January 20, 2016

Pebble Hunting

Flame of Thrown

by Sam Miller

In the fourth game of the 2005 World Series, viewers who tuned in saw a matchup of, in Joe Buck’s words, “two guys who throw hard.” Those would be Brandon Backe, whose fastball would register 90 to 92 for most of the game, and Freddy Garcia, who would get his up to 93 on a few occasions. Both pitchers were relieved for relievers in the eighth: Cliff Politte sat at 91-92 for the White Sox, and he was followed by Neal Cotts at 92-94, and he was followed by Bobby Jenks:

That pitch was 95. You can tell it was 95 because of the number they showed on the screen: 95. Ninety-five means it was 95. In case that isn’t clear.

Brad Lidge relieved Backe. He threw something harder than 95: He threw 96.

You can tell it’s 96 because of the flame. Flame means it was 96. Ninety-six gets a flame, because 96 will burn a hole through the bat, because 96 isn’t allowed in airplane lavatories, because 96 can break down the muscle fibers in meat. Ninety-six hooboy.

That was 2005, and so far as I can tell it was the first World Series with the flame.

Baseball has changed since then, and broadcasts have changed with it. For instance,

AOL is not where to go for the best sports coverage anymore, and so broadcasts no longer tell you to go there for it. What, then, of the flame? Has the flame evolved with the times? This is an article about the flame.

The Flame
The flame is an object. It is to us to first choose the appropriate way of defining it.

The flame is not matter. It was not born of a mother and it will not pass on its genetic material. It is information designed for a purpose, and that purpose is in a general sense to affect our experience of watching a baseball game. More directly, its purpose appears to be informative, rather than entertainment. If it was intended for the latter, then instead of the same flame over and over, it would produce, say, yo mama jokes or new episodes of Deadwood. It is binary information: The pitch was very fast or not very fast.

But it also exists only in the context of a broadcast of a baseball game, which is itself less about information than entertainment. (There are far more efficient ways to convey the information of a baseball game.) The reason the flame isn’t a new episode of Deadwood is presumably that those would detract from the entertainment experience of the overall baseball game production; the tail would be wagging the dog. In this sense, the informative value of the flame is part of the larger, entertainment value of the game, and is thus actually entertainment.

Of course, that baseball game exists only in the context of baseball players, owners, broadcasters, lawyers, etc., who are not, primarily, trying to entertain. They are trying to make money, which they can use to buy food and medicine so they can live long and attract fertile mates, thus giving them ample opportunities to pass on their genetic material. The entertainment of the baseball game—and, within it, the informative value of the flame—are tools by which all of the people involved in the game increase their lifespans and likelihood of finding mates. So, depending on the level you wish to observe, the flame is an object whose primary purpose is to inform, or to entertain, or to procreate.

The further one gets from the flame itself—the informative purpose—the less the flame matters as anything but economic vehicle, and the easier it is to throw our hands up and cynically assume we are being taken advantage of. So let’s stay strictly, for the bulk of this text, in the most specifically flame-relevant space: The informative one. We shall move forward assuming that the point of the flame is to provide us truthful information.

The Flame’s Information
In 2005, when Brad Lidge threw that pitch, the flame informed us that the pitch traveled 96 mph. This is not very fast—for a plane. It is extremely fast—for a rowboat. Were we to view the jet plane championships on Fox, we wouldn’t expect a flame to appear whenever a jet plane cracked 96, and were we to view the rowboat finals on Fox we would expect the flame to appear at a much lower threshold. So when the flame appears at 96 mph on a baseball broadcast, we know that it is not the number itself that is significant, but the number in the context of the specific act in question. The flame has decided that 96, but not 95, is a significant achievement for a baseball pitched, by a human, in competition, on Oct. 26, 2005.

