National League's Best Pitching Staff

Billy Beane

"My job is to get us to the playoffs. Everything after that is fucking luck." - Billy Beane

Billy Beane was pretty much right. Over the course of a 162-game season, the best teams tend to make it to the playoffs. But in a short series, it's much more likely that the weaker team will advance.

Dickey's "Angry Knuckleball" Added To Baseball Mogul 2013

Photo by MG_4618 on Flickr (Creative Commons license)

Thanks to R.A. Dickey's recent success, we're adding a 3rd type of knuckleball to Baseball Mogul.

The game already had a "Knuckleball" and "Slow Knuckler" as options (Wakefield throws both of these). We also of course have a knuckle curve, but that isn't actually a knuckleball. With Dickey having a Cy Young season throwing what some have called a "Power Knuckleball", we're taking a close look at the physics behind this pitch and adding it to the game.

Baseball Contract Analysis

We've been going through contract and salary data, in order to improve the artificial intelligence in Baseball Mogul 2013 and future versions.

Contract Lengths for Major League Baseball Players (n = 516)

We examined the 516 players on Major League 40-Man Rosters (as of opening day) that had reached either arbitration or free agency. The first fact worth noting is that the vast majority (84%) of all player contracts are between 1 and 3 years. The longer deals get all the headlines, but the shorter deals dominate team rosters.

Baseball Mogul's Simulation Engine

If you're a fan of the Baseball Mogul series, you might be wondering why, after 17 years, I chose this year to fully rewrite the simulation engine.

Well, first, some history:

Pythagoras Explained

Pythagoras of Samos, mathematician and philosopher, died about 2500 years ago. Nevertheless, his name is familiar to baseball fans. The "Pythagorean Expectation", invented by Bill James in the 1980s, predicts a team's winning percentage from runs scored and runs allowed. Despite the intimidating name, Pythagorean win expectations can now be found on mainstream sites like ESPN and MLB.com.

Baseball Mogul 2013 - Version 15.07

We have just released Version 15.07 of Baseball Mogul 2013. This is an update to Version 15.00 (which was launched on April 3rd). Simply put, if you were waiting to make sure the bugs were worked out of the initial release, this is the version to buy:

1. Fixes a crash bug that was occurring when viewing Scouting Reports for certain players.
2. Fixes a bug where a called strike sometimes didn't count on hit-and-run (or run-and-hit) plays.
3. Fixes calculation of season and career WAR ("Wins Above Replacement").
4. Includes year-by-year lefty-righty stats for every player going back to 1950 (alas, Retrosheet doesn't have complete data before that).
5. Improves manager AI and pitcher usage patterns.

If you already bought Baseball Mogul 2013, download and run this file to automatically upgrade to Version 15.07.

Happy simming!

Clay

Win Expectancy (Baseball Mogul)

Another new feature for Baseball Mogul 2013 is that all the game logs now show Win Expectancy after each play.

Win Expectancy is the chance that a team has of winning the game, based on the inning, score, outs and runners on base. In the case of the Baseball Mogul game recaps, the Win Expectancy for the home team is shown after each play.

Here's an example from the 9th inning of a game I just played in Play-By-Play mode:

At the top of this screen shot, the home team (Washington) is leading the game 4-3, at the start of the 9th inning. In this situation in the current run-scoring environment, this gives the home team an 87.2% chance to win the game.

Then the Washington pitcher walks the leadoff man (Sanchez), dropping their win chance by 9.6% (to 77.6%).

When John Buck doubles down the right field line, the game is tied and the Marlins have the go-ahead run on second base. The biggest play of the inning is when Jose Reyes singles with two outs to knocks in the (eventual) winning run. By the end of the inning, the Marlins are ahead 5-4 and the Nationals chance of winning the game has dropped to 17.4%.

In the bottom of the 9th, Washington's chances just get worse. The leadoff hitter (Marrero) strikes out, dropping their win chance below 10%.

At the end of the game, their chance to win has dropped to 0.0%. (This is self-evident, but it shows that the math adds up).

Strike Zone Heat Maps (Baseball Mogul)

As I mentioned here, this is the year that Baseball Mogul went from being a random number generator (like Strat-O-Matic or previous computer sports games) to becoming a true physics-based simulation of each pitch. However, it's difficult to determine if our model is realistic without having a way to visualize the results. So, I added heat maps to Baseball Mogul 2013, plus the functionality to record data for every single pitch.

Ideally, we should see the same patterns in our data that we see in Major League Baseball. For example, we know from actual pitch data that BABIP (Batting Average on Balls In Play) varies by pitch location:

Heat map by Dave Allen

This heat map is for right-handed batters only, with a view from behind the catcher. Imagine the batter standing to the left of the strike zone.

As you can see, BABIP is slightly higher in the lower part of the strike zone, because low pitches lead to more ground balls, and ground balls turn into hits a bit more often than fly balls.

And, the "hot zone" for BABIP tilts. That is, on low pitches, the highest BABIP occurs just inside the middle of the plate. On higher pitches, this region is a bit farther away from the batter.

This prime hitting zone tilts at this angle because the sweet spot of the bat passes through the strike zone roughly along this line. With the bat (and arms) pivoting on the batter's shoulder, it's easier to reach the high outside corner with the heart of the bat than it is to reach the high inside corner. So, with Baseball Mogul, I modeled the bat moving through the strike zone, by assigning different coeffecients of restitution to each point on the bat. The coefficient of restitution (COR) reflects how quickly the ball comes off the bat at each spot. It can be represented as a parabolic function along the length of the hitting surface, with the COR peaking at the bat's sweet spot.

