FORECASTING IDEAL 3-AND-D PLAYERS: PART TWO

In the first part of this short series, a step-wise linear regression approach was used to forecast this year’s draft participants’ 3PT catch-and-shoot ability (revisit the first part here: https://baselinejam.blog/2019/05/16/forecasting-ideal-3-and-d-players-part-one/ to see an updated table of nearly all wing players). In the second part, we will forecast draftees’ defensive ability by forecasting their defensive rating. I will also provide a couple of suggestions for the Thunder to draft with the 21^st pick. Then, in the final part, we will examine hidden 3-and-D treasures in the draft.

Step-wise Regression for Defensive Rating:

A similar regression approach was taken for the same sample size from our first part. I won’t be as technical in explaining step-wise regression in this post, but refer to the first part if you have any questions about the methodology taken to create the model. Using the same sample size, the independent variables that were examined were as such:

• Body Fat (%)
• Hand Length (inches)
• Hand Width (inches)
• Standing Reach (inches)
• Weight (lbs)
• Lane Agility (seconds)
• Shuttle Run (seconds)
• Height (inches)
• Wingspan (inches)
• College Career Average in Defensive Win Shares
• College Career Average in Defensive Box Plus/Minus
• College Career Average in Steals
• College Career Average in Blocks
• Years in College
• Three Quarter Sprint (seconds)
• Standing Vertical Leap (inches)
• Max Vertical Leap (inches)

The dependent variable (the variable that will be forecasted) was chosen as defensive rating. It’s well known in the basketball analytics world that there isn’t a great all-encompassing defensive metric just yet. Defensive rating can be briefly explained as how many points an individual player gives up per 100 possessions. Using this in this series as a defensive metric is sufficient but there are glaring issues with defensive rating. For example, it doesn’t take into account variables like matchups. A great example of this is Josh Okogie. He is perhaps one of the best defenders in the game in my opinion, and he was only a rookie this year. Yet, his defensive rating was relatively high at nearly 110. A huge reason for that is that he was the primary defender and had to guard the opposing team’s best players. However, the largest issue with defensive rating is that it doesn’t take into account for teammates’ defensive abilities and that can artificially raise or lower a player’s defensive rating drastically. Once again, for the purpose of this study, I think defensive rating is a sufficient metric to use but approach this content with caution.

That said, I ran multiple iterations to derive the below model, ensuring that all my independent variables satisfied the condition of having an alpha level of 10% or less (p-values < 0.1). This model had a much higher R-squared value than the 3PT catch-and-shoot model as its R-squared is 47.1%. A draftee’s career average defensive rating can be forecasted as such:

Interpreting the Model:

First and foremost, a lower average defensive rating is good (remember defensive rating is simply how many points a player allows per 100 possessions). Mathematically, this means that independent variables that have positive signs in front of their coefficients are bad in the sense that an increasing value in that independent variable adds to a player’s average defensive rating. For example, height (in inches) has a positive sign in front of its coefficient; so, as a player increases in his height, the worse his average defensive rating is. Conversely, this means that independent variables that have negative signs in front of their coefficients are good in the sense that an increasing value in that independent variable decreases a player’s average defensive rating. Once again, let’s look at an example: standing vertical leap (in inches). It has a negative sign in front of its coefficient; so, as a player’s standing vertical leap increases, the lower that player’s defensive rating is, signifying better defensive capabilities.

Let’s take a deeper dive into each independent variable:

• Standing reach (inches)—as a player has a longer standing reach, the lower his average defensive rating is. This makes sense as a longer standing reach allows the player to contest shots better.
• Lane Agility (seconds); Shuttle Run (seconds); Three Quarter Sprint (seconds)—these three seem counter-intuitive. Each of these variables have a negative sign in front of their coefficients, indicating that longer lane agility, shuttle run, and three quarter sprint times are actually better for average defensive rating. To understand why, one has to understand how these drills are conducted (see this video to understand how the lane agility is tested: https://www.youtube.com/watch?v=CJc7HOJGJvY ). Lane agility is supposed to measure a player’s lateral quickness; however, half of the drill is running up and down the length of the free-throw paint. Thus, players with longer strides will tend to have faster times. It makes sense that longer strides are actually worse for defense. Individual man-to-man defense is predominantly about reaction to the offensive player’s moves. Longer strides make it harder for a player to react. Moreover, a quick crossover and change in direction from an offensive player means that the defensive player with a longer stride typically has more ground to make up after making the initial defensive-slide. Same logic can be applied to the shuttle run and three quarter sprint drills. Players with longer strides will tend to have faster times and those longer strides are usually worse for reactionary defensive maneuvers.
• Height (inches)—as mentioned earlier, the taller the player, the worse his defense. Wait a minute: how does this make sense when taller standing reach provides better defensive rating values? My theory is a further extension on the previous point. The taller the player, the longer the stride. Standing reach is seen as a positive trait as it encapsulates a player’s wingspan.
• Years in College—The longer the player played in college, the better his defensive rating. This makes intuitive sense as that player builds more basketball IQ.
• Standing Vertical Leap (inches)—The higher a player can jump, the better his defensive rating. This also makes intuitive sense on two levels. First, a higher standing vertical leap means better shot contesting abilities and shot blocking abilities. Secondly, this can be a measurement for athleticism. The more athletic the player, the better he is on defense.

Forecasting 2019 Draftee’s Average Defensive Rating:

Now the fun part! Let’s take a look at draftee’s forecasted defensive rating. BIG NOTE: some participants elected to not partake in the strength and agility portion of the combine, so unfortunately some players’ forecasts are not available. From the available data, here’s the list of forecasted defensive rating in decreasing order (lower defensive rating values = good).

