The Lazy Statistician: STATRIBUTES SERIES PART 1 - SHOOTING

[garbageman]

The Lazy Statistician, STATRIBUTES SERIES 1: SHOOTING PERCENTAGES

Welcome back to a new series by the Lazy Statistician. I'd like to introduce to you, the Statributes series. Long have I desired to know how attributes actually correlate with statistics and for more reasons than just being able to use a lazy portmanteau.

For this series, most of the work has gone on in the backend. I took last year's regular season stats (2015-2016) season and last year's attributes (at close of season) and smashed them together in a table so I could see how things relate to each other. I will be using this table for a number of upcoming articles and analyses, maybe even beyond this season. If that is the case, I might still draw from the 2015-2016 season instead of updating with this year's stats and attributes because that was a ton of work for a lazy statistician to do. Don't worry, though. The stats will still be thrown together without an optimal understanding of more advanced statistical methodology.

ANYWAY

In the first series, I'm taking a look at offense. Especially scoring. So as to let you dip your toe in the water with something not super-exciting, here's a look at free throw shooting. I've separated the players into their (past) current letter-grade bands, summing up their free throw makes and attempts and seeing what the average FT% is for people in each letter grade.

+---------+--------+
| ft_band | ft_pct |
+---------+--------+
| A | 0.8643 |
| B | 0.7634 |
| C | 0.6462 |
| D | 0.5269 |
| F | 0.4117 |
+---------+--------+

So that looks like it makes sense. Every letter grade you advance up in fts attribute points, you get loosely 10-12% bump in your free throw shooting percentage.

Next up, we'll look at three point percentage in the same way:

+---------+--------+
| tp_band | tp_pct |
+---------+--------+
| A | 0.4195 |
| B | 0.3626 |
| C | 0.3198 |
| D | 0.2571 |
| F | 0.1278 |
+---------+--------+

That makes a whole lot of sense, too. For every letter grade band you jump, you can expect a high level increase of about 5 percentage points to your three point percentage.

NOW HERE'S WHERE THINGS GET TRICKY

I thought that would follow through to inside scoring. A lot of GMs out there are big on that inside focus, and I've especially been using that with my team this year. But...

+----------+---------+
| ins_band | ins_pct |
+----------+---------+
| A | 0.6385 |
| B | 0.6277 |
| C | 0.6247 |
| D | 0.6333 |
| F | 0.6579 |
+----------+---------+

Alright, so throw out the F band. That's only Theron Shannon and Vernon Norman's combined 76 inside field goals attempted from last season. Everyone else had a higher inside rating.

Still, that's pretty nuts. I broke out my z-score calculator, looking at the sample sizes of inside field goals attempted and made, and the slim margins that the A band holds over the B band are not even statistically significant.

Then, I looked at other attributes to see if they had more of an effect on inside shooting percentage: height, jump, strength, offensive rebounding...nothing.

I went back and tried to set minimums for inside field goals attempted to clear out some of the noise...100, 300, 500...still no dice.

I even made a lazy scatterplot graph to see if visualizations helped identify some kind of trend.

Screen Shot 2017-09-11 at 2.45.11 PM.png (82.83 KiB) Viewed 2439 times

X: Inside FG%, Y: Ins Attribute Rating, Results: Kyrie Irving's perception of Earth

The only thing that came close to setting my inside gameplanning nerves at ease was looking at per 36 minutes PPG.

+----------+---------------+
| ins_band | ppg_p36 |
+----------+---------------+
| A | 26.0982557500 |
| B | 18.6864100364 |
| C | 13.5331577733 |
| D | 11.1409735521 |
| F | 14.3538465000 |
+----------+---------------+

It should be noted that these numbers are pretty consistent if we throw out guys who played less than 12 mpg and even guys who played less than 24 minutes per game. Unfortunately, there are too many factors here to draw a real conclusion. A lot of dudes with A attributes in inside scoring are going to be keyed up or have teams focused on dishing them the ball a lot.

So what does that mean?

Does inside scoring potential matter? Probably, but maybe not as much as the other scoring potentials. Personally, I wouldn't go so far as to trade my inside specialists, but it IS making me think twice about how to spend my points (or how to save them for upcoming tax and insurance instead of making incremental gains to the INS ratings of my young guys who have room to grow).

In any case, I've got a lot of att's 'n' stats mashed together now. I've got per game, shooting, and efficiency stats all loaded up, and I've definitely got a few things I already want to look at next (especially considering athleticism), but if you have any specific things you'd like to see me zero in on, just ask!

[WigNosy]

For kicks, can you try correlating FG% against the Jumpshot attribute and/or the Strength attribute? If possible, excluding three-point shots from that mix would interest me as well (since those are part of FG% but seems clear that they are already controlled by the 3Pt shooting rating).

[garbageman]

Jump shots (not including threes) by JPS rating:
+---------+--------+
| js_band | js_pct |
+---------+--------+
| A | 0.4326 |
| B | 0.3954 |
| C | 0.3679 |
| D | 0.3330 |
| F | 0.2815 |
+---------+--------+

And by STR:
+---------+--------+
| str_band | js_pct |
+---------+--------+
| A | 0.3801 |
| B | 0.3766 |
| C | 0.3771 |
| D | 0.3581 |
+---------+--------+

[garbageman]

So it does look like there is some correlation with the JPS rating.

I didn't dig at all, but the STR rating really surprises me. I'd expect players with really high strength to be bigs with less of a jump shot rating and low strength players to be guards with better shooting.

