A lot of times, people who aren’t numerically-inclined will ask me how advanced stats can help them. It’s tough to come up with a useful response because statistical analysis is as much of an art as it is a science. That doesn’t mean there aren’t some basic things that analytics can do. Separating 2-point percentage from 3-point percentage and offensive rebounding from defensive rebounding are easy places to start.

But part of the art of analytics is trying to separate the real from the imagined. The thing I want to focus on here is the notion of a shooting slump. In general, people are too quick to throw the S-word around.

There’s a lot of stuff out there about Baylor’s Brady Heslip and his slump last season. This piece from Andy Katz illustrates the perception, and it was a running theme of Baylor-related talk during conference play last season. If you want to use ‘slump’ to describe a player whose shooting percentage decreases from one season to the next, then it applies here as Heslip made 45.5% of his 3’s as a sophomore and 38.6% of them as a junior. But generally I take slump to mean: “[Player] has lost his ability to make shots. He needs to get in the gym and work on his form.”

To that end, I wouldn’t be too concerned about Brady Heslip. Fellow stat-geek Jeff Goodman annually publishes a list of the best shooters in the country during the off-season. This might be his best contribution to the basketball community. Here’s how the top 25 on that list have performed the following season.

Season  3PM   3PA  3P%
2010   1931  4967 38.9
2011   2015  4985 40.4
2012   1842  4646 39.6
2013   2170  5358 40.5
Tot    7958 19956 39.9

That’s actually a pretty good track record there for Goodman and most people would be hard-pressed to do better. (You might, for instance, just try using a list of the 25 most accurate returning 3-point shooters. You’d do slightly worse, though.) The takeaway is that if you’re truly one of the best shooters in the nation, a good guess for your 3-point percentage in the upcoming season is 40%. The average player on Goodman’s list has gone 80-for-200 over the past four years. Heslip went 83-for-215 last season.

But Heslip didn’t just make Goodman’s list, he was ranked second on it. And the higher ranked players tend to shoot better, so we should account for that. I did a quick Excel-style regression that wouldn’t fly in a published journal, but it’ll do the trick here.

Expected 3P% = 41.88-0.916*ln(Goodman Rank)

Plug in 2 for the Goodman Rank and you get an expected 3-point percentage of 41.2. The players ranked second in prior seasons made 41.4% of their attempts in the subsequent season. Somewhere around 41% would have been a reasonable expectation for Heslip as a junior. In 215 attempts, he would be expected to make 89. (Technically, 88.6.) He made six fewer. That’s hardly worthy of panicking over Heslip’s shot. At least, that shouldn’t be one’s first guess to explain the decrease in accuracy.

The problem for Heslip is that he made 45.5% of his attempts in 2012 and non-analytics folks tend to assume that’s his true ability. But that’s not fair. According to College Basketball Reference, if Heslip maintained that percentage over his career, he would have had the sixth-highest career 3P% of the last 15 years. Almost every shooter with that level of accuracy in one season declines in the following season.

(If we were to get more sophisticated with the model, I think we’d find taller players and guys who do more than just shoot 3’s probably don’t drop as much, and shorter one-dimensional players tend to drop more. Doug McDermott’s expectation this season might be to shoot one or two percent better than my crude model suggests.)

We can never know Heslip’s true shooting ability with complete certainty, but even after making 45.5% of his attempts, a reasonable assumption would have been that he wasn’t quite that good. In order to shoot over 45% in a season, you need to be a good shooter and you need some other things to go your way that have nothing to do with your shooting ability. It’s possible Heslip has the mechanics of a guy that should shoot 41% against a normal defense.

Of course, that figure is a composite of a variety of shots. Maybe Heslip makes 60% of his wide open looks and 30% of his contested shots. Maybe he had more wide open looks as a sophomore than as a junior. Yes, that’s a reasonable explanation, but it’s just as possible there were things occurring we can never fully understand. Even a player whose true ability is to make 60% of his wide open shots will rarely make exactly 60% of his wide open shots. And some weeks maybe his form is slightly worse, so he’s a 55% shooter when open and other weeks it’s better and he’s a 65% shooter. But those kinds of differences are nearly imperceptible and I doubt they could be reliably noticed by all but the very best shot doctors. This kind of natural variability may explain more of Heslip’s decline in accuracy than changes in shot quality or his role in the offense.

So one way I would use stats is to be more thoughtful about what they mean. For most players, I don’t think it’s helpful to assume they are slumping when they are not. Maybe such thoughts would motivate some players to improve in the offseason. But what if there’s nothing wrong with the slumping player’s form? Now they’re focused on something they shouldn’t be because they think they’re in a slump. I don’t think that would be productive, either. Assume no slump until the data or other information (injury, role change, defensive adjustments) make a convincing case otherwise. It’s not convincing in the case of Heslip.