Sunday, June 6, 2010

Replacement level in college basketball

In baseball, the concept of replacement level, popularized in the 1990s by the likes of Keith Woolner (now working for the Cleveland Indians), has developed into the standard way to evaluate players. Arguably, the main reason for using replacement level is to balance rate stats and counting stats, and to provide an accurate assessment of a player’s total marginal contribution to his team.

If you use average as the baseline to rate player performance, then it looks like an average player has no value. We know that an average player does, in fact, have value. A team of average players should win 50% of their games. What we really want to compare a player’s performance to is some baseline below average, preferably at the theoretical replacement level.

Though there is still much debate in the baseball analysis community over how to precisely define replacement level, the overall idea is relatively simple. Here’s a basic definition of replacement level, offered by Woolner (linked above):

Replacement level is the *expected* level of performance the average team can obtain if it needs to replace a starting player at minimal cost.

Here’s a basketball-specific example:

Player Points/game Games
Player A 20 25
Player B 15 40

Obviously, using points is simplistic, but the numbers are for illustration purposes only. So, which player is more valuable? Let’s assume a replacement level player scores 5 points a game. In a 40 game season, Player B, playing in every game, provided 15 points/per. However, player A only played in 25 games, so we have to add in 15 games of replacement level production (5 points/game). His new average is 14.4. Or we could say Player B is 400 points better than replacement level (a RL player would score 400 points less, in the same number of games) , and Player A is 375 points above replacement level. Same thing. So, these two players, with different levels of playing time and performance, are basically equal.

But how do we truly define replacement level in basketball? As we have discussed before, basketball analysis is not baseball analysis, for a variety of reasons (most notably, they are different sports!). And, more specifically, college basketball is not NBA basketball.

In the NBA, when a player goes down mid-season, the organization has many options on how to replace that player. They can elect to do it entirely with players already on the roster, simply changing playing time and/or positions around. Or they can go out and look at available free agents or players in the D-League. Trades are also a possibility. In college basketball, once the recruiting period is over and the season has begun, teams are essentially restricted to playing out the season with their roster.

So, we are left with (at least )two questions. How do we define replacement level in basketball? And, specifically to college hoops, how do we apply this concept? Should replacement level be the expected performance of the last player on the bench, or the best player on the campuses club team? These questions are not easy, but if we are able to define replacement level in college basketball, we can gain a better understanding of player value.

Spam – Not good in any form

I just noticed that the comments section of this blog has been flooded with spam. Nothing worse than poking around a blog, only to see it filled with spam.

I have now enabled comment moderation, which means I’ll have to moderate any comment before it appears on the site. This is not a big deal, as this blog has received a total of about five comments. Please, my fellow humans – no spam-bots allowed! – I encourage you to comment on any posts, whenever you have something to add. I have not been posting much recently, as you have probably noticed, but hopefully that will change.

Bearcat Blogger

Goodbye, Lance

It has been a while since I’ve written here – the spammers have let me know, flooding the comments section. It has also been a while since I’ve diligently followed the Bearcats, as my interests have given way to baseball, school, and other “priorities.”

Apparently, Lance Stephenson is not returning, as he will enter the NBA draft and has hired an agent. Stephenson had an interesting season with Cincinnati, at times dazzling with his superior athleticism and understanding of the game, while at other times playing a non-factor.

Overall, Stephenson averaged 12.3 points, 5.3 rebounds, and 2.5 assists in 28 minutes a game. He averaged 2.4 turnovers and one steal per game, while shooting .440 from the field, .664 from the line, and an ugly .219 from three (he shot 50% on 2pt FGs).

While Stephenson’s presence will undoubtedly be missed, note this nugget: His effectiveFG% of 46.2% was below the team average of 47.6% (neither figures particularly good, by the way), and Lance shot a team-leading 26% of the time while on the court. He is going to be missed, but his shoes are not impossible to fill.

Best of luck in the NBA, Lance.