RE: 3 point shot
This is a topic I've been meaning to write about for a while now. Since I'm shut in I've got some time so here it goes... this is going to be long but hey, there's no sports so hopefully someone enjoys this.
The recent trend in basketball all around the country is going with the Analytics approach of prioritizing threes and layups because they yield the highest points per attempt compared to mid range jumpers which yield the lowest. This trend started in the NBA with teams like Houston fully embracing it, and has trickled down through college and we're seeing it in the MAC. So to start, I'll say I'm a big time believe in analytics. However, I think we haven't been seeing this strategy applied properly in the MAC and that's part of why a lot of fans feel that, while the league is very competitive, the games are often frustrating and unenjoyable to watch.
Let's start by looking at the points per shot on 2 point attempts vs. 3 point attempts. Free throws also need to be incorporated into this of course. That's the tough part about this type of study. Just saying "if you can shoot 33% from three that's as good as shooting 50% from two" is dangerously inaccurate because 2 point attempts yield more free throws so the true points per 2-point attempt needs to factor in free throws. To start I break it into four categories. Points per 3 point attempt (3*3PM)/3PA. Points per 2 point attempt (2*2PM)/2PA. Points per FT tuple (2 * FTM * Conversion Factor)/(Conversion Factor * FTA). So this would be an estimate of 2 point shooting fouls and on the floor fouls in the bonus. The Conversion Factor is a multiplier to convert number of free throws to number of possessions. For the NBA the accepted factor is 0.44. For college it is 0.475. Finally the 4th category is the points off free throws that doesn't fall into the FT tuples. These are assumed to be and-ones.
So the trick is to take the last three categories and come up with a "true" points per 2 point possession. First calculating the points scored. 2 PT FGM + And-Ones obviously all count here. And some portion of the FT tuples need to count here as well. For this I'm just going to say an arbitrary 90% of these count. This is just an educated guess to try to get "close". Crediting all of these doesn't feel right because we know fouls occur outside the arc especially at the end of games. Calculating the possessions I'll use the same logic. 2 pt FGA + 90% of the FT tuples. Now let's look at the data for the NBA because that's where these trends started.
The splits below are Points Per 3 Pt Pos / True Points per 2 Pt Pos / Actual Points Per Pos.
Starting with the NBA for this season the league averages look like this:
* 1.072 / 1.150 / 1.127
These are the top 5 teams in three point attempts this year:
* HOU: 1.045 / 1.236 / 1.157
* DAL: 1.108 / 1.195 / 1.162
* MIN: 1.007 / 1.155 / 1.102
* MIL: 1.067 / 1.221 / 1.166
* BKN: 1.020 / 1.143 / 1.100
Average of the top 5:
1.051 / 1.189 / 1.138
These are the bottom 5 teams in three point attempts this year: NOTE: The New York Knicks were actually 3rd fewest here, but their numbers are so abysmal across the board they drag the average down significantly. So I removed them because they are a dumpster fire that clouds the results. In statistics talk, I'm removing an outlier. In real talk, the Knicks are an outlier in terms of being an actually NBA franchise.
* LAL: 1.065 / 1.193 / 1.159
* IND: 1.088 / 1.141 / 1.130
* SAS: 1.113 / 1.140 / 1.137
* OKC: 1.064 / 1.186 / 1.153
* DEN: 1.073 / 1.147 / 1.128
Average of the bottom 5:
1.080 / 1.161 / 1.140
Looking at spots 13-18 in three point attempts (the middle) it looks like this. (SAC, UTA, BOS, DET, LAC, WAS)
1.108 / 1.150 / 1.140
So my takeaways are these:
Teams that shoot the most threes:
1. Significantly lower yield per three pointer than league average.
2. Significantly higher yield per two than league average.
3. Overall, score more per possession than league average. Three of the 5 teams were over league average at least.
4. So while the increase in 3 point shooting is lowering the teams yield per three, in general the teams are more efficient than league average.
Teams that shoot the fewest threes:
1. Slightly higher yield per three pointer than league average.
2. Actually a higher yield per two pointer than league average as well.
3. Overall, they score significantly more per possession than league average and slightly more than the top 5. EVERY single team here scored more per possession than league average.
4. So they shoot less threes and across the board put up better efficiency than league average.
Teams that shoot an average amount of threes:
1. Significantly higher yield per three pointer than league average.
2. Exactly the same yield on two pointer as league average.
3. Overall, score significantly more points per possession than league average. The same score as the teams shooting the fewest threes.
4. So they are the most balanced teams in the league and they're offense is performing every bit as efficiently as the teams that are most imbalanced one way or the other.
