I feel like stepping back for a minute, always a good thing to do at this part of the NBA season, and talk about PER. John Hollinger's Player Efficiency Rating has been an accepted "catch-all" stat for long enough that I feel like we have forgotten how important a stat it actually is.
In short, there is no singular stat that captures how good a player is better than PER. Despite its shortcomings, PER is still vitally important when analyzing a player. Ignoring its monumental importance is one of the easiest ways of making bad roster decisions.
Obviously, you want to have guys on your team with high PERs. Almost always you want to ride the three or four players on your team with the best PERs, if they are capable of playing large roles. The success of a team is based mainly on this, and should not be forgotten.
There are always exceptions, but a player with a PER above 25 is better than a player whose PER hovers around 20 (if they play approximately the same amount of minutes per game.) And that player with a PER around 20 is almost always better than any player whose PER is below 17. This may seem like common sense, but we forget it too often, get too fixated on certain lineups, +/- stats, floor spacing stats, etc. These stats are great, but in general they just point to the truth that PER makes abundantly clear in the first place:
Get three players on your team who play 30+ minutes a night and have PERs of at least 18, hopefully higher. You should be very good.
Last year there were only twenty players who played at least 2,000 minutes and had a PER above 20, according to Basketball-Reference. The year before there was sixteen, and the year before that there were twenty seven. The number of superior NBA players is limited, and there is no way around this.
Players improve and regress, but PER tells us more about a player's value than anything else. It's an objective analysis that usually is very accurate. Short sample sizes and obvious athletic dominance cannot tell us nearly as much about a player as PER can. Nor can defensive and spacings stats. It is always worth remembering this.
When judging a player, the first thing to look at is his PER. Then look at another accurate "catch-all" stat, like Wins Shares Per 48 Minutes, to make sure it ranks the player's value similarly. If it does not, adjust your thinking accordingly. Often "simple" advanced stats like PER, as well as WS/48, are the only things worth really looking at if we want to understand the big picture of the NBA, and see the whole forest instead of only a few trees. Deeper analysis often leads nowhere, although we wish it could. Sometimes the more we know only makes us more easily confused with what is truly important.