Why aren't there more 20 strikeout games?

Yesterday Beyond the Box Score's own Neil Weinberg examined this very question as he looked into why we haven't seen someone make a run at the 20 K mark in recent seasons. This is especially relevant with the rise in strikeouts, something Neil touches on much more in depth in his article which is definitely a must read. Neil summarized the current situation thusly:

Strikeouts are up, but we don't have a 20 strikeout game in this super high strikeout era. We don't even have an 18 or 19 strikeout game in a super high strikeout era. I don't havethe right answer here, but there are a few related factors that might be playing a role.

Neil Weinberg - Beyond the Box Score

Neil's article made Tom Tango's wheels start turning and he offered up some follow up research for anyone interested in this topic. Mr. Tango brought up an additional point that is worth looking into further:

If let’s say there are fewer 16+ K per game (at the team level), then one reason might be that it’s easier to get an out via K!  For example, let’s say a pitcher today gets 0.25 K per PA, but in the other 0.75 PA, they get 0.63 outs per PA.  A pitcher might get hot on one side, say bumps it up to .55 K per PA and cold on the other side (of the other 0.45 PA, they average 0.5 outs per PA).  That gives us a pitcher with 0.55 + .45*.5 = .775 outs per PA.  To get 27 outs, he’d have to face 35 batters, of which 19 are K.

But, imagine a pitcher that is hot on both sides, so, he gets his .55K per PA and in the other 0.45 PA, he averages 0.8 outs per PA.  That gives us a pitcher with .55 + .8*.45 = .91 outs per PA.  With 27 outs, that means he faced 30 batters and got 16 K.

It’s why for example a Nolan Ryan might be a better shot at getting the high K outing, because he performs badly enough in his non-K outcomes that he gets to face more batters.

Tom Tango - Tango Tiger Blog

This brings up a good point, but one that might not be easily understandable in its current format. The image I've created below lays out the scenario(s) Tom described with Player A being the baseline, Player A - HOT being the improved pitcher, and Player B being the pitcher mentioned in the second paragraph. I've also added the two most recent 17 strikeout games in the form of Brandon Morrow (on 8/8/2010) and Anibal Sanchez (on 4/26/2013):

The crux of Tom's point is this: pitchers that are less efficient at getting outs when they don't strike a batter out have a better chance of reaching a high strikeout total. That is, if a pitcher gets most of the batters out that he doesn't strike out, it gives him fewer opportunities to actually get up to 20 Ks as those precious outs are taken up by ground outs, fly outs, and any other kind of out you can think of.

This brings us to Morrow and Sanchez, the last two MLB pitchers to strike out 17 batters in a single game. Morrow faced 31 batters and allowed just 3 baserunners while striking out 17 over 9 innings. In facing those 31 total batters, he struck out 0.55 batters for every plate appearance (K/PA). In the other 0.45 plate appearances he averaged 0.71 outs per PA (Non-K/PA). This resulted in a very high 0.87 Outs/PA, and one of the best outings over the past 5 seasons. His 0.71 outs per PA though limited his ability to rack up strikeouts as he retired one out of every three batters he faced that he didn't strikeout.

Sanchez on the other hand only lasted 8 innings striking out 17 of the 29 batters he faced. That means he struck out on average 0.59 batters per PA (K/PA) while getting outs on other plate appearances at an average of 0.58 per plate appearance. That results in an Outs/PA of 0.83 which is worse than Morrow's number. With three fewer outs, his higher rate of outs by strikeout means that his performance was actually better from a maximizing strikeouts standpoint. This should seem obvious as he had the same number of Ks despite only throwing 8 innings. Basically Sanchez's higher K/PA and lower Non-Ks/PA figures highlight that he had a better shot at striking out the elusive 20 batters, and also racking up a high pitch count.

Using some reverse engineering from Sanchez's numbers we can estimate a nine inning performance for Sanchez. Had he not thrown 121 pitches through those 8 innings and stayed in the game, the numbers suggest he would have faced 33 batters, striking out 19 of them. Note that these figures are simply being pulled from his overall game stats and extrapolated out over 27 outs, just as we did for the hypothetical pitchers originally.

The other part of Tom Tango's post asked for some data which Scott Lindholm kindly looked up and offered to anyone interested. Luckily Scott did the hard part, and all I have to do is drop the table below:

To explain the data in the table I'll give some context. The left column is the time frame, broken into five year increments per Tom. Then the number of games from that five year sample. The third column is the number of 16+ strikeout games by a team combined. Then the percentage of games in that time frame that had 16+ strikeouts. The next two columns show the same data with one big exception: it's for individual pitchers only.

Overall the number of 16+ K games by teams has gone up, though there was a slight dip from 1974 - 1983. The timing is approximate, but generally the number of games where a team struck out 16 or more batters has gone up over time.

Tom Tango on the importance of this data:

If let’s say there are fewer 16+ K per game (at the team level), then one reason might be that it’s easier to get an out via K!

Tom Tango - Tango Tiger Blog

Over the past five seasons the number of 16+ strikeout games has gone up significantly while the number of 16+ strikeout games by individual pitchers dropped for the third straight five year period. This suggests to me that the potential causes outlined by Neil are all valid. That is, pitchers aren't going as deep into games thanks to pitch counts (see: Sanchez, Anibal), relievers are bringing more Ks to the party, and it's an extremely rare occurrence anyway.

It's possible that run on 16+ strikeout games from 1994 - 2003 was a convergence of strikeout rates increasing and awareness of the importance of pitch counts still being in its infancy. Then again, it's also possible that it's simply statistical noise, largely a function of increasing strikeout rates over time by teams as a whole. Based on the data above it doesn't seem that the hypothetical mentioned by Tom Tango comes into play here. Perhaps there is another reason that can be teased out of the data above. What are some other possible causes of the increased team 16+ strikeout rate, but decrease in individual performances of the same caliber?

Special thanks to Neil Weinberg and Tom Tango for the inspiration, and Scott Lindholm for the data that he graciously offered!

All statistics courtesy of Fangraphs, Brooks Baseball, Baseball Savant, and Baseball-Reference.

Jeff Long is a writer at Beyond The Box Score and Baltimore Sports and Life. You can follow him on Twitter at @BSLJeffLong.