Of the 16 teams that fall into these categories 11 would be in the playoffs right now.
4 of the 5 that shoot the fewest threes would be in.
4 of the 5 that shoot the most threes would be in.
3 of the 6 balanced teams would be in.
So I guess I would say the conclusion I draw from this is simple. You don't need to shoot a lot of threes to be successful. Rather you need to maximize your points per possession. If you build your team to shoot threes, then go all in and shoot threes you'll be successful. If your team isn't built on strong three point shooting, then you need to identify this (analytics can help) and go strong the other way, and you can still be successful and it can actually help you be better when you do shoot threes.
So what does this mean for the MAC? Well, in my opinion the influx of analytics and three point shooting in the NBA is why we're seeing so much three point shooting in the MAC. The problem is MAC teams don't have a team of analytics professionals telling them how to run their team. So what you get is teams that shoot more threes than their three point shooting ability warrants.
Here's the number of threes the average MAC team attempts per conference game by year, as you can see three point shooting is clearly on the rise across the MAC.
* 19/20 - 22.75
* 18/19 - 22.26
* 17/18 - 22.20
* 16/17 - 21.04
* 15/16 - 20.40
* 14/15 - 18.33
* 13/14 - 18.15
* 12/13 - 17.69
So now, let's come back and just focus on Kent last year. I have all the play logs in a SQL DB so I went ahead and ran some queries to see how Kent scored per possession, on the different types of shot. The playlogs categorize shots as DUNK, LAYUP, JUMPER, and 3PTR.
In conference games...
Kent shot 59.8% on Dunks & Layups which comes to 1.197 points per possession.
Kent shot 43.7% on 2 point jumpers which comes to 0.873 points per possession.
Kent shot 34.2% on 3 point jumpers which comes to 1.025 points per possession.
So now, if we put 2 point jumpers, dunks, and layups together, Kent averaged 1.054 points per possession on two's, which is significantly better than their yield on three pointers.
In total Kent attempted 439 threes vs. 681 twos. So 39.2% of Kent's shots were three pointers. Think about that... Kent was scoring more efficiently on two's than threes but they still attempted nearly 40% of their shots from three. And this is WITHOUT even factoring in the bump two pointers get when you include free throws. So very clearly, Kent shot way too many threes this year. WAY too many. To put that in perspective, the average NBA team shoots from three 38.2% of the time. So we shoot threes at a rate higher than the average NBA team despite the importance placed on three point shooting skyrocketing in the NBA in recent years.
So is this a Kent problem, or a MAC problem? In conference games the MAC averaged 39.1% of their shot attempts coming from three this year. Nearly identical to the rate that Kent fired them up.
Looking at individuals here's the top 5 in threes attempted in conference play:
* Simons (141 attempts): 1.128 points per three. So Simons efficiency was phenomenal especially for the volume. He deserves the neon green light he had.
* Pippen (95 attempts): 0.947 points per three. This was a massive drag on our team's offensive efficiency.
* Roberts (75 attempts): 1.120 points per three. Also phenomenal. Would've been interesting to see if he could keep that up if we increased his volume.
* Williams (46 attempts): 0.978 points per three. Not very good, but for someone who played as much as Williams I'm content with this number of attempts. He's an elite slasher and needs to keep them honest.
* Williamson (37 attempts): 0.973 points per three. Not an extremely high volume so tough to get on him too much.
I know there's a lot here to digest, but I think the conclusions are this. The MAC as a whole is trying to buy into the "analytical approach" of shooting more threes because threes are more efficient than mid range shots. MAC teams don't get the top prospects so we're not loaded with guys that can hit volume threes at an efficient rate. There's more of a premium on three point shooting than ever, so that's probably hurting the MAC in terms of how good of shooters we're getting too. But we're still shooting it at an insanely high rate. No team should EVER average more points per 2 point attempts than 3 point attempts WITHOUT FACTORING FREE THROWS into the 2 point attempts. But we did that this year. And we still shot 39.2% of our shots from three.
So I guess to sum it all up. I'm a big believer in analytics. You should use analytics on the national level to determine what direction you want to go when building your team. Then you should use analytics specifically on your team to help determine how to game plan with what you have. I fear what's happening is MAC coaches see the national analytics and are trying to apply them to their game planing which DOES NOT WORK, period. Once you have a team of 8-10 guys the national numbers don't matter. It only matters what those 8-10 guys are capable of. So national analytics may tell you that a three pointer is more efficient than a two pointer but that's not true if that three is being taken by a 31.6% three point shooter. So that shooter shouldn't have a license to shoot 95 threes in conference.
|