Page 23
What is Replacement Level?
By Kevin Pelton
Mar 17, 2004, 14:30
Email this article
 Printer friendly page

I regularly throw around the term "replacement level" in my columns, but just what exactly is replacement level?

Baseball Prospectus 2004 defines replacement level as "the expected level of performance a major-league team can receive from one or more of the best available players who substitute for a suddenly unavailable starting player at the same position and who can be (or were) obtained with minimal expenditure of team resources."

Things are slightly more subtle in basketball. For one thing, the difference between a starter and a reserve in basketball is less sharply defined. So I would first adjust the baseball definition to refer to "a suddenly unavailable rotation player". Position is generally less important in the NBA, as most players can play multiple position.

The most important distinction in the NBA is that replacement talent generally comes from outside the organization. There are non-rotation players who can step in, but even they will generally need to be replaced with a new player. The farm system concept separates replacement level in the NBA from baseball.

So where do the NBA's replacement players come from? The most obvious source is the NBA's various official and unofficial "minor leagues", the CBA, the NBDL, and the ABA, in the form of call-ups, often 10-day contracts after teams can begin signing them in January. The other location is so-called "freely available talent" during the summer -- players signed to contracts for the minimum salary, generally non-guaranteed deals with invitations to training camp.

Is replacement level important? It's absolutely critical.

An understanding of replacement level is necessary to balancing quantity and quality of production.

For an example, let's turn to NBA.com's "exclusive efficiency rating", calculated per 48 minutes. Kobe Bryant (28.6) and Tracy McGrady (28.5) have virtually identical ratings per minute, but McGrady has played 800 more minutes. In that case, it's pretty obvious to tell who has been more valuable.

But what about McGrady versus Shaquille O'Neal? The Big Aristotle has been far more productive on a per-minute basis (33.2) but has played, again, almost 800 less minutes (1737 vs. 2508). That's 800 minutes where McGrady's been keeping Keith Bogans and DeShawn Stevenson on the bench but O'Neal's been watching Jamal Sampson. By the method I've used (more later), a replacement-level player by this rating system would produce about 16.3 efficiency points per 48 minutes. Both O'Neal and McGrady are valuable to their teams by the extent to which they exceed this level, and how long they do so far.

Thus, we can calculate any player's Value Over Replacement Player (VORP) by (EFF/48 - 16.3) * (MIN/48). McGrady scores 637 VORP, O'Neal 610. Despite being less effective on a per-minute basis, McGrady has been more valuable to his team.

This is extremely important information, both for value-based awards (O'Neal's low minutes total was why I didn't consider him for MVP last season) and for accounting for the value of players like Marcus Camby who are expected to miss a significant chunk of time because of injury. Where you set replacement level matters. Set it high, and efficiency is the dominant factor. Set it low, and minutes played play a bigger role.

The method I've taken to estimating replacement level is to find the efficiency level, by my rating system, of the player ranked number of teams * 10th in per-minute efficiency. This year, it would be the player ranked 290th -- at the moment, Bulls forward Marcus Fizer. The thinking is that rotations stretch about 10 deep, and players below this are generally replacement-level guys.

Does that really work? I'd like to study this year's replacement players to find out. I've found players I believe fit into the two groups I identified, freely available talent and in-season free-agent acquisitions, and we'll look at both groups.

Freely Available Talent

There is no hard and fast rule for this group. I went generally with players who I believe are on minimum-salary contracts (I can't say I checked every one) and were not major free-agent players. Undrafted rookies are a great example. Brian Cardinal, good as he's been, showed just how freely available he was when I asked him during an interview whether there was much interest in him this summer and he replied, "My friends were interested in me."

Player                Team   Eff   Min
Brian Cardinal         GSW  .533  1416
Udonis Haslem          MIA  .465  1410
Milt Palacio           TOR  .357  1136
Bobby Simmons          LAC  .453   973
Eddie House            LAC  .435   899
Tony Massenburg        SAC  .452   764
Daniel Santiago        MIL  .459   636
Amal McCaskill         PHI  .371   611
Michael Ruffin         UTA  .352   603
Jason Hart             SAS  .418   550
Loren Woods            MIN  .440   495
Marquis Daniels        DAL  .497   428
Devin Brown            SAS  .456   415
Keith McLeod           MIN  .377   391
Richie Frahm           SEA  .541   386
Darvin Ham             DET  .388   377
Scott Padgett          HOU  .504   351
Linton Johnson         CHI  .377   262
Tremaine Fowlkes       DET  .368   257
Robert Pack            NJN  .383   219
Mitchell Butler        WAS  .414   216
Ben Handlogten         UTA  .522   172
Bimbo Coles            MIA  .295   169
Jabari Smith           SAC  .398   154
Jamal Sampson          LAL  .488   130
Mike Wilks             HOU  .410    95
Jeff Trepagnier        DEN  .342    89
Rodney Buford          SAC  .325    73
Shane Heal             SAS  .348    72
Quincy Lewis           MIN  .346    65
Sean Lampley           GSW  .559    63
Britton Johnsen        ORL  .338    61
Theron Smith           MEM  .350    53
Alton Ford             HOU  .470    41
Mengke Bateer          TOR  .424    40
Tracy Murray           POR  .323    35
Alex Garcia            SAS  .254    13
Mark Pope              DEN  .061     9
Weighted avg. (38 players)  .424     -

As Cardinal demonstrates, some of these guys have been very effective. Udonis Haslem and Marquis Daniels, a pair of players who went unselected in their respective drafts, also look to have bright futures. Richie Frahm and Ben Handlogten are interesting stories -- guys who have played extremely well in limited minutes, though Handlogten is now injured and Frahm has tailed off since Ray Allen returned. Tony Massenburg helped hold down the fort during Chris Webber's absence, while Daniel Santiago was surprisingly adequate during a two-month stint as Milwaukee's starting center.

