Post by Andrei Kirilenko on Jan 23, 2018 14:18:40 GMT
D5 Analytics
#3 - Score Area
As I mentioned in my previous post, I wanted to dive deeper into some of the correlations I found and introduce a new element into my study: Score Area. As some background, each player in NBA Live is assigned a certain “Score Area”, which I like to think of as offensive playstyle. There is this reference guide available online for people similar to us who like to spend their time editing roster files that are now 12 years outdated. The link above provides the following reference chart, which is what Walt utilizes in assigning Score Area to players when doing rating updates:
A little outdated, huh?
The purpose of this post is to try to gain a better understanding of how Score Area impacts actual performance in-game. I will try to be brief in my approach but am happy to answer any questions regarding the math. To start, I took updated player ratings and updated player statistics and combined them all into one big worksheet. Ian was kind enough to provide me with a worksheet containing the Score Area for each player, which I incorporated into my larger data set. Then, since I am only examining offense in this study, I removed any player that averages less than 20 minutes per game and any player who average less than 10 ppg. I also removed any player who has played less than 10 games. This left me with a data set of 113 D5 players.
Before I get into the results, I want to make a caveat and mention that I do not really think this is a big enough sample size. There are 7 different Score Areas, meaning the average data set for each Score Area is only 16 players. In fact, Score Area 2 only contained 7 players, which isn’t really enough to draw the type of conclusions I want to make. However, some Score Areas, such as area 7, contained up to 28 players, which is probably a little better sample size.
Without further ado, here is what I found:
Score Area 2:
As mentioned above, only 7 players were in Score Area 2 so I am not going to draw many conclusions. The reference guide claims that Score Area 2 is mostly perimeter players, and gives examples of two point guards. In D5, we have included in this Score Area players such as Evan Fournier, Eric Gordon, JJ Redick, Buddy Hield, and Malik Monk. All of these guys really suck in our sim, and in fact the average field goal % for Score Area 2 is just 41.9%. Everyone in Score Area 2 has a FG rating in the 70s, and the average 3p rating is an 82. These guys actually shoot decent from deep (38.8% average), but as mentioned above, the data set was too small to draw any good conclusions.
Score Area 4:
Score Area 4 is supposedly Centers and Forwards with a midrange game. These guys are the most dominant group in D5, with an average FG% of 46.2%, higher than even the “inside players” group (Score Area 9). And that is because the correlations actually suggest that Score Area 4 should be the Inside players and Score Area 9 is actually mid-range players. FG rating is only a 30% correlation with FG% for players in Area 4, yet it is a ridiculous 87% correlation for players in Area 9. The Inside rating is a whopping 70% correlation in Area 4, yet only 3% correlated in Area 9. These groups are entirely reversed in the reference table!
The stats back this up as well. Nene, despite being in the “mid range” group but only having a 56 FG rating, manages to put up 13 ppg in D5 on 53.7% shootin, because his Inside scoring is rated an 89. On the flip side, a guy like Dario Saric, also in Area 4, shoots only 37.6% from the field, despite having one of the highest Area 4 FG ratings of 74. Saric is only rated a 61 for inside scoring. Every single player in this group with an Inside rating above an 80 is absolutely dominating in D5. Every player below a 70 Inside rating is putting up horrible statistics.
Score Area 9:
I am going to jump to Area 9 since it was discussed in conjunction with Area 4 above. This group is supposed to be Post-Up players, and includes the likes of Andre Drummond, Hassan Whiteside, and DeAndre Jordan. Almost every player in this group has a good Inside rating (average 81), yet these guys are only the third best group FG%-wise in D5. That’s because the Inside rating doesn’t appear to really matter much for Area 9. It only holds a 3% correlation with FG%. Clint Capela has a 93 inside rating but shoots 41.8% from the field. Dwight Howard has an 80 inside rating but shoots 41% from the field. Andre Drummond is shooting 37.5%! These guys all have between a 17 and 35 FG rating.
