Monday, April 18, 2022

Did Stats, LLC., Change the Definition of "Stuffs"— Or Did ESPN?

By John Turney 

In 1992 Stats, Inc., as it was known then, introduced a statistic called "stuffs". Simply put it was a tackle for a loss on a run play. Stats, LLC., (STATS) as they are now known tracked store plays ever since. 

For years ESPN used those stats on their online player profiles. Here is the Aaron Donald page screengrab—

We've highlighted the "STF" or "stuff" column for 2015 and 2021. You'll notice that his career-high is 16 in 2021. Next is 13 in 2015. Well, not so fast.

In 2021 either STATS changed their definition to include tackles for no gains (and for pass plays for no gain and for a loss) or ESPN is using another source for stuffs. We don't know which. 

Here are the 2015 plays in question. You can see the ones for losses total 13.5 (The above chart shows 13, we think that is a rounding thing for the chart). But if you add in the tackles for no gain, using the method that STATS used the total would be 17—

For 2021 the plays in question total 10.5 (using the STATS method for stuffs) that were for a loss. But if you add in the tackles for no game it totals 16—the career-high in the ESPN bio above—

The point? Simple. You cannot always compare stat totals in the same bio unless you know how they were kept and some of the media outlets don't let you know when changes were made.

Using the old method (counting tackles made in the backfield only Donald's 2021 total is 10.5 stuffs, not 16. If you use the "new" method his career-high is 2015, not 2021.

it does not matter which method is used, whether one considered a "stuff" to be for no gain and for losses or for just losses, it just matters that they all be counted consistently.

Don't get us started on NFLGSIS's tackles for loss, that is confusing. 

SO, when someone tells you Aaron Donald had 16 stuffs in 2021, they are right. But they are also telling you something that puffs up his total in relation to previous seasons. And this applies to all the ESPN bios found online. Someone, we suppose, just decided that plays for no gain count as a stuff and didn't go back and conform all the previous years to that same standard.

That makes it confusing. It does not take away the terrific play by Donald in 2021 (or any year) it doesn't matter in the abstract if his career-high in stuffs in 2015 or 2021 but if you are going to count things it only makes sense to count the consistently. 
 
If anyone knows the source ESPN is using for their bios, please let us know. We'd love to get further information to sort this out—If STATS is the provider and they changed or if ESPN is using their own inhouse stats (ESPN Sports & Information) to populate their bio charts.

Now, a discussion for another day is how Donald was credited with 46 assisted tackles in 2021 when his career-high before this year was 26. Bottom line: Tackle stats are hard to trust.

2 comments:

  1. meticulous analysis as usual John....now can you explain the variables and protocols used by espn and the other networks to come up with stats like "83.6% chance of winning as of x moment in y game"?.....from where do these numbers derive and what proof is there of validity? (and if not, what's the point of these being promoted/pushed out there by media?

    ReplyDelete
    Replies
    1. It's based on historical data using score, time remaining, down-distance, etc. Now there are areas where they have to "fill in" the data. E.g.: There may not be in their history a 4th and 32 in Q2 down by 28-points, so in some cases they have to use extrapolation. Also they often use base odds so if a 15 - 0 team is playing a 2-13 team they'll start with a win probability skewed to the former. It's generally well thought out, but there are certainly ways to poke holes in it.

      Delete