The average efficiency I list at the bottom is weighted by minutes played, because an important point here is that while many players have low efficiencies, once teams realize they aren't playing well they can look for another replacement player and stop playing the first one. Bad performance by a replacement player doesn't hurt a team as much as good performance helps it.

The value is .424, which is actually very close to the .430 value I estimated last year. This year, I would have the choice between rounding up or down from Fizer's .427, as I only use twentieths of a point (.425 or .430). This is a good sign. Let's look at these players graphically, minutes against efficiency (click to enlarge):

Last year's replacement level is highlighted in red. Once we get past the point where performance becomes reliable, most players exceed the standard. Past 600 minutes, only Milt Palacio -- whose efficiency rating underrates his defense, but is arguably not replacement level anyway -- misses the standard.

Given the off-season, teams seem to be able to come up with talent that meets my estimate of replacement level.

In-season free agents

Again, I tried to be as accurate as possible here, but might have slipped. I also left out players who started the season on the roster of one team and then were released and signed elsewhere, unless they were replacement level in the first place (in which case, like Robert Pack, they ended up on the first list).

Player              Team   Eff   Min
Rod Strickland       ORL  .463  1027
Doug Overton         LAC  .372   645
Ronald Dupree        CHI  .418   522
Oliver Miller        MIN  .475   455
Matt Barnes          LAC  .431   351
Mark Jackson         HOU  .342   310
Zendon Hamilton      PHI  .534   295
Torraye Braggs       WAS  .432   154
Mikki Moore          UTA  .386   152
Mamadou N'Diaye      ATL  .512   136
Omar Cook            POR  .319   130
Paul Grant           UTA  .375    98
Desmond Penigar      ORL  .466    89
Darrick Martin       MIN  .334    78
Anthony Goldwire     MIN  .404    66
Jason Collier        ATL  .469    64
Cherokee Parks       GSW  .357    64
Dermarr Johnson      NYK  .371    62
Maurice Carter       LAL  .410    50
Matt Carroll         POR  .362    47
Hiram Fuller         ATL  .330    43
Tyrone Hill          MIA  .446    38
Dan Langhi           MIL  .350    37
Paul Shirley         CHI  .286    30
Derrick Dial         ORL  .495    29
Ime Udoka            LAL  .374    28
Eddie Gill           POR  .254    23
Rusty LaRue          GSW  .339    22
Kirk Penney          MIA  .220    18
Damone Brown         NJN  .279    17
Kaniel Dickens       POR  .657    12
Josh Davis           ATL  .103    10
Weight. avg. (32 players) .412     -

It shouldn't be surprising that these players, for the most part, are less effective than the freely available talent. During the season, choices are limited. Many players have gone overseas to play, while others have already made their way onto the roster of opposing NBA teams. There are a few interesting players -- a slimmed-down Oliver Miller has been effective; I've always liked Matt Barnes, and he's played well; and Zendon Hamilton has done nothing but produce during his limited NBA opportunities. The graph:

Most of these players haven't played enough minutes to be reliable. Even of those who have, however, many have not met my estimate of replacement level. Given limited options, NBA teams have not been as successful at scrounging up replacement-level talent during the season.

On the whole, I'm encouraged by the results of this study. I was concerned that I was dramatically overrating replacement-level talent by my method. There's an argument to be made that I should lower replacement level -- perhaps 11 players per team? -- but unless you put a heavy weight on in-season pickups, what I'm calculating for replacement level seems to be pretty close.

"The Page 23 Club"

One of the unfortunate things about this column is that, because of my schedule, I can't commit myself to a specific day or time for publishing columns. To help my readers, I'm starting an e-mail list. If you want, you'll receive an e-mail whenever a new column is up with an introduction to the column and a link. If you're interested, e-mail me at kpelton@hoopsworld.com and let me know. I will, of course, make every effort to protect the privacy of your e-mail address.

Kevin Pelton is an intern for the Seattle SuperSonics and is responsible for original content on Supersonics.com. He writes "Page 23" for Hoopsworld.com on a semi-regular basis.


If it happens in the NBA, it's happening at HOOPSWORLD.

Watch Kobe Bryant, Duncan, Jason Kidd, Webber live. Buy your NBA tickets from Buyselltix.com.

We have biggest inventory for NBA Teams Tickets.  Purchase tickets for Lakers, Knicks, Sixers, Kings,

Mavericks, Spurs, Pacers and Pistons. Buy NBA Championship and NBA Playoffs Tickets also.

Copyright © 1998-2003 HOOPSWORLD.com, All Rights Reserved.