Reply With Quote
O/D WS/48, O/D BPM
I don't think anybody just goes to 1 advanced stat as "the" advanced stat to get a picture of a player's effectiveness, and surely people look through a few to come up with a picture.
Curious what people think are more or less useful ones. We've come a long way since the days of PER being the one "advanced stat."
Just due to ease of access, the ones I find myself looking at are those available of basketball reference (defensive/offensive rating on-off figures, win shares/48, BPM, VORP), and the RPM stats published on ESPN. the RAPTOR numbers on 538 are another one i can think of that i generally look through
to be frank, i have no clue which of these are thought to be more credible or meaningful than the other
O/D WS/48, O/D BPM
VORP/RAPTOR
Also Win Shares has never lied to me.
WS is a quick and dirty, but if you want to compare players, WS/48 is more useful.
Thanks for this thread - would love to understand advanced stats better. Can ya'll explain why you like the ones you do?
From the early Box score variations, we've generally moved into using variations of lineup adjusted on-off (RAPM, RPM, etc), often with a box score informed prior (essentially, assumes that if two guys share the court all the time, the one driving the good performance is the one putting up 20, 5 and 5 rather than the 6ppg role player) to minimise the error. RAPM does a decent job of that. RAPTOR was good, though 538 is stopping updating it. I generally use the ESPN RPM first for ease of finding things.
After those catch all totals, you get into looking at particular skills that are valuable. For rim protection / centres, look at number of shots contested and field goal percentage allowed at the rim. For off ball players, look at 3pt %, and break it down by standstill / catch and shoot, off the dribble, and off screens. PGs, Assist to TO% etc, points per possession on the pick and rolls they run, etc. Varies by role.
Overall, generally go through a progression of RPM / Raptor etc for "Is this player good" then to the eye test / more detailed stats on why.
E.g, If a big with pedestrian numbers pops in +/- stats:
- Are they good at boxing out for teammates (Lopez special), or
- do they set awesome, intelligent screens (flipping the angle, good timing, etc - I think this is why Nick Collison always looked awesome) or
- Is their floor spacing really valuable for the space it opens up (Matt Bonner - who was a huge + by RPM type stats).
You can do similar things on the value of off ball movement / cutting, defense (on ball, and how well people roam off ball to help), etc.
For scouting / draft evaluation, you're trying to guess development - how good is a player likely be be, rather than how good are they now.
Generally, use FT% as a proxy for eventual 3pt% development, as you have a high rate of attempts and that's less prone to statistical noise. Steal % works well as a proxy for athleticism / feel for using it (There'll be gamblers, so it's imperfect). Rebounding translates. Then, often assume an aging curve, so 19 year olds with decent numbers are likely to become better than 23 year olds with slightly better numbers now.
the NBA is cool because they do a lot of geometric vectoring and plotting. A big part of a players offensive game is based on a player finding his spots so he can practice. Those gradient charts metered on shooting percentage are useful.
I also imagine that vector analysis would be very helpful for finding angles of attack and where to defend.
I found timvp and his use of analyzing players by lineup to be interesting in finding best lineup.
winshares and similar stats that come from the approach of correlating to wins are useful for ranking players over an interval based on wins, I don't see them as fungible from interval to interval so they are not very useful in roster building or predictions. player contract efficiency and rankings are what they are good for.
New Silver Spurs Dance Team vitals.
Win/loss column
PPG
Another thing with Bonner is that he took pretty good care of the ball he knew he wasn't a guy who was beating people off the dribble so you didn't see him put it on the floor in traffic and he knew to hit an open teammate when the defense closed out on him.
Hitting your teammates in their spots and passing to an open man or the space your teammate is cutting to is underrated.
Exactly. No crap players ever averaged 27 ppg
Two words: Carmelo Anthony
Purvis short.
11 time all star, 19 seasons
yeah dude sucked
12 seasons in the league. yep horrible
He is cancer personified that made Iverson look like a team player.
Sure, he was popular but he was a crap player.
Height, weight, bust, waist, hips, cup
Definitely better than magic and Pippen. Who never averaged 27 ppg. Definitely ppg is the way to use to judge players. Youve convinced me.
Arm length
Pit strength.
It's underrated!
Dude was empty calories, and rarely made his team better unfortunately. Others I’d put in this group are LaMarcus, Amare, and possibly Harden when it’s all said and done.
All stats generally have some sort of weaknesses and advanced stats are no exception. In general, advanced stats such as WS, WS/48, RAPTOR, VORP are all affected by the quality of your teammates. If you have a system that you fit well in, then your advanced stats will shine. Case in point being Chauncey Billups, was he really the 4th best player in the NBA in 07-08 season, better than Lebron, Duncan? Was Manu really better than Duncan that year? I wouldn't say so but WS/48, VORP and BPM, when taken together as a whole, does point in the right direction.
BPM, aside from the quality of your teammates and how well you fit within the team, also depends on your sub and again how the team was built. Let's say Curry, the Warriors built their offence around his and Thompson's ungodly outside shooting, and when one of them sits, it is hard to get another player to duplicate that effect, which gives them a large BPM (OBPM in this case). if a team splurges on a strong starting lineup and got a crap bench, naturally your BPM is going to be huge as well as your replacement is crap, so these are factors that are outside of the player's control and doesn't say one player is better than the other.
There also is the sense of whether you ARE the system, and how easy it is to build that system around you? Players like Duncan, Jokic, Shaq, Lebron, Hakeem and Giannis are relatively easy to build around as the complementary pieces to help them win lots of game are comparatively easy to find, whereas guys like Dirk, the Admiral, and Barkley are tougher to build around because of the weaknesses you have to cover. I really haven't found a stat that measures this yet.
I find PER to be just like a weighted system for traditional stats and isn't overly insightful, especially when comparing across different seasons. Differences in pace and styles of different eras produce different results. I felt the low scoring 90s and early-mid 00s have generally deflated PERs, and today's league and those in the 60's have inflated PERs due to the pace and also the heliocentric style of play. You take a look at the best PER seasons ever, and of the top 20 seasons, 13 of them were either in the 60s or over the last 10 years, the other 7 were all Jordan and Lebron's who are widely known as the two best players ever. Whereas when you look at WS, WS/48, VORP and BPM stats, the best seasons tend to spread across multiple eras (still favouring recent era and those in the 60s, just not as prominently).
To answer the question, you just really have to look at them all, and I find that WS/48 and BPM compliments each other quite well, while RAPTOR and VORP have some level of overlap. You take 3 of these together and it tells the story pretty consistently, I would say. For example, last year, Jokic, Embiid were in the top 3 in WS/48, VORP and BPM, while Luka was top 3 in VORP and BPM (Luka was 11th in WS/48 because the Mavs sucked).
There are currently 1 users browsing this thread. (0 members and 1 guests)
Posting Permissions