I hope I didn't lose you with the technical jargon. Simply put, once this COR model is added to the simulation, we should see results that look somewhat like real life. So here's the same heat map from one season of Baseball Mogul:

Heat map by Baseball Mogul 2013

Wow.

I honestly didn't expect it would look that good. As with the first heat map, BABIP is highest from the middle of the plate down to the bottom edge. And, we see the same tilt: on low pitches, the batter is more likely to hit the ball solidly if it's over the inner half of the plate. As the pitch moves up in the strike zone, BABIP improves as the pitch moves toward the outer half.

Very cool. I'm really looking forward to using these heat maps to continue to dig into the new simulation engine.

Pitch Tracking and the "Charts" Tab (Baseball Mogul)

Baseball Mogul 2013 now tracks and records every single pitch of your simulation. Pitch type, speed, location, result. etc.

In conjunction with this, we have now added a 'Charts' tab to every player's Scouting Report:

The above chart was created by simulating from the 2012 season through 2025. It shows BABIP (Batting Average On Balls In Play) for all lefties that Strasborg faced.

You can also limit the search to certain pitch types (fastballs, sliders, etc.) and situations (2 outs, RISP, "close and late"). For example, here's Jason Heyward's Slugging Percentage on just fastballs:

Finally, you can limit the search to at-bats against one team or one specific player, such as Dustin Pedroia versus the Yankees from 2012 through 2020:

Baseball Mogul 2013: Under The Hood, Part 2

For years, people have been asking for direct access to the Baseball Mogul database. Many of them want to be able to build mods that read and parse the data.

For example, Mizerak created a very cool app called the Box Score Parser that sifted through multiple seasons of box scores in order to derive data that can be easily viewed inside Baseball Mogul. Unfortunately, he had to use box scores because they were one of the only things saved in a human-readable form.

Baseball Mogul 2013: Under The Hood, Part 1

When I first wrote Baseball Mogul, it simulated each game by simulating the result of each plate appearance. This isn't unusual. This is how Strat-O-Matic works. This is how other computer games work. This is even the method I used when writing my very first baseball simulation, using paper and dice, back in 1976.

But the thing is, baseball isn't played with paper and dice. It's played inside televisions. And the game on television isn't determined by comparing player stats and generating a random number. It's determined pitch-by-pitch. Each pitch has a velocity, a spin direction (and magnitude), and the location where it crosses the plate.

Photo by Wall Street Journal

So, for Baseball Mogul 2013, I rewrote the entire simulation engine to calculate:

1. The velocity and path of each pitch (similar to that recorded by PITCHf/x).
2. The timing and velocity of the bat swing.
3. The plane of the bat swing (and the location of its sweet spot).
4. The angle and velocity of the hit that results from the above.

Cliff Lee For MVP?

Photo by aturkus on flickr

Cliff Lee is now 14-2 for an Indians team that has won only 45 games. Lee has over 31% of his team's wins.

The last pitcher to finish the season with more than 30% of his team's wins was Fergie Jenkins, winning 25 games for the 1974 Texas Rangers. And Jenkins pitched in a 4-man rotation, allowing him to get 41 starts.

In other words, no pitcher has ever done what Cliff Lee is doing: winning 30% of his team's games while pitching in a 5-man rotation.

Over-Active Game Design

Photo by Bari D

It's not surprising that everybody wants to be a game designer. What's bad is that everyone thinks they can be a good game designer.

Probably the most telling aspect of bad game design is "over-active" game design: the desire to "improve" games by adding complexity.

The media response to tonight's All Star Game is a perfect example of over-active game design. Some suggestions I've heard on T.V. and radio:

Varitek Isn't The Weak Link

I don't live in Boston anymore. But I still live in Red Sox Nation. This means that the radio next to my desk gets WEEI. The hot topic this week is how Jason Varitek's offensive drought is hurting the team.

Tek is definitely worrisome. However, his OPS of .672 is only 118 points below his career average.

How You Can Fix The Wild Card!

It's been almost ten years since we came up with the right way to fix the current wild card system.

Will The Real Guillermo Mota Please Stand Up?

By now we all know that Mike Lowell and Josh Beckett are headed to the Red Sox in exchange for Hanley Ramirez, Anibal Sanchez, Jesus Delgado and Harvey Garcia. But it was just announced the Red Sox insisted that Guillermo Mota be added to the deal, in exchange for even more players from Boston's farm system.

Mota is an interesting commodity. In 2003, and part of 2004, all with the Dodgers, he had an ERA near 2.00. Since being traded to the Marlines in mid-2004, his ERA has ballooned to 4.74.

So the question is, who is the real Guillermo Mota? The dominating setup man that helped the 2004 Dodgers win the N.L. West? Or the struggling reliever who was relegated to a reduced workload in the 2005 Marlins bullpen.

The Maddux Gambit

In chess, a gambit is the sacrifice of material (such as a pawn) in order to gain another type of advantage in the game (such as a positional advantage). This strategy can also be used in sports such as football or baseball, where a player or coach may sacrifice something on one play (sometimes even allowing one or more points or runs to score) in order to gain another kind of advantage.