However, guards are undoubtedly attempting way more shots when bigs could theoretically only be taking high percentage jump shots. Gonna add in some actual numbers here.

+---------+-------+----------+--------+
| str_band | makes | attempts | js_pct |
+---------+-------+----------+--------+
| A | 8469 | 22282 | 0.3801 |
| B | 9631 | 25571 | 0.3766 |
| C | 10111 | 26814 | 0.3771 |
| D | 2223 | 6208 | 0.3581 |
+---------+-------+----------+--------+

And now let's double group (FROM BEST TO WORST):

+----------+----------+-------+----------+--------+
| jps_band | str_band | makes | attempts | js_pct |
+----------+----------+-------+----------+--------+
| A | A | 860 | 1899 | 0.4529 |
| A | B | 1759 | 4096 | 0.4294 |
| A | C | 994 | 2343 | 0.4242 |
| B | A | 2458 | 5825 | 0.4220 |
| B | C | 5431 | 13790 | 0.3938 |
| B | B | 2604 | 6725 | 0.3872 |
| C | A | 2968 | 7873 | 0.3770 |
| B | D | 1431 | 3816 | 0.3750 |
| C | B | 3816 | 10238 | 0.3727 |
| C | C | 2518 | 7018 | 0.3588 |
| D | C | 937 | 2769 | 0.3384 |
| C | D | 553 | 1660 | 0.3331 |
| D | D | 196 | 589 | 0.3328 |
| D | B | 1146 | 3453 | 0.3319 |
| D | A | 1928 | 5822 | 0.3312 |
| A | D | 16 | 50 | 0.3200 |
| F | A | 255 | 863 | 0.2955 |
| F | D | 27 | 93 | 0.2903 |
| F | B | 306 | 1059 | 0.2890 |
| F | C | 231 | 894 | 0.2584 |
+----------+----------+-------+----------+--------+

Let's see how this approach translates to INS

+----------+----------+-------+----------+---------+---------+
| ins_band | str_band | makes | attempts | ins_pct | players |
+----------+----------+-------+----------+---------+---------+
| D | A | 2424 | 3668 | 0.6609 | 24 |
| B | A | 4655 | 7112 | 0.6545 | 15 |
| D | D | 441 | 677 | 0.6514 | 15 |
| C | D | 735 | 1143 | 0.6430 | 15 |
| C | A | 6791 | 10718 | 0.6336 | 35 |
| B | B | 3563 | 5624 | 0.6335 | 15 |
| D | B | 2575 | 4102 | 0.6277 | 37 |
| C | B | 7732 | 12455 | 0.6208 | 49 |
| C | C | 5293 | 8584 | 0.6166 | 54 |
| B | C | 4168 | 6895 | 0.6045 | 22 |
| D | C | 1648 | 2746 | 0.6001 | 26 |
+----------+----------+-------+----------+---------+---------+

For this, I threw out bands where there were less than 10 players, but it looks like strength does indeed play a bigger factor in inside shooting than it looks like when you look at just strength alone. However, at this level of granularity, there are so many other factors for each player that even the bands with 15 players can be influenced by a few overperformers or underperformers.

[WigNosy]

Are you looking just at Jump shots or at inside shots too?

I would expect JPS to correlate with jump shot percentage and STR to correlate with inside shot percentage.

[ballsohard]

I think inside has always been mislabeled , if you read the manual it basically says ; inside means the ability to score.. which correlates with your info.

[garbageman]

Are you looking just at Jump shots or at inside shots too?

I would expect JPS to correlate with jump shot percentage and STR to correlate with inside shot percentage.

Looking at jump shots on their own (separate from inside or threes) using the appropriate categories from the shooting category in the on page stats.

[garbageman]

I think inside has always been mislabeled , if you read the manual it basically says ; inside means the ability to score.. which correlates with your info.

Anywhere I can find this manual?

[garbageman]

And just for some more followup...here's jump shot percentage by inside band (ignore the F band...small sample size) and three point percentage by inside band to add more to Scott's comments about general scoring:

+----------+-------+----------+--------+
| ins_band | makes | attempts | js_pct |
+----------+-------+----------+--------+
| A | 3743 | 9430 | 0.3969 |
| F | 85 | 219 | 0.3881 |
| B | 7896 | 20473 | 0.3857 |
| C | 12864 | 34543 | 0.3724 |
| D | 5846 | 16210 | 0.3606 |
+----------+-------+----------+--------+


+----------+-------+----------+--------+
| ins_band | makes | attempts | tp_pct |
+----------+-------+----------+--------+
| A | 1765 | 4623 | 0.3818 |
| B | 5789 | 15172 | 0.3816 |
| C | 7312 | 20848 | 0.3507 |
| D | 3799 | 10985 | 0.3458 |
| F | 54 | 187 | 0.2888 |
+----------+-------+----------+--------+

[false9]

Good stuff, Josh!

[kucoach7]

If you ever want some data, HMU.

[garbageman]

If you ever want some data, HMU.

Just noticed you can dl reports from the Tools section of TV

[ballsohard]

I think inside has always been mislabeled , if you read the manual it basically says ; inside means the ability to score.. which correlates with your info.

Anywhere I can find this manual?

Should be on the game somewhere in the node.

[kucoach7]

If you ever want some data, HMU.

Just noticed you can dl reports from the Tools section of TV

Interesting. I've never poked around in there. If you can't get historical data on team or player stats/atts for past seasons, I have them all compiled.

[Conroy]

[Inner_GI]

You can claim 5 points.