http://fivethirtyeight.com/features/the-hidden-value-of-the-nba-steal/

So, I came upon an interesting article that talks about how undervalued the steal statistic is in the NBA. I linked it above, and stats geeks might kind of enjoy it. But it gave an interesting model to apply to NBA stats that I decided to apply to this year's team. For a lot of reasons, I don't expect that this model translates perfectly to college basketball, but it got me thinking and I decided I would apply it to this year's team anyways--just in case there's some insight to be found.

So to give a short premise of the article, this statistician has come up with scaled values of some of the Box Score statistics, and (supposedly) has found that steals are a HUGE factor in determining a team's success--even relative to scoring. Now this model leaves out a TON of stuff, (Field goal percentage, 3 point percentage, Free Throw percentage, Fouls, etc, but I decided to take what it gave me and apply it mathematically to this year's team). The model he came up with goes as follows:

Points are straightforward. 1 point=1 point in the model.
Rebounds are worth 1.7 points/rebound
Assists are worth 2.2 points/assist
Blocks are worth 6.1 points/block (sounded high to me at first, but a shot blocker can change shots that he DOESN'T block, so I decided to roll with it).
Steals (the main subject of the article) are worth 9.1 points per steal! To me, this sounded (and still does sound) WAY high. Certainly steals can lead to transition points, but I'm not sure where the other 7 come from. My guess is lurking variables that don't show up on the box score. Maybe a defender that gets a lot of steals also closes out better on 3 pointers and contests more shots, or gets beat to the basket less. Either way, it's a correlation number, and I rolled with it and applied it to this year's team.
The only other stat included was turnovers, which are worth 5.4, so I used our players' relative number of turnovers as points against them.

In the end, I calculated a raw score which consisted of a value of the GOOD STATS: points, rebounds, assists, blocks and steals, each adjusted to account for the difference in minutes played, and then I subtracted the BAD STAT (turnovers), also adjusted for the number of minutes played. Here's the data. Mind you, this is a defense-loving model, but I was still surprised at what it came out with in terms of raw scores. These numbers themselves mean nothing, but they provide player-player comparison of (supposed) player contribution based on the model's values.

Good - Bad = Raw Score
Haws 1.270 .298 .972 (low in rebounds, assists and disappointingly low in steals. Must get more active on D)!
Collinsworth 1.672** .423* 1.249 (yes, higher than Haws)
Carlino 1.672** .396 1.276 (highest on the team!)
Mika 1.303 .316 0.987
Halford 1.042 .297 0.745
Winder 1.068 .175 0.893
Bartley 1.254 .219 1.035***
Austin**** 1.171 .220 0.951
Sharp: I did his numbers, but they're not really worth including. It's enough to say that the only thing he seems to do all that well is not turn the ball over. No one else played enough for me to bother taking the time.

What the model would suggest:
*When it comes to Collinsworth, it is what it looks like. Collinsworth, along with Carlino, contribute the most across the combined offensive/defensive spectrum; but KC's turnovers/minutes played are the highest on the team. He must learn to play more under control.
**Yes, Carlino and Collinsworth's combined 'good stats' are equal down the the thousandth decimal. Carlino wins in points, assists and slightly in steals; Collinsworth wins in rebounds and blocks, but as mentioned, turns the ball over a little more. I have more complete stats if needed, but chose not to relay them here as this thing is already too long.
***Bartley's raw score beat Halford's; perhaps more PT is in order for him, especially as he seems to be more defensively active.
****Nate Austin is frankly underachieving. His points/minutes played are the lowest of everyone I included numbers for. He is our most efficient rebounder and shot blocker--but those stats don't cover up his deficiency in scoring and assists. He needs to become a better offensive player.

As a way of scrutinizing the model itself, I decided to include some numbers for Jimmer and Jackson's senior years as well. Scientifically, they felt close enough to me to be positive and negative controls, Jimmer being an offensive guy and Jackson being known more for his defense. This way, we can see if this model favors defense too heavily. (That may be a debate worth having). In my opinion, it goes to show that the model, although it can point out some interesting things, ought to be taken with a grain of salt.

Good Stats - Bad Stats = Raw Score
Jimmer '11 1.577 .535 1.042 (less than Carlino, Collinsworth--just edging Bartley)
Emery '11 1.552 .157 1.395 (higher than Jimmer or anyone this from this year)

Should Jimmer's turnovers count that much against him? Should Jackson's steals count that much for him? I don't know. The article mentions the 'replacability' of a scorer compared to someone who gets steals, etc. There's probably no good way to know. But what the model does suggest that I agree with--especially pertaining to steals, is that defense needs to become more of a priority at BYU. And individual numbers would show that Haws needs the biggest kick in the butt as far as that goes. Anyone would figure Jimmer smashed everyone in the points/minutes category--and he did; but Jimmer's steals and assists numbers were also better than Haws this year. Jimmer was known for neither steals or assists. Getting hands in passing lanes is a good way to start.

Here's to hoping for some improvement in the offseason!