Of course this season has seen a class of rookies up there with the very best years (at least at this stage of the careers), and we are seeing big minutes played by a number of guys, with 9 of the 29 first round picks cracking the starting lineup with regularity!

At the same time, some veteran players with perhaps only average personal statistics have become coach favorites on the grounds that they know what it takes to win in the NBA. So the question is, do players have to learn how to win?

Rookie numbers of note
We're not going to worry about individual stats for the rookies at the moment since what we are trying to ascertain is are they a drag on a team's chance of winning. Instead we'll focus on a couple of key "how the team plays with this player" metrics:

• Plus/Minus -- the raw points scored minus points allowed for the team while the player is in the game
• Won-Lost % -- charting the plus/minus on a game by game basis, we record a Win for the player if the +/- is +1 or more, and a Loss if it's -1 or worse
• Roland Rating -- the difference between the player's on court and off court "Net48" ratings, which is just plus/minus scaled to 48 minutes. In other words does the team play better (a positive rating) with the player on the court, or worse (a negative rating)

Well at a glance you might be tempted to say yes the rookies as a whole need some seasoning before they can develop the all around NBA game that contributes to a team winning. After all, the first rounders from this exceptional rookie season have combined for a -709 plus/minus number to this point, with a 43% W-L percentage, and a weighted average Net48 of -2.4 points per game.

However, it has to be recognized that the rookies with the most playing time typically play for the teams with poor records, whereas the teams with good records give fewer minutes to their first year players. This bias would skew the plus/minus on an overall look, which is why turning to the Roland Rating is a better way to examine the question. Are teams playing better when the rookies are on or off the court?

The answer turns out to be basically neither -- the rookies as a group are on a weighted basis (take the rating times the minutes) show a virtual breakeven -0.1 Roland Rating. In other words there's no sign there whatsoever that Rookies know less about what it takes to win than your typical NBA non-rookie player. Indeed if we made the weighted average across all players in the league it would be a balanced 0.0 net.

Now you can still challenge this assessment by stating that the players who are not producing will see less playing time and thus we are weighting the average in favor of the players who are, and on a player tally we find that 11 players have positive Roland Ratings, 13 have negative ratings, and 3 have virtual even ratings (between -0.9 and +0.9). Of the players in for at least 20% of the total team minutes so far, 5 are positive, 7 negative, and 2 even.

So our initial reaction is that perhaps the notion of this needing to learn how to win is part myth. Of course other objections to the premise we are building on (that the teams do not play better or worse with rookies on the floor overall), notably that the rookies who play the most are indeed super-talented and their Roland Ratings if you will at this point may be slightly positive, but will move into the very positive domain as they get more experienced. This does seem believable and likely, but unfortunately we only have last season's data and this season to work with.

Still, even if it is the case that the rookies will get better with years on their uniforms, there does not seem to be grounds for not playing them on the theory that they cause undue harm to a team's performance.

Another key objection with merit is that perhaps the minutes they do receive are more commonly in garbage time and thus to be disregarded if the team does show a positive score, especially if the rookie comes into the game with his team trailing by a huge margin. We could do some checks on this, but the counterargument is that both teams probably slide their rookies in on such occasions, which should even out the equation.

Finally one year of rookies is not by any means a convincing sample, and in light of the almost undisputed brilliance of the 2003-04 class, we may find different results in different years. Hence we will turn to the 2002-03 rookie brigade, who at this point are regarded as a down year in the rookie heritage (may they prove this assessment wrong!)

So with the "weak" 02-03 class we do find some similarities -- the -3.3 net48 isn't that far from the -2.4 of the 03-04 crew, and likewise the 42% game-by-game W/L plus minus is in line with the 43% we are seeing this year. The Roland Rating was however somewhat worse for last year's group, as a -0.7 is starting to become a bit more significant, and there were 9 positive, 15 negative and 3 even ratings in the mix.

Unfortunately then without more years to refer to the results are inconclusive: the strong 03-04 rookie class defies the conventional wisdom, while the uninspiring 02-03 first rounders suggests that rookies as a generalization may have some level of damage to a team's winning outlook.

Again, using complete, unadjusted by circumstance season stats may not be the best approach. We could look only for instance at our current "clutch time" performance (or any other time/game situation breakout), but we'll save those numbers for another day.