I got into a "discussion" about my stance that Eric Walker's theory of the Sillyball era is correct. You can read the full theory here at his website, as well as the studies he has found which examined the composition of the ball. Seeing the complaints about it, I realized that while Walker's analysis made sense to me - and still do - there are some who don't see what I saw.
Walker's finding that got me believing him is this: starting roughly in the 1993-94 timespan, baseball suddenly started scoring more runs than it did in the 15 year period before. And not just a little, in 1994, it went beyond the max of before and stayed there for the next 15 or so years. Basically, the offense suddenly started scoring, on average, 12% more runs than they did before on average.
Walker also examined a number of angles on how that could happen and came to the conclusion that the only explanation that makes sense is that the baseball changed in the 1993-94 time period, that is, it got juiced, leading to the offensive era that most people call the steroid era. Hence why Walker calls it the "sillyball" era.
That "discussion" I got into took a look at why an offensive era might have popped up, and why it has been dying down, but there was no explanation given for how the offensive era suddenly sprung up, like a giant beanstalk, in roughly a year or two. Until I find a better explanation of this sudden jump, this "sillyball" is the theory I put my money behind.
Any theory involving the changing composition of player talent, or new player techniques (like new pitches or maple bats) or usage of PEDs forget one fundamental truth about humans and newness: there is an adoption curve to the increased usage. It don't happen over a 1-2 year period. It changes over time until the market is saturated with users. All the other theories I've seen trying to explain other issues, like the scoring downturn in recent years, appear to make sense at the surface, but until they have a theory that covers the start of that era, as well as the end, and particularly how baseball suddenly, over a 1-2 year period, started scoring a significantly larger number of runs.
Stats to the Rescue
While I was fuming, I realized that the run scoring data could be analyzed statistically to see whether the new era data is proved or not proven by the null hypothesis that the new era data is the same as the old era. Maybe that will give me an answer of some sort that will convince some more people. Here is the data for the period 1978-2013:
The population is the years 1977 to 1993, which I will call Regular era. Unlike Walker, I included 1993 into the pre-offensive period because, if you look at his table of ratio against the average, 1993 was not larger than the high for the period, in 1987. 1994 was the first year in sequence that was above the high for the prior period.
Given this definition, the offensive era therefore starts in 1994, but when does it end? I decided to do as I did and end it with the first season below the Offensive era starting 1994. Since 2009 is the first season lower than any in the period starting 1994, I used 2008 as the end point of the Offensive era and 2009 as the start of the return to the Regular era.
I tested four periods against the population mean and standard deviation, two-tailed, with the null hypothesis being that the new period is not different from the population, that they share the same mean. First, I tested 1994-2000. 1999-2000 was the high point for the Offensive era. I then tested 2001-2007, since it was noted that scoring was declining during this period. Then I tested 1994-2007, since that is the offensive era. Lastly, I tested 2008-2013, since that appears to be the New era, in a return back to the pre-1994 Regular era.
I tested NL, AL, and MLB. In hypothesis testing, if your z-stat is above +/- 1.96, the hypothesis is rejected at the 5% level, if it is over +/- 2.58 the hypothesis is rejected at the 1% level, and over +/- 3.30, at the 0.1% level. Basically, the higher the z-stat, the more likely the new sample stats does not belong to the population it was tested against (i.e. 1977-1993). If your z-stat is 3.30, that means that the probability of the population having the same mean as your sample is 0.1%, in other words, very unlikely. And in the table for z-value, if your z-stat is 3.90 or higher, it is pretty much 100%
Hypothesis Testing: 1994-2000
Here the z-stat was these:
- NL: z=8.189
- AL: z=9.026
- MLB: z=8.984
Hypothesis Testing: 2001-2008
Here the z-stat was these:
- NL: z=6.689
- AL: z=5.616
- MLB: z=6.398
Hypothesis Testing: 1994-2008
Here the z-stat was these:
- NL: z=10.492
- AL: z=10.266
- MLB: z=10.811
Hypothesis Testing: 2009-2013
Here the z-stat was these:
- NL: z=0.722
- AL: z=0.510
- MLB: z=0.570
Other Hypothesis Testing
I also decided to test how the new era compared to the Offensive era and the z-stat ranged from -5.832 to -6.543, which mean we can reject the hypothesis that the new era is from the same population as the Offensive era.
In addition, I tested the New era against 2000-2008, to see if those two periods are just part of a continuous decline, which was mentioned, and it was even more clearer that they were different data sets. The z-stats ranged from -7.250 to -9.325, which is even more unlikely than the offensive era. There may have been a decline (and if you look at it, it was more up and down, but that is how it was described), but the period 2000-2008 is statistically significantly different from the 2009-2013 New era.
Wow, I did not know what would happen when I did this. But this data confirms without a doubt that, statistically speaking, major league baseball moved from one era of scoring, the Regular era, as defined as 1977 to 1993, to the Offensive era, as defined as 1994 to 2008, and then returned to the same level of scoring as before in the period after that, the New era, from 2009 to today, 2013.
I didn't know exactly what would happen when I dug into the numbers. I half panicked at one point and wondered if I would have to eat a big plate of crow because what if people were right that Walker seemingly broke up the data at odd points or that the change in scoring was not that significant. But once I had the epiphany that my conundrum could be studied via simple Hypothesis Testing, I had to get that done and see what the answer was.
