TL;DR version - don't just look at free-throw totals and cry. Arm yourself with the facts. Harden isn't an outlier, and even a super rudimentary statistical model proves it in multiple ways. I've been embroiled in Harden free-throw controversies in conversations with co-workers and social media friends recently. I have THREE people (Mavs fan, Warriors fan.... European fan?) sending me private messages about it (c'mon man!). Living in the crucible of Utah has forged me into a reluctant Harden defender, so I decided to do a quick stats exercise (a simple linear regression analysis of Usage vs Free Throws Attempted. If you don't know how to do regression, it's easy to run and has a small learning curve to interpret, I recommend it. I hope even JC Denton can be proud of this). So: 1. I pulled the top 50 players in minutes per game (who have played at least 15 games) as of now, for the 2018-2019 season. This even includes the PJ Tuckers and Trevor Arizas of the world. I didn't even want to bias the data by starting with just stars or superstars, although that would be a valid analysis. 2. I ran a linear regression on FT/G (Y) as explained by Usage (X) for these 50 players. Hypothesis: People say Harden has a lot of free-throws. But his usage also leads the league. I'm thinking Usage has a high impact on free-throw rates. 3. Regression output (I didn't include my messy file, you can replicate this): R-square: 0.589 (e.g. 58% of variation in free-throws per game is explained by usage rate alone. not terrible.) XY Correlation: 76.76% (e.g. these two variables have a strong positive relationship, 100% = perfect, 0% = nothing) Intercept: -3.087 Usage% (X) coefficient: 0.3229 (e.g., for each % increase in usage, a player will take roughly 1-more-free-throw for every 3 games played) Usage% (X) p-value: 7.8835E-11 (e.g. usage effect on FT/G is statistically significant, with a very high confidence. Because p <0.03, or even 0.000001 for that matter, we "accept" that usage has some sort of meaningful impact on free throw rate) 4. So, using this model (linear regression: Y = aX + b): Y (predicted FT/G) = .3229*Usage% + (-3.0878) We can get some very quick and dirty predictions of what the MVP group's FT/G should be based on the usage-vs-free-throws relationship of the top 50 minutes-played NBA players: Harden (38.3% usage) - 9.3 FT/G predicted A Davis (36.9% usage) - 8.8 Embiid (31.6%) - 7.1 Giannis (30.3%) - 6.7 Durant (35.6%) - 8.4 Lebron (30.8%) - 6.9 5. How do these predictions differ from actual values? Harden - 9.3 vs 11.1 = +1.8 (e.g. Harden takes about 2 more FT/G than we'd expect from usage alone) Davis - 8.8 vs 9 = +0.2 Embiid 7.1 vs 10 = +2.9 Giannis 6.7 vs 9.3 = +2.6 Durant 8.4 vs 7.6 = -0.8 Lebron 6.9 vs 7.6 = +0.7 Each player in this group, excluding Durant, shoots a little bit of a 'premium' of free-throws versus what their usage rate would suggest - NOT just Harden. And there's more - Harden doesn't even have the highest premium - Giannis and Embiid actually outpace him. At the minimum, Harden is not an outlier.) 6. Let's visualize this! Notice how the guys on the left of the trendline ('premium FT shooters') thrive in the paint. Harden's distance on the top right doesn't matter - it's the diagonal distance from the trend-line that is really apples-to-apples. (Drummond, Simmons, Love, Griffin, Davis, Giannis, Embiid... there are 18 players shooting a premium amount of FTs vs their usage.) With some exceptions (Mitchell) the guys below the trend line are primarily jump shooters (McCollum, Curry, Booker, Durant (not labeled)). So the data jives with intuition... 7. So - what has this 15 minute exercise taught us? * Are Harden's FT/G really so crazy? The data says probably not. * The difference gets even less stark when you compare more similar players. * Don't just look at FT/G and cry! If you want to waste so much energy hating Harden, channel some of it into understanding the nature of your complaints. * Usage explains a huge piece of free-throw rate - and given the visual, paint-presence probably really expands this. I / we should include that in the next model when we have time.
Great post. Would it be possible to extrapolate his usage vs his playoff shooting FTA to see if there is a correlation between the '2'?
It would be interesting to see how this changes in the playoffs. Maybe I'll do that when I get a free minute. But for that I'd also want to pull data for more seasons, and that would take longer. I don't know if you meant to say 'extrapolation', though. That means that we're making up data points or filling in missing data points.
KOC of the Ringer mentioned something about Harden's FTs. I don't have the data but if I remember correctly it was something like "Harden doesn't get as many FT's per "drives" as some other players. He is not #1 but in the top 5." So, then how is he #1 in FTAs? I am guessing that is because of FTA's he gets on his STEP BACK shot. Where defenses are really sloppy in defending that. Most of the time, they just land in his shooting zone. And sometimes they hit his arm. I wonder "per game" how many FTAs he gets on the Step back alone. That number right there could be one of the BIGGEST inflators. Something other players don't get....because they don't shoot that shot.
Also, I don't see a whole lot of "rip throughs" this season. NBA basically cancelled that out. The only way to really weaponize that is what CP3 does late in a quarter when a team is in the penalty. Right now, most of the fouls on Harden on the perimeter aren't rip throughs...but just defenders closing out undisciplined.
If there's data some where on shot type that includes step backs, that would be amazing. I don't think that's tracked publicly anywhere, sadly. I don't think that would be persuasive with general fans though. They're inclined to think that all jumpshot fouls are bogus. :/ There's still something to be said for just how much of a workhorse Harden is. When people say OMG HIS FREE THROW TOTALS what they're really saying is OMG HIS USAGE RATE.
Yeah, I think KOC was citing Second Spectrum data. I just can't completely remember what he said. But it was along those lines. So yeah, it is a cocktail of: 1) General Usage 2) The shear amount of drives to the basket. 3) Fouls on jump shots that are unconventional because of the type of unconventional shot.
he also shoots every technical and late game fouling situation free throw..... that could equal 5+ atts in those types of games.... but if you dont watch the games you wouldn't know that
I actually didn't think about that. That would be another interesting thing to look at. Not sure if Technical shots are separated anywhere.
Refs don't call so many free throws in the playoffs. You've got to make shots. James Harden per game stats Regular season 9.3 free throws Playoffs 7.8 free throws Decreased by 19% That is a lot.
Revised Cocktail of: 1) General Usage 2) The shear amount of drives to the basket. 3) Fouls on jump shots that are unconventional because of the type of unconventional shot. 4) Takes all Technical FTs as well as late game FTs when teams are fouling to get the ball back.