The most likely factor in this determination is scarcity. I don’t mean to suggest that a chyron producer calculated exactly the speed that represented, say, the 95^th percentile of fastball velocities, but that’s presumably what they were going for here. Maybe they didn’t get it right, because, again, there probably wasn’t a ton of work put into finding the right mark, but if you want to star the outliers in any particular dataset, you’ll pick a spot near the edges.

That spot appears to have been 96 from the beginning. So far as I can tell, there was no flame in 2004 or before. Here’s 96 in 2003; no flame. Here’s 97 in 2001; no flame. The flame was probably born in 2005 at 96.

The flame was born two years before PITCHf/x, and three years before full coverage of PITCHf/x, so we can’t say how frequently pitches topped 96 at that point. We can say how frequently—as a percentage of all pitches—they topped 96 in 2008, and in the years since, and surprise! The arrow goes up:

To match the 4.1 percent frequency of 2008 baseball, the 2015 flame would have to be set at 96.998 mph. To match the frequency of 2005 baseball—well, that’s trickier. I think we probably all assume that there were fewer 96+ fastballs thrown in 2005 than in 2008, but how can we be sure? The relative flatness of the line from 2008 to 2012 makes it hard to extrapolate with confidence to the years before 2008. But Baseball Info Solutions’ measured pitch velocities, as published at FanGraphs, show a steady rise in pitch velocity from 2002 (at 89.8 mph) to today (92.2 mph), including rises from 2005 (90.0 mph) to 2008 (90.9). Anecdotally, I can tell you that a sizable difference between 2005 and 2008 seems obvious; the reason I can’t say with certainty that 2004 didn’t feature the flame, for instance, is that I watched multiple 2004 broadcasts without seeing a single pitch over 96 mph. Taken together, we can confidently say that 4.1 percent frequency—the 2008 flame threshold—is probably higher than the founding flamers intended, if we’re going to get all originalist on this question. But to be conservative, and to stay consistent with our data, we’ll treat 2008 as Year 1, and 4.1 percent as the intended frequency.

Here's where the flame would have to be set for each year since then to maintain 4.1 percent:

• 2008: 96.0
• 2009: 96.15
• 2010: 96.18
• 2011: 96.07
• 2012: 96.17
• 2013: 96.35
• 2014: 96.505
• 2015: 96.998

This suggests that, if 96 was a good marker in 2008, then it stayed a good marker until 2014, when by frequency the flame should have been bumped up to 97. So now let’s see how the progression actually went:

2008: 96 mph flames. (In all cases, pitchers that came in one tick lower—95 in this case—were confirmed to have no flame.)

2010: 96 mph flames (or, more exactly, “red” velo reading, as compared to the standard yellow that was used that year).

2011: 95 mph flames.

Wait—95? It’s moving backward? It’s moving backward.

2012: 95 mph flames.

2015: 95 mph flames.

In 2008, 4.1 percent of all pitches got a flame. By 2015, more than 13 percent did. It’s runaway flame inflation.

Which Leaves Only Four Possibilities

1. Just nobody in production is paying attention. But this seems to be contradicted by the fact that, in 2011, they proactively made a change—albeit in a surprising direction. The fact is, screen space is extremely limited and quite valuable; that’s why we have the flame in the first place, so that more information can be included without expanding the amount of real estate required. Just look at the GIFs from 2005 and 2015 and see how much attention has been paid to producing the corners of the screen: They've added four crucial pieces of information (hitter, pitcher, pitch count, batting average) while making the graphic less obtrusive. Somebody pays a lot of attention to that space.

2. They were “wrong” the first time. There is, of course, no such thing as “wrong,” just as there is no such thing as “fast” other than relative to the norm. But perhaps it was decided that 4 percent was just an unacceptably infrequent use of the pre-paid flame service, especially because they would tend to arrive in clods, e.g. 15 flames when the closer came in in the ninth but none in innings 1-7. If that’s the case, then we could assume that the “adjusted” standards of 2011 are closer to the ideal, perhaps the very exact ideal: 8.3 percent of all pitches. Which still makes 2015's flames way too frequent.

