A Story on Mother’s Day

“We love our parents because they loved us first.” – Gary Vaynerchuk

At my grandmother’s funeral a woman I had never met told me a story:

One Sunday about ten years ago, I went to St. Paul’s Church for the first time. After mass a woman came up to me and said: “I don’t recognize you, are you new?”

She said she was, had just moved to town. This was her first Sunday at her new home and wanted to pick a church to attend, but she was not sure which one to go to. She could not ask any of her friends, because she had none here.

My grandmother replied, “My name is Helen, so now you do now.”

On Suicide, the Meaning of Reality, and Hanging Curveballs

“There is but one truly serious philosophical problem and that is suicide.”

– Albert Camus

“Achieving the end of the exercise was never the point of the exercise to begin with.”

– Adam Savage

“The journey is what brings us happiness, not the destination.”

– Peaceful Warrior

The philosopher Albert Camus said that the only real question that we have to ponder is whether or not we should kill ourselves. He came to the conclusion that the answer should be no, but that isn’t to say we should not ask the question.

If you are reading this I assume that you, like Camus, have not chosen suicide. And so other—if less important—questions arise. One of these is What is luck? or more specifically: There is randomness everywhere. As far as we can tell, nothing is 100% guaranteed to happen. What then determines whether one event should or should not take place?

More than in any other arena (no pun intended) randomness is exhibited in sports. Randomness is a batter hitting a curveball for a line drive that just so happens to be right at the shortstop who catches it for an out. It is random that the pitcher threw a curveball rather than a fastball. It is random that the batter swung at the pitch rather than not swinging, that he hit the pitch rather than missing it, that hitting the ball resulted in a line drive rather than a fly ball, that the shortstop happened to be standing in precisely the right spot, and that he did not drop the ball. So many factors could change the entire situation.

Do we, as observers, have the ability to collect enough information surrounding this situation that we can predict, in advance, the outcome of each pitch? Certainly at the current juncture we do not. And it appears doubtful (though not 100% guaranteed) that we ever will reach that point, even if we can factor in variables such as what the hitter had eaten for breakfast that morning.

Presently, all we can say is that based on observation over the last decade, there is about a 70% chance that a batted ball will result in an out. This, of course, means there is a 30% chance the ball will find grass and result in a hit. The shortstop who knows that the batter tends to hit balls up the middle, and thus positions himself in its likely path, might tilt the odds. But tilt them as he may they will never reach 100% one way or the other. So why does the ball do what it does? Luck. Chance. Randomness. A weighted cosmic coin flip. These are all, of course, the same thing.

---

In the spring each year, Baseball Prospectus takes every stat there is about baseball and puts them into a simulation of the upcoming season. Then they run it a bunch of times. The results are a series of percentages that each team has to win their division, make the playoffs, and win the World Series. What this actually means is that if the season were played 100 times, assuming everything to be identical at the start—the same players with the same talent on every team—the outcome would always be unique from every other time the season was played. Due to the large amount of randomness involved in the sport, every simulated season will be different than every other simulated season.

There has been a statistical analysis explosion over the last few decades. New stats are developed and analyzed. New studies are conducted daily. Each of these is a mini-simulation. An attempt to learn which players have contributed the most to their teams and which ones will continue to do so in the future. As more is discovered about which stats are good indicators of future performance, the simulations will be adjusted and become more accurate. The end-goal, it seems, would be to build a computer simulation capable of predicting, with perfect accuracy, the outcome of every pitch.

---

But, of course, we find that this has a catch. For if we build a perfect computer simulation and watch it play through a season like some video game that is indistinguishable from reality. It will tell us, in advance, the result of every pitch. Why then should we continue to play the game? For we have become the baseball gods and already know the outcome.

That we would ever come to this seems incredibly unlikely, for the outcome of a series of random events is unpredictable by definition. If we could simulate randomness, it would cease to be randomness. So then, what is reality, if not another simulation? Carried out to its furthest potential, we could not only simulate the events of a baseball field, but every atom of the universe. This would make us not only baseball gods, but gods in a much truer sense of the word. For who is to say our simulations are less meaningful than our reality?

