Was anything else of importance?

If you are a coach that has somehow stumbled upon this page, it might be best to cease reading after the In-Game Stats section. If you are a Tennessee fan, continue all the way to the bottom.

Depending on what you’d call important, maybe. I couldn’t quite fit these observations anywhere else in the post, so I’ll leave them here in a miscellaneous area at the bottom:

Pregame Win Likelihood

• Without any context added, teams that were at home won 54.7% of these coin-flippy games. The home win rate across all of college basketball the last two seasons is a hair under 60%, so while this is a useful thing to know, it’s not that much more important than anything else out there. 54.7% roughly equates to a 1.3-point advantage, by the way, which is under half of the ~3-point home-court advantage example college basketball has set for itself.
  • Here’s a good reason why: in our dataset, home teams received 3.5 more free throw attempts per game than the road team did. Going back to our earlier stat of the average college basketball team shooting around 70-71% from the free throw line, that’s about a 2.5-point advantage to the home team. On average, home teams were probably a tad worse than the road team, but overcame it by way of beneficial officiating.
• There was a fairly real correlation between how likely you were to win the game before it tipped off and if you actually ended up winning it. Teams sitting at 50-69% to win pregame were 52.7% successful in doing so; teams at 70% or better were 64.9% at surviving a closer-than-expected battle. Teams at 80% or better: 69.6%.
• To piggyback off of this, your odds of surviving a close game if you were at home and you were favored before tipoff were legitimately elevated. Those teams won 59.7% of these close games out of a 2,125-game sample, while all others were successful 44.1% of the time. Road favorites, meanwhile, did still manage a 55.9% success rate.

In-Game Stats

• Unsurprisingly, teams that won the 3PT% battle won 57.6% of the time, besting the 56% success rate of 2PT%.
• Also unsurprising: the more Four Factors you won, the better your odds of winning were. Teams that only got one Four Factor win had a 12.6% chance of winning their games; those who won at least two were at 70%.
• The two best Four Factors to win, unsurprisingly: eFG% and TO%. In games where a team only won two Four Factors, teams that took eFG% and TO% won 80.7% of their games. eFG% + FTR, the most important combination in close games, isn’t as effective when applied to a larger sample, sitting at 69.3%.
• Here’s how all the Two Factor combinations came out:
  • eFG% + TO%: 80.7% win rate
  • eFG% + OREB%: 79.5% win rate
  • eFG% + FTR: 69.3% win rate
  • OREB% + TO%: 33.1% win rate
  • TO% + FTR: 25.6% win rate
  • OREB% + FTR: 24.7% win rate
• There were some various other goofs I checked into, too. Across the full sample size of 2019-20 and 2020-21, not just close games, teams that forced opponents to turn it over on 25% or more of possessions won 66.1% of the time, which feels about 33.9% lower than it does when it happens to your team. Teams that rebounded 40% or more of their misses won 72.4% of those games. Teams posting a 60% or higher eFG% – 10% above the national average – won 87.1% of the time.
• Georgia Tech scored 120 points against Georgia State and lost, 123-120. Manhattan scored 45 points against Quinnipiac and won, 45-42. No point total, offensive or defensive, is safe; only efficiency totals reasonably can be. (Teams scoring 1.34 PPP or more were undefeated. Unfortunately for college coaches, this only happens in about one of every 1.6% games, or roughly once every two seasons for the average team.)

What About Tennessee?

This is a Tennessee-heavy readership, so naturally, I assume people want to know how Tennessee shook out. Perhaps unsurprisingly, Tennessee lines up from an efficiency perspective with just about everyone else in college basketball, at least in the Rick Barnes era (2015-16 to present).

• eFG%: 48% of raw efficiency; 52% of opponent-adjusted efficiency
• TO%: 27% raw; 27% opponent-adjusted
• OREB%: 16% raw; 14% opponent-adjusted
• FTR: 9% raw; 7% opponent-adjusted

eFG% matters a hair less for the Vols, while FTR matters a tiny bit more. Still, this equation looks pretty darn similar to the one all of college basketball uses. Would you like to know how many times Tennessee has won each stat in the 194-game Rick Barnes era, along with their win percentage for each when they ‘win’ this stat? Of course you would.

• eFG%: 121 ‘wins’; 85.1% win rate
• TO%: 119 ‘wins’; 72.3% win rate
• OREB%: 98 ‘wins’; 77.6% win rate
• FTR: 96 ‘wins’; 80.2% win rate

Shocker: shooting the ball well wins you games. Perhaps Tennessee should explore this less-than-ideally-mined territory on offense in an upcoming season.

When it came to close games, we’ll keep it fairly short. Tennessee has played in 57 games decided by six points or less in Rick Barnes’ six seasons; they’re an almost-perfect 29-28 (50.9%) in winning them. Here’s their win rates for the Four Factors in these 57 games, for those somehow monitoring such stats.

• eFG%: 15-12 when they win it in close games; 14-16 when they don’t
• TO%: 18-13 yes; 11-15 no
• OREB%: 15-9 yes; 14-19 no
• FTR: 21-4 yes; 8-24 no

Because these are very small sample sizes (I would wait until you get to, like, 150 close games before drawing anything from this), it does look a little out of whack. However, it’s quite convenient for Free Throw Rate to be of the utmost importance, apparently.

What have we learned? Here’s my major takeaways from this exercise.

1. eFG% is the most important Four Factor by a wide margin.
2. Free Throw Rate is the least-important one, but it takes on an outsized importance in tight games.
3. OREB% may be the most talked-about stat, but perhaps the least important of the Big Three.
4. A lot of what you just read can also be explained away with “there’s a lot of luck involved” and it would make just as much sense.

Your personal takeaways may differ. I’d love to hear about them, whether at statsbywill@gmail.com or on Twitter @statsbywill.