Player	College	Forecasted NBA Avg Defensive Rating
Miye Oni	Yale	101.1
Nassir Little	University of North Carolina	101.4
Shamorie Ponds	St. Johns	101.5
Reggie Perry	Mississippi State	101.9
Eric Paschall	Fordham/Villanova	102.1
Marial Shayok	Iowa State	102.1
Talen Horton-Tucker	Iowa State	102.3
Charles Matthews	University of Kentucky/Michigan	102.9
Oshea Brissett	Syracuse	103.2
Terance Mann	Florida State	103.4
Quinndary Weatherspoon	Mississippi State	103.4
Jared Harper	Jared Harper	103.6
Zach Norvell Jr.	Gonzaga	103.9
Ky Bowman	Boston College	104.7
Jalen McDaniels	San Diego State	104.7
Admiral Schofield	University of Tennessee	104.8
Quentin Grimes	Kansas	105
Grant Williams	University of Tennessee	106
KZ Okpala	Stanford	106.2
Luguentz Dort	Arizona State	106.2
Tremont Waters	LSU	106.4
Cody Martin	NC State/University of Nevada	106.6
Ty Jerome	University of Virginia	106.6
Dylan Windler	Belmont	106.6
Ignas Brazdeikis	University of Michigan	106.8
Kyle Guy	University of Virginia	106.9
Jaylen Hands	UCLA	107
Louis King	University of Oregon	107
Cameron Johnson	University of North Carolina	107.2
Brandon Clarke	Gonzaga	107.3
Carsen Edwards	Purdue	107.8
Jaylen Nowell	University of Washington	107.9
Kevin Porter Jr	USC	108.2
Kris Wilkes	UCLA	108.8
Jordan Poole	University of Michigan	109.2
Devon Dotson	Kansas	110.5
Jordan Bone	University of Tennessee	110.8

 

Combining it all for Thunder’s First Round Pick:

There are many rumors circulating that the Thunder promised Matisse Thybulle that the organization would draft him. I love this selection. He’s forecasted as a 34.5% 3PT catch-and-shoot shooter, which is arguably his glaring weakness. He elected to not participate in the strength and agility portion of the combine, so a forecast of his defensive rating isn’t available. However, his college résumé speaks for itself. He played all four years, insinuating that his basketball IQ is relatively mature. He averaged 3.5 (!!) steals in this past season and averaged 2.3 (!!!) blocks. That stat line is absolutely unheard of.

Here’s the beauty of the draft range that the Thunder is in: each potential player has a high floor. I truly believe that only Matisse has a high ceiling, but there are a few candidates that can immediately contribute off of the bench. That said, let’s look at those other candidates at that 21^st pick spot.

Grant Williams: Personally, I would like a Grant Williams pick. I can see him playing all four positions from point guard (1) to power forward (4), which is crucial. His forecasted 3PT catch-and-shoot percentage is 34.5%, same as Matisse Thybulle’s. However, his defensive rating is forecasted to be rather average at 106. He’s not overly athletic, but I believe he can be a Draymond Green-esque player (others also believe this comparison is appropriate: https://www.thestepien.com/2019/05/13/draft-notes-grant-williams-player-interest-golden-state/ ). I know the Thunder fanbase isn’t a huge fan of Dray, but we can’t deny that Draymond is a jack of all trades, a defensive guru, and an extremely high IQ player.

First, Grant has some amazing active hands from both the post and the perimeter. Watch the two clips below as examples:

Not only that, but Grant has often served as Tennessee’s anchor on defense this season and his help defense instinct is pretty good as shown below. Once again, this could be great as the Thunder can often utilize him in small ball lineups.

Shifting gears to Grant Williams’ offense.

Let’s continue with the Draymond comparisons. Grant has high basketball IQ—he seems to always make the right plays as he is able to dissect the defense by understanding where and when the double is coming. I understand that he will probably not be doubled in the NBA as much, but the point is that he has high playmaking ability. Watch the two clips below. The first one is particularly impressive as he’s able to read the flare screen and hit the open three-point shooter. As for the second clip: instead of kicking it back out to the wing upon the double as most college post players would do, Grant stays patient and notices that the weak-side defender’s eyes are focused on Grant. He makes the pass to the back-door cut for an easy layup.

The biggest issue with Grant seems to be his lack of confidence in shooting the deep ball. Grant can easily be a great pick-and-pop player but watch the below clips as he often exhibits hesitation and lack of confidence on wide-open 3PT shots. Unfortunately, confidence is something that you can’t teach.

Cameron Johnson: Aside from Tyler Herro, Cameron Johnson is probably the best pure catch-and-shoot shooter in this draft. His shot mechanics are fundamental and he’s always ready to catch and shoot, often shaving micro-seconds off of his release. Check out his quick release in the clip below.

 

Not only does he have a quick release, but Cameron also displayed some ability to have a quick release from longer-distance 3pt shots as shown below.

So that all seems good, but defensively he seems to be a major liability. He is shown to be very slow laterally and his defensive stance is hardly fundamental as he’s completely standing upright. Instead of showing GIFs, check out this very instructional video from Draft Express: https://www.youtube.com/watch?v=lbFltAehSUs

I want to say that a suitable Thunder comparison for Cameron Johnson would be an Alex Abrines-type player. That’s honestly not all that bad as the Thunder needs extra shooting off of the bench, and it’s much easier to mask sub-par defensive players by making them come off of the bench.

These are just two of my personal favorites, but readers can examine both the first and second part to come to their own conclusions.

I hope you all enjoyed the first two parts of the blog series. In my last part, I will try to identify some hidden gems in this year’s draft. As always, thank you so much for taking time to read the blog! Also, if you have any additional questions, I’m always happy to chat. Send me an email:

KangJ20@darden.virginia.edu