What would occur if we found our simulations could predict our reality? Knowing, in advance, everything that will happen to us would strip all enjoyment from our lives. Unless we come to the same conclusion that Camus came to. That even while we know we will never be able to know if there is inherent meaning behind our existence, we should embrace our existence regardless and carry on with a passion for life.

We should embrace life (and baseball) knowing how marvelous it is to not know. And yet we should never stop studying anything, baseball statistics included. Never stop trying to become gods. For it is the pursuit, the season, the journey, the climb up the mountain that is the meaning of our lives.

The Impossible Double Grand Slam

They say no matter how many games you have watched, you have never seen it all in baseball. This play last week was proof of that. There is virtually no way to discover if this has ever happened before either, as it would not be recorded in the play-by-play data.

This does remind us of the odd rule—or lack of a rule—concerning two runners occupying one base. Which is that the runner is not out until he is tagged.

Therefore if the situation were to arise that the defense refused to tag out the trailing runner, and instead pitch to the next batter, both runners would legally be occupying the same base. Unless the umpires stepped in, and we’ll assume for fun that they do not, all nine players on the team could theoretically be on base simultaneously.

Or just shy of that, eight runners could end up on base and the ninth could hit a 9-run home run. Unfortunately we’ll never get to see that happen… just like we will never see anyone run backwards from second to first base.

Head-To-Head Fantasy: You’re Never Good, Only Lucky

We’re two weeks into the baseball season, which means most teams have played about 13 games. Nobody would ever judge a baseball player’s season on such a small sample size—but 13 games is how long most fantasy football regular seasons are.

Because of this, it requires a great deal of luck to be successful in your fantasy football league. Most leagues feature weekly head-to-head matchups. So my team does not have to perform well—it just has to be better than whoever I happen to be playing. My team could have its worst week and win, or its best week and lose. Whoever I happen to be playing each week and how they happen to do is simply determined by luck.

Fantasy baseball though, with its long season requires much more skill—doesn’t it?

You may recall the Law of Large Numbers from statistics class (or from this Wikipedia article). This Law states that if you are looking to find the average of whatever, the more times you can run the experiment, the better. For example, if you want to find out the batting average for the league, collecting data from every at-bat over an entire season will get you a much more accurate picture than if you calculate the batting average of players during one random inning.

This is why nobody labels a baseball player as being good or bad after just 13 games.

So you would think that despite the luck-filled head-to-head matchups in fantasy leagues, the team with the best players will eventually come out on top at the end. Matthew Berry did an experiment with this very concept after last year’s baseball season.

Berry allowed readers to create teams with any players they wanted, based on the stats from 2012—which had already been completed. Already knowing every player’s stats, readers easily picked the best players at each position. Then without repeating any players, Berry was able to create teams to beat them all.

How? This can happen with any scoring system, but I’m going to use a points system for simplicity’s sake. In this league each team has three players and Team A plays Team B three times. A player with a great week will be over 80 points, whereas average is somewhere in the 40s.

In Week 1, Players A, B, and C all have fantastic weeks and Team A destroys Team B, which has three players who were merely average. During Week 2, all of the players have average weeks, but Team B is able to squeak out a win by a few points. The teams are now both 1-1. In the final week, Players E and F have above average weeks, whereas Players A, B, C, and D are all average once again. Team B wins for the second straight week.

Team-wise Team B is 2-1 and ahead of Team A in the standings. However, if you take a look at individual performance over the season, all three players on Team A are superior to all three players on Team B. Even more surprising is that Team A has far more overall points than Team B, even though they are 1-2.

Very few fantasy teams will consist of just three players, but despite an infinite number of team combinations and scoring systems, this basic pattern of teams with better players losing more games can and likely will manifest itself in almost every league.

Who cares though, what does this mean? Having the best record in a head-to-head league is all that counts as far as winning the league goes, and it is entirely luck. Because even if you are smarter than everyone else in your league, have the most accurate projections, and construct the best team, as we have just seen you can still lose.

What you want is a team that is good week in and week out, not great one week and suck the next. The when they are good is just as meaningful as how good they are. But while many projection systems can predict in the ballpark (pun intended) how well a player will perform, nobody can predict when they will do well or poorly.

