In 2011, in an effort to have non-wood baseball bats perform more closely to wood bats, the NCAA mandated that those bats conform to a BBCOR (coefficient of restitution) rating of 0.50, replacing the previous BESR (ball exit speed ratio) method, which estimated the batted ball speed. The National Federation of High Schools (NFHS) will require compliance with NCAA standards in 2012, although some states instituted the change this year.
According to the NFHS, “the new standard ensures that performances by non-wood bats are more comparable to those of wood bats. It’s also expected to minimize risk, improve play and increase teaching opportunities”.
I’m the developer of the Hardball Times Forecasts, which include college batting and pitching statistics as part of the player analysis. The change in the bat standard created a discontinuity in the statistics which would prevent me from using data from before and after the change until I was able to quantify to differences in the two sets of data.
Dr. Alan Nathan was a member of the NCAA Baseball Research Panel, a group of scientists who provide advice to the NCAA on issues regarding bat performance. He had written an article at Baseball Analysts in January 2010 which described how the BESR and BBCOR measurements work. He explained to me:
“By spring 2008, the BESR standard had been in place for some seven-eight years, and I was looking for a way to improve on that performance metric. It had been known for some time that there is not a direct correlation between the BESR of a bat, as measured in the laboratory, and performance on the field. The latter is defined by batted ball speed (BBS). That is, two bats with identical BESR values would not necessarily have the same BBS.
“So, I was looking for a way to improve that situation. Armed with lots of data on bats that had been tested up to that time, I did lots of data analysis and calculations. The net outcome of all that effort was that I was able to show that for NCAA-type bats (which have a fairly narrow range of lengths and weights), BBCOR provided a direct correlation with BBS. By that I mean, if bat A has a higher BBCOR than bat B, then bat A will have a higher BBS than bat B. Having found the metric that I was looking for, I proposed it to the rest of the panel. The full panel saw merit in it and we agreed to recommend it to the NCAA. By later that summer, the NCAA Rules Committee agreed to adopt it.”
Dr. Nathan told me that lowering the bat’s BBCOR to 0.50 would reduce the batted ball speed by about 5 percent, which would be 5 mph for a typical hard hit ball, which would reduce the distance of a long fly ball by 25 to 30 feet. This was expected to reduce home runs by up to 60 percent, and a quick and dirty check of home runs per game at the end of the 2011 seasons showed an approximately 50 percent reduction.
With a lower batted ball speed, other offensive statistics could be expected to change as well. The problem this posed for me, the data analyst, was how to mix statistics from 2010 and before with 2011 and later when projecting player performance.
To measure and account for the difference in performance of the bats, I generated next-season projections for all the batters in my college database, which is mostly Division I schools from 2002 until 2011. I used the same projection methodology as in the weekly THT Forecasts, except that only college data were input.
One complication was that the projections require a birth date, which is normally not a problem as I project the players only after they turn pro. But to conduct this study, I needed as many additional birth dates as possible for players who had not played past college. Some schools publish players’ birth dates on their roster pages, but others don’t. After much digging I had birth dates for about 12,000 of the 40,000 players in my batting table.
After the projections for seasons up to 2010 were best fit with the actual performance in the projected seasons, those adjustments were made to the 2011 projections and then compared to the actual 2011 stats. Those results were gathered into matched pairs of projected and observed totals for each player, scaled to the smaller of the plate appearances.
I had a total of 1,977 players with 2011 projections who subsequently played in that year. The following table shows the ratio of observed to projected totals in all the batting categories, at various minimum sizes of weighted plate appearances.
size players PA BH XBH TR HR HP BB SO SH SF GDP 0 1,977 268,128 0.939 0.931 1.048 0.567 0.945 1.008 1.040 1.115 0.872 1.178 100 1,410 247,545 0.938 0.930 1.042 0.567 0.938 1.016 1.043 1.088 0.886 1.200 200 966 195,796 0.937 0.931 1.057 0.563 0.925 1.022 1.047 1.046 0.889 1.208 300 540 117,458 0.944 0.920 1.010 0.575 0.927 1.024 1.035 1.010 0.897 1.229 400 296 67,676 0.942 0.911 1.002 0.586 0.918 1.026 1.037 0.974 0.943 1.212 500 109 26,341 0.951 0.920 0.932 0.566 0.930 1.006 1.033 0.921 0.817 1.107 600 20 4,930 0.927 0.855 0.825 0.492 0.905 1.027 1.020 0.869 0.842 1.125
Indeed, when measured on a player-by-player basis, home runs were down 43 percent. Base hits per balls in play as well as extra base hits (doubles and triples) per base hit were down, while triples per extra base hits were up. I’d guess that outfielders might be playing closer, allowing fewer balls to fall in, but when a ball did get by an outfielder, it was more likely to go for three bases. The only stat which varied by sample size was sacrifice hits. Players with smaller sample sizes (bench players or those with little experience) bunted progressively more often as the sample size decreased.
I now multiply both batting and pitching stats from 2010 and before by these factors (at size = 250) to get a version normalized to the 2011 baseline, which can then be inserted into the same table with the 2011 (and later) stats for furthering processing. As I am now using a new baseline, the league factors used to create major league equivalencies (MLEs) have changed, but they can now be applied to data from all seasons.
My measurement of a home run rate reduced by 43 percent is not as much as the 60 percent that was anticipated from the theoretical model of bat performance, but it is still a significant reduction which likely meets the goals of returning amateur baseball to the offensive levels seen with wood bats. A legitimate question would be, “Why not just use wood bats?” as the new BBCOR bats cost more than $400 each. Prominent collegiate leagues and the All American Amateur Baseball Association have mandated wood in recent years, but even with the large cost of each BBCOR bat, the rate at which wood bats break and need replacement keeps the long term cost of BBCOR bats lower than wood.
References & Resources
References and Resoruces
Dr. Alan Nathan’s article at Baseball Analysts on measuring bat performance
“Steve the ump” offers some advanced math on composite bats
NCAA bat standards
Article by Amy Farnum at ncaa.com
High school standards and player reactions