On the flip side, the best players in this group, despite having lower overall ratings, are guys who have decent FG ratings. The best FG ratings in the group are Jahlil Okafor (72 FG, 47.2% shooting, 16.3ppg), Jakob Poetl (70 FG, 48.6% shooting, 13.3ppg), and Cristiano Felicio (67 FG, 46.5% shooting, 19ppg). These scrubs are dominating D5 despite being scrubs. I think this group needs a complete overhaul.
Score Area 10:
I am going to move on to Area 10 since we are talking about big men. This is supposedly bigs who can shoot, and contains players like Towns, Jokic, Embiid, Cousins, and Love. FG% for this group had the highest correlation with the Inside rating of any group (80%), but was also pretty highly correlated with FG rating (66%). That means that if you add FG and Ins together, you can figure out who is going to be the best player. Every player in this group is actually pretty dominant, so instead of looking at the group as a whole I decided to focus on a few specific players who are exceptionally good or bad.
Kevin Love is the worst statistical player in this group, shooting 38.8% from the field. This is because his Inside is only a 65, while his FG is only a 72 (total 137). I’m not sure why his ratings suck so much, but he needs an update. Joel Embiid actually sorta sucks too, shooting only 42%. His Inside is only a 72 and his FG is only a 70 (total 142).
The two potentially most dominant players in D5 (who happen to be on the same team) are Towns and Jokic. Towns has 80 FG, 84 Inside (total 164) and averages 49.8% shooting for 29.2 ppg. Jokic has 85 FG, 85 Inside (total 170) and averages 49.6% shooting for 24.4 ppg. Enes Kanter also dominates, putting up a ridiculous 53% shooting. He has 74 FG and 88 Inside (162 total).
Score Areas 6 and 7:
I combined these two groups for analysis because they are pretty similar. 6 is supposedly slashes/shooters and 7 is supposedly multi-talented wings. I found that Area 6 is more highly correlated with the FG rating, whereas Area 7 is more highly correlated with the Inside rating. However, none of the correlations stick out too much as unusual.
My main conclusion here is that it sorta sucks to be a wing player in D5. These groups have really low FG% almost unanimously. I think this is because the Inside rating is particularly low across the board for both of these groups. Goran Dragic (33 inside) is shooting 32.6% from the field. Kemba Walker (25 inside) is shooting 35.8% from the field. Jeff Teague (39 inside) shoots 37.2%.
If you want to have good wings on your team, you have to find a guy with a decent Inside rating. If I wanted to use this study to my advantage, I would go trade for Jae Crowder (nice defensive ratings and a 90 Inside rating, shooting 44.8% from the field), TJ Warren (82 inside rating, 44.4% from the field), or Harrison Barnes (81 inside rating, 50.4% from the field). These are some of the most efficient wings in the sim, and provide the most "bang for your buck".
One last thing I will point out for Area 6: this was the only group with high correlations between 3p rating and 3p% (70% correlated). I don’t have any conclusions from this except that you probably want your shooters to be in Area 6.
Score Area 5:
This last group is perimeter players who shoot midrange and threes. This includes Devin Booker, Steph Curry, Klay Thompson, and CJ McCollum. This group seems to be decently correlated with both FG and Inside, and like 6 and 7 above, players who have received a low Inside rating struggle. Steph Curry all but maxed out in shooting ratings but has just a 45 Inside, causing him to shoot 41.2%. No one in this group really shoots a high percentage, but Klay Thompson is one of the better players at 47.5%. Klay has a 86 FG, 95 3p, and 69 Inside.
As mentioned above, Area 6 actually correlates better with 3p shooting than does Area 5. I don't know if that means we should move these shooters to Area 6, but it is an interesting observation.
Final Conclusions:
Wing groups (2, 5, 6, 7): Inside rating needs to be examined more carefully because it causes wings to shoot artificially low percentages.
Big men groups (4, 9, 10): 4 and 9 are potentially backwards, and the FG rating needs to be examined more carefully for all groups because it causes big men to shoot artificially low percentages.