In my wildest thoughts, I would not have thought that the results would be so clear cut. I thought that maybe there would be a lot of grey area. That is what happened when I looked at managers' performance versus Pythagorean, while the likelihood that managers could out perform their Pythorean was high, in the 60-80% range, if I remember right, statistical testing, at least in school in elementary stat classes, start with 95% level or go higher, up to 99% level. As I noted above, anything above 3.90, plus or minus, is pretty much 100% that you reject the Null Hypothesis, as the normal table I consulted basically ended at 3.90, it did not bother to go any higher. And all the comparisons resulted in very high z-stats, the lowest in magnitude was 5.616.
I played around also with 1993, 2008, and 2009. 1993 statistically probably belonged to the Offensive era. I'm not going back and updating everything because this was conservative, changing it would make the z-stats even larger by reducing the average and particularly the standard deviation of the population. But for example, 2000-2008 would have had a z of 7.651 instead of 6.689. 2009, however, was more of a Regular era season than an Offensive era season, while 2008 was definitely an Offensive era season.
So what happened to cause scoring in baseball to suddenly flip the switch from Regular era to Offensive era back to New Regular era? Most of the reasons I've seen don't fit the data.
Steroids usage would have the normal S-curve adoption rate, and there should have been a gradual rise, over time, instead of a jump, as there was in 1993-1996, plus a sharp drop in 1997-1998, before another rise in 1999-2000. And we know that usage had started at minimum in the late 80's when Canseco and presumably McGwire were using, yet 1988 and 1989 were two of the lowest scoring years in that time period, then 1992 was also that low again.
Usage resulting in increased offense should not be jumping about like this and one would think that the offensive era would be gradually rising sometime in the late 1980's, yet things were pretty much the same until 1993-4. And in any case, no matter what anybody with anecdotal evidence says, there is no scientific proof that it works, but if you look in the studies in Eric Walker's steroid website, there are some that show that it does not work like how people think it does.
Maple bats was another technology that maybe contributed to the offensive era. However, maple bats were not adopted by every player, just a small but significant subset of players. And they were boosting usage in the 2000's right when the so-call decline period happened (really, if you look at 2001-2008, it was just random ups and downs, bouncing around the mean).
Now, I can see the logic that the decline period could be credited to the changes to the strike zone, as the data does indicate that there was some sort of decline, but still, that entire Offensive era period, both halves, are not like the Regular period, which is what is the focus of this study. So while there is an apparent decline in the 2001-2008 period, the data from that period is still significantly different from the Regular era population.
And with pitching technology, whether it be medical advances or new pitches or new combination of pitches, it would not make sense that everyone would change that quickly, basically in one season, then bounce around, the decline should be pretty clear over time, if the technology is that good, the information would spread, adoption would increase, and RS should be falling steadily over time. But RS was 4.700 in the NL in 2001 and 4.706 in 2007, shouldn't the new and improved pitching caused some significant decline, if that is the theory, that the new pitching technology caused a decline in RS. Same with the change in strike zone noted above, why was offense so similar in 2001 and 2007? The only clear period of decline has been from 2006 to 2013, with a pause in 2012, though, so maybe there is something to this theory, only later than what I had see postulated, which is that it started in 2001.
So the only thing I can see that makes sense is Walker's theory that the MLB changed the ball in some way in 1994. How does scoring suddenly surge and stay in that new range?
Plus, the randomness of both eras were approximately the same. The standard deviation for the Regular era was 0.187 for the MLB vs. 0.162 in the Offensive era, which is pretty close, that works out to a difference of 4 runs over a 162 game season (out of 674 runs on average for the Regular era). It would be even closer had I shifted 1993 into the Offensive era.
This all suggests one level of play during the Regular era, then a higher level of play during the offensive era, but with the same randomness in data. And the Standard deviation for the New era is 0.165 for the MLB, right in line with before. So from one regular plateau to a higher plateau, then back to the regular plateau, once more. You cannot change all of baseball scoring like that except by one common thing: balls and bats. If everyone switched to maple bats, I could maybe accept that reasoning, but only a vocal minority did.
So that leaves the baseball. And it makes sense. Baseball went to the live ball in the 1920's (plus rule changes on replacing ratty balls more frequently and outlawing the spitter) and the surge in revenues made it into the national pastime. The public's cynicism about baseball was at an all time high in that time period, as Bud Selig took over as Commissioner in 1992 and the divisive strike occurred in 1994.
So, and this is my supposition, I can see why he would not want to announce that the baseball was changed, because there is only one reason to do that, it's to boost up the offense in the game, and there is really only one reason to do that: to curry the public's favor. And that would be a very blatant move and cynically received by the fanbase. The McGwire-Sosa battle to catch Roger Maris's record would not have been so fascinating to and well received and followed by the fans (sucked me back in, I had stopped following baseball as intently after the strike, and only started coming back when Sabean was named GM, and the record chase got me addicted again) had they known it was due to a livelier ball.
So there you have it. I've analyzed the scoring in baseball, and in both leagues and thus the majors, scoring was elevated by something significantly during the offensive era, which is as people saw, but now there's statistical evidence instead of just talk about such a change or a discussion of the relative sizing (which had convinced me, but apparently did nothing for others, and thus I took it a step further). The offensive era was a real break from prior levels of baseball performances, statistically significantly speaking.
I also found that the scoring in the normal regular way the game was played prior to 1993 is basically what we have returned to in recent years, since 2008. It is almost spot on what it was before, with very similar mean and standard deviation.
What caused baseball to rise suddenly to a new plateau of scoring during the offensive era, in basically 1-2 seasons, 1993-94? Then returned to the prior level of play over a 1-2 season period again, 2008-2009? The only explanation I've see that made sense is Eric Walker's theory that the ball was juiced. Every other explanation I've seen could explain a rise over time, but not a statistically significant jump plus decline in a short 1-2 seasons timeframe. I'm still waiting to see a better explanation.