WinShare should be super familiar for guys following baseball stats as it is basically the more original version of the now highly utilized WAR (wins above replacement.) basketball reference has a version of it out as well which is different from how it's done in baseball but the general approach is similar.
In short, it's a stats that attempts to quantify the players overall contribution to the team (combining both offense and defense.) to a single number, that number, when you add the whole team up, will equal exactly their total actual wins. It's a daunting challenge obviously, but taking a hard and broad look into the number and you see a considerable deal of consistency in the bigger picture .... some points you may ask.
1. What's the average? : Winshare is base on overall playing time, while WS/48 is more of an average per 48 mins played, .100 is the average for WS/48 by default, a starting player playing roughly 30 some mins a game with a WS/48 of .100 for 82 games (let's call that player Leremy Jin.) will come out to roughly 5.6 wins, which makes sense since the average team = 41 wins, starters playing 30 some min = 60ish % of the total game and divide that by 5 you get 41 x 0.6 x 1/5 = 4.92 , most starter actually play more than 30 min so it's a bit higher than that.
2. is it consistent or predictive? it actually is.... for players in a similar role anyway. For example, you'd think Lebron would have higher WS/48 with the Heat than with the Cavs right? but that's not true, his 3 years in the Heat his WS/48 is exactly the same as his last 3 year average with the Cavs. it's just that the guys around him got much better. Obviously it stands to reason that better teams get more winshare to give, that is true but it seem to mostly benefit quality role players, and espeically good reserve player, for example on the Heat the Birdman has a crazy high WS/48 , but it's obvious that doesn't mean you can plug him into other team as a starter and he'll be a star. so take guys who aren't 30 min + players WS/48 with some grain of salt.
So, here's the main Rocket's player's Winshare and WS /48 this season.
Harden : 12.8 / .208 (he was a super star and Win share totally reflect this.)
Parsons: 7.0 / .125 ( solidly reflecting a well above average player.)
Asik : 5.5 / .108 ( reflects a slightly above average player which is interesting. mostly because it doesn't seem quite as impressed by his D as the eye test. )
Lin: 5.4/ .099 (the definition of average, which can be taken anyway you want.)
Greg Smith: 4.0 / .172 (see what I was talking about in terms of making quality backups look good? having said that, having a super high TS% goes a long way here obviously. )
Delfino: 3.5 / .101 ( definition of average again )
Patterson: 3.0 / .120 (interesting , though WS really hates his defense. but it had him as a pretty darn good player, though one also must take into account he also didn't start all the time..)
Beverly: 2.2 /.148 (a very good reserve, though in this context you would need to think more carefully of how that could potentially translate to full time.)
Morris : 1.5/.081 (just not that good.... )
Douglas : 1.0 / .085 (see above.)
no one else had a WS of more than 1 for the team, though that's mostly due to playing time, Garcia was basically the same as Delfino but with much less time obviously, Jones had the best WS/48 among the rest but also had the fewest minutes.
It's not a perfect stats, but if you take a wide look at different players and what they show up as you'd realize it's probably a lot more consistent than you'd think.
Another interesting point is that Harden had almost 1/4 of the WS, which when you consider that going forward a max contract is generally 1/4 of the cap, and you'll see how it all lines up.