WHAT IS THIS REPLACEMENT LEVEL GARBAGE

[Kathoro]

WHY IN THE WORLD WOULD RATING SYSTEMS USE COMPARISONS TO THE REPLACEMENT LEVEL PLAYER? THEY SHOULD COMPARE TO THE LEAGUE AVERAGE. SURE,RELATIVE TO "REPLACEMENT LEVEL" MAY BE THE IMMEDIATE VALUE OF THE PLAYER TO THE TEAM AFTER THE TRADE DEADLINE, BUT THATS ABOUT THE ONLY TIME IT COULD APPLY. ALL PLAYERS SHOULD BE ASSUMED EQUALLY CAPABLE OF BEING AQUIRED WHEN YOU THINK ABOUT THE GRAND SCHEME OF THINGS. COMPARING TO LEAGUE AVERAGE IS WHAT MAKES SENSE.

[Mike G]

Why in the world would you write in all caps?
It's hard to read.

[Kathoro]

It was early in the morning and I was having fun with it.

[EvanZ]

Not fun since 1994.

As to your question, I calculate WARP so that the total for a team roughly equates to the team's total wins. It makes interpretation easier for some people compared to calculating wins above average (which I also did once upon a time, and people didn't like it).

[bbstats]

I AM ONE OF THE FEW DISSENTERS.

I think value above replacement level is pretty worthless. It doesn't give us any new information.
Using an "actual replacement" (i.e. Danny Green for Ginobilii) is beneficial to teams. Other than that, knowing a player's value above 0 is their value above the average Player-Minute. Replacement level unfairly paints negative players as slightly better.

Here's how I view it:

Rating Above Replacement Level= Player's impact above the median player
Rating Above Average = Player's impact above an average NBA possession with a rational coach

"Replacement" players get below average minutes, so I think it's all a bit silly.

If you adjust for strength of schedule, for example, you wouldn't give a team extra credit because their opponents' coach subbed in players rationally.

Maybe somebody can give me an example where Above Replacement numbers increase the amount of information at hand? The only thing I can think of is making players appear more positive.

[billb]

ALL PLAYERS SHOULD BE ASSUMED EQUALLY CAPABLE OF BEING AQUIRED WHEN YOU THINK ABOUT THE GRAND SCHEME OF THINGS.

Which grand scheme is this, exactly? All players are equally capable of being acquired? Let's just say the Rockets want to speak with you about this alternate universe you've developed.

[bbstats]

What does acquisition have to do with anything?

If that's what we care about, then player level should be regressed against Cost ($)

[Mike G]

.. I calculate WARP so that the total for a team roughly equates to the team's total wins..

So, a team of replacement level players would be expected to win no games? Or they'd likely win a couple due to randomness?
If so, that's how I figure it. Everyone better than a D-league AllStar adds some (non-zero) chance of winning an NBA game.

Winning 41 of 82 games, a team is capable of beating any team, and likely to slaughter a team of players who could be found available to replace a whole NBA roster.

[EvanZ]

.. I calculate WARP so that the total for a team roughly equates to the team's total wins..

So, a team of replacement level players would be expected to win no games? Or they'd likely win a couple due to randomness?
If so, that's how I figure it. Everyone better than a D-league AllStar adds some (non-zero) chance of winning an NBA game.

Winning 41 of 82 games, a team is capable of beating any team, and likely to slaughter a team of players who could be found available to replace a whole NBA roster.

Charlotte's won 7 games. You'd think at least a couple of those guys are above replacement level, right? Maybe not.

If replacement level for an individual player is somewhere around -3, that team would be -15. Charlotte is -9 right now according to my Vegas ratings. Their SRS is -13.5.

As bad as Charlotte is this season, they could be worse, if they started all their bench players. The Warriors have been losing a lot of games lately, and I don't even think very many of their players are replacement level.

[knarsu3]

A little off topic but I was wondering what is a replacement level RAPM?

[Mike G]

The Hornets are winding up as the 2nd worst team ever, after the '93 Mavs. No team has quite hit -15 SRS (point differential + strength of schedule).
So if a zero-win team should ever appear, it would have to be worse than -15; and their avg player would be something below -3.

Others may argue that Manu Ginobili is a "replacement player", since he comes off the bench.
But if you're looking for anything absolute, it's probably the average player who can be found to replace your 12th-15th man.

[bbstats]

C'mon Mike, let's not get ahead of ourselves as North Carolinians...yes we *want* to rebrand back to the Hornets. But not yet

[Crow]

Comparisons to replacement level or league average both have their place. I occasionally compare to championship level.

And with respect to scoring, using some of Evan's data, it doesn't that you need or maybe even want the superstars on the very edge of the individual production frontier. A step or two inside that boundary is more commonly successful as part of a well-functioning team.

[deepak]

I think "replacement level" is mean to capture fringe rotation players or guys who can be picked up very cheaply mid-season as an emergency replacement. If you have a rotation player getting 20 minutes a game and he's goes down with an injury, what impact will that have on your wins if the team replaces him with a replacement-level player?

If you're talking about how good a player is relative to average, that may not be relevant for bench players who are making less than mid-level money. I may have a player on my team that is "below average", and yet losing him will hurt my win total because my depth is 1 less and an even lesser player will have to fill in.

To me, replacement-level its a sensible baseline from which to judge players.

[Crow]

Replacement level RAPM depends on your definition of replacement level. If it is 12th man level, it might be -1.5 to -2 per 100 complete game possessions. If it is the level of the guy who gets called up to take the last job in the league., it might be a bit lower. The very worst guys in the league RAPM are either having an usually bad year or measurement. They probably don't deserve to be in the league if their performance continues at that level long-term. There probably are guys who would come in and hurt less in that situation. Maybe the true replacement level guys don't have upside, but often the guys given that benefit of a doubt don't have much either. Teams sometimes won't admit it for a long time.

Hakeem Warrick has been almost at -4 or worse on prior informed RAPM for every season but one. Phoenix bit on him the season after he "rose" from -7 to to -3. He probably doesn't deserve the role he is given or takes or may not deserve to be in the league at all based on his RAPM history. On 10 year RAPM he is the 4th worst guy measured. But I am sure that some / most boxscore stats say he is average to good and teams apparently continue to believe / use those evaluations until team results cause them to want to get rid of him.