And this makes it very similar to real life: The goal is not to be the best team, the goal is to win the championship. And the best team very rarely wins the championship.

A Beginner’s Guide To DIPS: Part II – FIP

In looking at BABIP, we divided at-bats into two categories: times when the ball was put into play and times when it was not. Whether balls in play will result as a hit or an out is largely out of the hands of the pitcher. With this in mind, multiple stats have been developed that focus on at-bats when the ball is not hit into play.

The most well-known of these stats is Fielding Independent Pitching (FIP). Which looks only at walks, batters hit by a pitch, strikeouts, home runs allowed, and innings pitched—in other words, things that are much more within the pitcher’s control than balls hit into play.

FIP = ((3(BB + HBP) + 13HR – 2K) / IP) + constant*

As described by Fangraphs, FIP is used to “to calculate what a player’s ERA should have been over a given time period.” By ignoring balls in play it is assumed that factors like defense is equal for all pitchers.

*The extra number added on at the end, which is typically about 3.1, puts FIP on approximately the same scale as ERA, which makes it easier to recognize a good one from a bad one. It changes every year based on a number of factors.

By ignoring so many at-bats one would think FIP only gives us part of the picture of a pitcher, whereas ERA or WHIP take into account all batters a pitcher faces. However, what we find is that FIP is actually a better indicator of the degree of success a pitcher has in subsequent seasons than ERA. This is likely due to walk- and strikeout-rates being two of the most consistent stats for pitchers from year-to-year.

xFIP

A strikeout or a walk in one stadium is exactly the same in any other, but a home run at Fenway Park may not be one at Safeco Field. This makes home runs a bit tricky to deal with when we are trying to limit the number of factors other than the pitcher himself in FIP.

On average 10% of fly balls will end up as home runs. Some pitchers may end up finishing the season with 8% of their fly balls having left the yard, while others will be unlucky and have 14% of theirs flying over the fence.

In order to eliminate this element of luck, we have a variation of FIP called xFIP. Rather than using the number of home runs a pitcher allows, xFIP assumes every pitcher has the average fly ball to home run ratio.

FIP = ((3(BB + HBP) + 13(Flyballs * League-average HR/FB rate) – 2K) / IP) + constant

Though the difference between FIP and xFIP will not necessarily be a lot, we find that xFIP is more predictive of subsequent performance than FIP.

A Beginner’s Guide To DIPS: Part I – BABIP

When a batter steps up to the plate two things can happen: He will hit the ball or he will not hit the ball. If he does hit the ball it will lead to something like a ground out, fly out, or base hit. If he does not hit the ball that means he has been hit by the pitch, walked, struck out, you get the idea.

We know that the best pitchers—the Roy Halladays and Randy Johnsons of the world—strike out a lot of hitters and do not walk a lot of hitters. Weaker pitchers will do just the opposite. The Mike Mohlers and Tyler Greens (sorry guys!) walk almost as many guys as they strike out.

But when the ball is hit, we also know that if the ball was thrown by Randy Johnson it is less likely result in a base hit than if it was thrown by Tyler Green. Good pitchers are better than weak pitchers regardless of whether the ball is hit or not; this has been common knowledge for a hundred years. The only problem is: It is not true.

Most people will reject this upon hearing it the first time, I was skeptical too. So let’s bring statistics into the discussion to settle this. The numbers do not lie.

First, Randy Johnson struck out 28% of the batters he faced. Tyler Green struck out 15%. We were both correct in that observation and by all accounts Johnson is the better pitcher, but that is not in question. The question is: If a batter is to hit a ball thrown by Randy Johnson is it more likely to be an out than when that batter hits a ball thrown by Tyler Green?

Common stats like batting average use all at-bats, regardless of the outcome. But we need only to consider part of at-bats, the ones in which balls are hit into play. This stat is called Batting Average on Balls In Play (BABIP). Because fielders have no opportunity to turn them into outs, home runs are not considered balls hit into play for BABIP.

Our hypothesis, based on it having been ‘common knowledge for the past hundred years’ is that hitters who faced Johnson would have a much lower BABIP than those who faced Green.

