<
>

Exploring categorical variance

Big men such as Shawn Marion specialize in low-variance categories like rebounds and FG%. Jerome Miron/USA TODAY Sports

The NBA trade deadline? An exceedingly distant memory.

Your fantasy league trade deadline? Maybe it already happened. Maybe it's about to happen. In some leagues, it won't happen (I hate those leagues.)

But regardless of how you spin it, we're faced with a looming truth; we're dealing with a rapidly diminishing amount of games left in the campaign.

As the regular season inches closer to the fantasy playoffs, I like to take a look a certain statistical tendencies in weekly or nightly increments. The margin for error is shriveling.

Every single statistic -- a single rebound or missed shot -- could determine the margin of victory or defeat. Every decision.

When making lineup decisions, matchup plans and waiver-wire moves, it's always helpful to know which statistics are the most and least reliable. That is, which stats fluctuate the least on a weekly basis? Which fluctuate the most?

Both sides of the spectrum are important. Because we're talking about variance. As in the least amount of deviation from the mean.

Which categories have provided the smallest amount of night-in, night-out fluctuation this campaign?

And why do we care? Because we want to know what we can count on. And we want to know what we can't count on.

If there was a statistic that was flatlining on a weekly basis -- where it was holding relatively steady -- we'd want to prioritize that stat in a favorable matchup. No need to swing for the fences in those numerical areas.

You want players who provide reliable, steady production. You want a statistical base you can rely on.

Let's say you're in a head-to-head matchup in a points-based league. You lose a regular to injury, and want to add a player for one game. You just need this player to contribute in a steady, reliable fashion; you don't need him to post an atypically productive stat line. You just want a guy you can plug in and bank on for some low-end production.

You'd want a player who excelled in statistics that don't vacillate on a game-by-game basis.

Conversely, what if we're in a head-to-head matchup where we don't like our chances? Or a rotisserie league where we needed to make up some ground in a hurry?

In those spots, we'd want to prioritize he swing-for-the-fences stats.

Let's say your team averages 50 steals a week and is matched up against a team that averages 55 steals per week. If there was an average variance of just 5 percent per week in steals, you'd have a real chance of beating the odds and winning that category.

Maybe your team snares 5 percent over its average and posts 53 steals. And the other team underperforms by about 5 percent and posts only 52 steals. There's a category win.

With all of this in mind, I sent a request to the heroes and heroines of ESPN's Stats & Info department. I asked them to take a look at the weekly NBA-wide totals across the major fantasy basketball categories (FG%, FT%, AST, 3PT, REB, BLK, STL, TO, PTS) and measure which stats had the most and least amount of variance. Their percentages of variance by category are outlined in the chart above.

Here's a graph illustrating the differences:

What do these numbers tell us?

1. It's smart to build around percentage categories, even in head-to-head matchups. You should do it even when only considering a small block of games.

It's a strategy I always emphasize at the beginning of the fantasy season when talking draft strategy. Not only do the percentage categories tend to be undervalued by fantasy hoops owners, but they also tend to be the most reliable throughout the course of a season. Not to mention an NBA career.

2. When we want to swing for the fences, we should emphasize 3-pointers, steals and assists.

These tend to be less reliable categories, but that can play to your advantage in a one-game, one-week or even one-month scenario.

If you're up against a juggernaut in a head-to-head matchup, you can load up with players who produce 3-pointers and hope for a big, random jump over a one-week period. You can do this while hoping the other team experiences a random drop in 3-pointers.

On the other hand, you can go the other way and load up on percentage-strong players and hope the other team has an off week.

3. Big men offer steadier, more reliable fantasy production.

Look at the four categories that offer the least amount of variance: FG%, FT%, BLK and REB. Which players dominate three of those categories? Power forwards and centers.

Take a look at Serge Ibaka; in his last five games, he's put up the following shooting numbers from the field: 8-of-11, 3-of-11, 4-of-9, 5-of-11 and 10-of-16. Some variance, but not as much as you'll see in the next four categories.

Here are some players available in more than 50 percent of leagues that would fit this low-variance scenario:

Shawn Marion, SF/PF, Dallas Mavericks (available in 52.0 percent of leagues)

Amir Johnson, PF/C, Toronto Raptors (available in 58.1 percent of leagues)

Marvin Williams, SF/PF, Utah Jazz (available in 84.3 percent of leagues)

Miles Plumlee, PF/C, Phoenix Suns (available in 62.7 percent of leagues)

Andray Blatche, PF/C, Brooklyn Nets (available in 81.0 percent of leagues)

4. Guards -- especially PGs -- offer more boom-or-bust production.

Look at the four categories that offer the highest amount of variance; 3PM, STL, TO, AST. Point guards check all four of those boxes.

Players who handle the ball with a higher degree of Usage Rate are going to produce more in these categories. You have to have the ball in your hands to generate assists. The negative side of that equation is that you'll also generate turnovers.

Take a look at the assist totals in Goran Dragic's last five games: 14, 6, 5, 3 and 3. That's a wild swing over the course of 10 days. Over the same five-game span, his 3-point production went as follows: 1, 2, 0, 5 and 1.

Here are some players available in more than 50 percent of leagues that would fit this boom-or-bust scenario:

Danny Green SG/SF, San Antonio Spurs (available in 50.9 percent of leagues)

Lou Williams, PG/SG, Atlanta Hawks (available in 67.7 percent of leagues)

O.J. Mayo, SG, Milwaukee Bucks (available in 61.3 percent of leagues)

Vince Carter, SG/SF, Dallas Mavericks (available in 82.9 percent of leagues)

Kent Bazemore, SG, Los Angeles Lakers (available in 87.0 percent of leagues)