3. It’s not about frequency at all! Rather, it’s about the effect on the batter—a flame representing a certain level of difficulty to the pitch, no matter whether it happens once a game or 100 times. If this is the case, then the 2005 levels—or, perhaps, the 2011 levels—might be just as appropriate today, assuming that batters aren’t getting better at hitting high-velocity fastballs. Are they?

Let’s use three measures: Balls in play per “strike,” where strikes include called strikes, fouled pitched, swinging strikes and balls put in play; balls in play per contact, or, put another way: BIP/(BIP+fouls); and slugging percentage on balls put in play. We can’t simply look at how hitters did on all pitches 95+ (or 96+), because a much higher percentage of the 95+ pitches in 2005 were close to 95 (rather than 98, or 100, or 103) than in 2015. So we’ll limit it to pitches that are exactly 95-95.999 mph.

2008-2009

• 28.3 percent of strikes put in play;
• 46.0 percent of contact put in play;
• .500 SLG on BIP

2015

• 27.4 percent of strikes put in play;
• 46.0 percent of contact put in play;
• .513 SLG on BIP

All Years

• 28.0 percent of strikes put in play;
• 46.3 percent of contact put in play;
• .523 SLG on BIP

This is, I have to say, pretty surprising to me: Hitters aren’t getting better at this. I figured they would be. You get used to it. You look for it. You’ve been raised around it. You adjust to it. You force the pitcher to adjust back. But, in fact, 95 mph is no easier for 2015 hitters to hit than it was for 2005 (or, at least, 2008) hitters. Pitchers might not ever have to adjust again, if they can just keep throwing more and more pitches like these pitches.

For the purposes of our chyron producer, this might justify the use of the flame for 95 no matter how common it is or becomes. What is fire? In one sense, fire is defined relative to things that are not fire. Fire is hotter than a tennis ball. It is brighter than a bag of hair. It is less dense than a Kit Kat. So in a world with tennis balls, bags of hair and Kit Kats, you could define a fire by those features. This is the premise of flame as signifier of rare velocity. But on, say, the sun, where fire and brightness are everything and everywhere, fire would not be interesting; it would not be a useful signifier at all.

But fire can also be defined by what it does. Fire consumes matter. Fire produces smoke. Fire burns my fingers. It is its effect. So whether fire is surrounded by fire or not, it remains that same consumptive, destructive thing that it is. And if 95 mph is just as effective in a world of 95s as it is in a world of 92s, then it arguably merits a flame regardless of how common it is.

4. In fact, we have been wrong in presupposing that the flame is primarily information. Perhaps it is entertainment, and somebody has determined that we, the viewer, are more entertained when we see a flame. This is, after all, the network that gave us When Buildings Collapse. Maybe only a little entertained, maybe only subconsciously so, but entertained. More likely to keep watching, and more likely to watch again. Which means the procreative television producer wants to put as many flames in front of you as he or she can without you noticing that you are being played.

Of these, only reason no. 3 justifies a continuation of the 95 mph flame. I guess, then, each of us has to decide whether we're a Reason No. 3 person or not. If we think that super-fast fastballs are most interesting for their rarity, we are seeing wanton abuse of the feature. If we think they're interesting for the effect on the game itself, on the outcome of the batter/pitcher matchup, then the flame has (perhaps accidentally) gotten it exactly right. Perhaps someday there will be a flame on 80 percent of pitches, and perhaps every game will go scoreless into extra innings. In which case we will debate whether it represents a sport finding its level or a sport desperately in need of the heavy hand of government to save it.

I don't know. I answered the question you had about whether the flame graphic has kept up with increasing pitcher velocity. I failed, yet again, to answer the question of what it's all about.

Thanks to Rob McQuown and Harry Pavlidis for research assistance.

Sam Miller is an author of Baseball Prospectus. Follow @SamMillerBB
Click here to see Sam's other articles. You can contact Sam by clicking here