previously, and will continue to develop these technologies. We plan to release more information on the others as time allows.

To support that goal, here are our current and upcoming public presentations:

• We are presenting #Ballparkview at #SABR51 in July 2023.
• We are presenting #Pitcherview at #ASMEIMECE2023 in November 2023.

More details to come, and presentations to add to this list. Until then, thank you very much for reading, following along, and supporting us. We sincerely appreciate it!