Thanks to Fangraphs, which provides every pitchers’ BABIP for every season, we can test this out pretty easily. In 1995, Johnson pitched 214 innings. He led the league with a 2.48 ERA and won the Cy Young Award. The hitters who he faced had a BABIP of .301. In other words, 30% of the guys who hit a pitch that Johnson threw reached base safely. The other 70% resulted in groundouts or flyouts.

That same season, Green threw 140 innings and had an ERA of 5.31, which was good for 19^th best on the Phillies. The hitters who he faced had a BABIP of .313. That means 31% of the guys who hit a ball thrown by Green reached base safely. Greene’s BABIP differs by just 12 guys out of 1000, so his BABIP is virtually the same as Johnson’s.

Obviously this is a fluke, right? I just picked two season I knew would work… Let’s try it again.

In 1998, Johnson split time between the Mariners and the Diamondbacks, but his season-long ERA of 3.28 would have ranked in the top 10 in either league. Hitters who faced Johnson in 1998 had a BABIP of .320.

Green’s 1998 ERA was not fantastic at 5.03. This would be the last season that Green would pitch in the Majors, but hitters who faced Green in 1998 had a BABIP of .254. Almost 7% lower than Johnson’s BABIP.

You are welcome to look at the numbers yourself, but this example is no fluke. Regardless of whether they are a ‘good’ or ‘poor’ pitcher, the typical pitcher BABIP is between .290 and .300. As we have seen with Green and Johnson most seasons will fall outside of that range and there is little relation from one season to the next. Over a career, though, things tend to average out.

Johnson played for over 20 years and had seasons where his BABIP ranged from .247 (in 1990) all the way up to .348 (in 2003). Based on those numbers it appears that it increased over his career, but the season after his highest season BABIP in 2003, it fell back down to .264 (the second-lowest of his career) the following summer. Johnson’s career BABIP ended up at .291. Meanwhile Green pitched for just four seasons in the 1990s and his career BABIP came out to .289. Again, virtually identical to Johnson’s.

Since this was first discovered by a fellow named Voros McCracken in the late 1990s, people have been just as surprised as you probably are. And they have been trying to disprove it ever since. While it has not been disproven, numerous studies have shown that pitchers do have some effect on balls hit into play—albeit much less than what we had thought in our original hypothesis.

Numerous factors play a role in whether or not a ball in play will fall in for a hit. The quality of the fielders, the size of the park (remember, home runs don’t count as balls in play), and the pitcher (groundball pitchers do tend to have lower BABIPs).

The biggest factor, which tends to be frustrating for a lot of people, continues to be luck. Sometimes a batter will hit a line drive—absolutely drill the ball—right into the glove of a shortstop. Other times the shortstop is two feet to the left and the ball flies into the gap for a double. The difference between those two scenarios is luck (possibly it is good positioning by the shortstop, but more likely luck).

This should, in theory, extend to everyone. If you or I walked out on the mound of a Major League game and started lobbing the ball in there, the balls hit into play would not result in hits 95% of the time like we might assume. They would likely be higher than .300, but perhaps not much higher. In fact, this is demonstrated every year during the home run derby. If you were to put a normal defense behind the home run derby pitcher—who wants the batter to kill the ball—the batter would have a good batting average, but he would still be putout plenty of times.

See also - Part II: Using what we know about BABIP to come up with a better pitching stat.

Pirates Prediction: 2013

Last year I made a 100% guess of a prediction that the Pirates would win 80 games. They won 79.

As part of that prediction though, I also said the Bucs would exceed the elusive .500 mark this year. I have learned a decent amount about statistics over the past year though, and all that new knowledge is pushing me to retract that prediction. But I won’t.

Because as accurate as predictions have become over the last few decades, there is virtually complete agreement that luck still plays a role in all this. Which is why I predict the Pirates will finish the 2013 season with a record of 82-80. Not because of improved players or new additions, but because they will luck into it.

The final part of my pre-2012 prediction was that the Pirates would make the playoffs next year in 2014… I don’t know what I was thinking when I wrote that, but it is going to take a lot of luck to pull that off. Get your green weenies out now.