Breakdown: Back-To-Backs (HV Weekly: 1/29/21)
Slower pace, more fouls, and a betting trend that seems too good to be true
Welcome back to the HV Weekly!
Tonight we have one of the best matchups of the season between Iowa and Illinois. Whether you’re reading this newsletter before tip-off or after, stay tuned to Twitter for some eventual post-game analysis on that one.
For today’s edition of the newsletter, we are getting back to our data roots by breaking down back-to-backs and uncovering a betting trend that seems too good to be true.
Before we get into that, though, the quick sales pitch:
Do you want to receive our weekly Starting Five newsletter and plenty of other subscriber-exclusive content, along with discounts to other Hoop Vision products? Subscribe to HV+ today and help us keep Hoop Vision independent and ad-free.
Ok, let’s go!
A Very Different Season
Theoretically, this should be a great season for basketball research.
Have you ever wondered how big of a role the crowd plays in homecourt advantage and how it swings games? Now we have games being played with no fans.
Or if you’ve ever wondered about the importance of being in “game shape” when coming back from an injury, we now have teams being shut down for two weeks and then thrown back into live action right away.
When you sit down and begin researching though, there’s one massive issue: everything is abnormal this season. With so many different variables in play, isolating cause-and-effect, finding correlations and discerning signal versus noise are all nearly impossible.
Attendance and fan support are obviously different this season, but so is travel and scheduling, and practice time, and player availability, and pretty much anything else you can think of. The chaos has created an almost never-ending cycle of potential narratives.
Ben Falk at Cleaning the Glass wrote about these challenges as it related to the NBA bubble over the summer:
Players have had to deal with slick floors and coaches have had less time to prepare given the quicker turnaround times between playoff games. Young players have had months off to work on their game and improve, and haven’t faced quite the same pressure of playing in front of a large crowd on a major stage. Some players have surely benefited from the lack of distractions outside of basketball. Others have surely felt a major impact on their mental state, potentially sapping energy or motivation. Chemistry and familiarity built up over a regular season have been disrupted. And player conditioning (or lack thereof), particularly as a result of whatever players did during the long hiatus, has taken center stage.
While complicated in its own right, at least the teams in the NBA bubble were able to operate under similar circumstances, relative to opposing teams. Things were unusual, but every team was in the same boat.
In college basketball, individual teams and conferences have varied widely when it comes to scheduling and logistical decisions.
Long story short: the sort of analysis we’ll be doing below is tricky.
Now that we have that big caveat out of the way, we decided to narrow in on a specific scheduling pattern that has emerged as a result of the pandemic:
Back-to-backs.
What a difference 24 hours makes
To avoid travel, some smaller conferences — especially those unable to charter flights — have teams playing back-to-back games against the same opponent.
Overall, we’ve had 272 back-to-back series played so far this season. That only includes series that were played on consecutive days (zero days off in between).
To make matters even more confusing, the Patriot League is the only conference that is (sometimes.. not always) playing back-to-backs in different locations!
On January 9th, Holy Cross hosted Army for a conference game. The next day, the two teams both travelled to West Point to play another. That two-location style of back-to-back has happened nine times in Patriot League play thus far.
Worth noting: the Patriot League has also done the usual single-location back-to-backs, as well.
On January 2nd, Colgate opened their season with a dominant 101-57 beatdown at home over Army. (The NET rankings especially loved it.) Just 24 hours later — in the same location — Army got revenge with a 75-73 victory.
Biggest turnarounds from game #1 to game #2
1/2 & 1/3 — Colgate beats Army by 44, loses rematch by 2
1/1 & 1/2 — Wright St. beats Oakland by 39, loses rematch by 10
1/1 & 1/2 — S. Alabama beats Ga. Southern by 29, loses rematch by 13
12/11 & 12/12 — Iona beats Fairfield by 28, loses rematch by 15
The graph below plots the margin of victory in all of the back-to-back rematches.
Overall, the team that wins the opening game has a win percentage of 59% in the second game.
But of course, we can (and we will!) go deeper than simple margin of victory and win percentage.
Next: How efficiency and the four factors change during the second game of a back-to-back.
Game #2… slower and physical
Teams do play back-to-backs in a normal season, but usually only within tournaments. This season’s back-to-backs, however, are an especially interesting test case.
The table below shows pace and points in the first game of a back-to-back compared to the second.
On average, the second game is played at a slower pace.
Again, determining causality can be tricky (if not impossible), but it makes intuitive sense that players have tired legs and no days off, which may lead to a slower pace.
But why the higher efficiency? We can go a level deeper by looking at the four factors.
The biggest factor in driving efficiency up is an increase in free throw rate.
In a vacuum, fouling is almost never a good thing — free throws are one of the most efficient forms of scoring for an offense. But for aggressive denial/pressing defenses, the risk of fouling can sometimes be worth it in order to generate more turnovers.
That doesn’t appear to be what’s going on here. Turnover rate actually decreases in the second game of a back-to-back.
Instead, I think it’s more likely that fouling is a byproduct of fatigued players.
NBA studies in the past have shown that fouling is connected to fatigue. More specifically, the rate of shooting fouls increases the longer that play continues without a timeout break.
This is mostly just a theory, but the back-to-backs seem to be leading to some combination of slower, sloppier, and more physical play during the second leg.
(Quick note: One coach pointed out to me that in these back-to-backs, the referees are the same in both games. Does that play a role in all of this? Truthfully, I’m not sure.)
The market (might be) over-adjusting
Next up, I wanted to look at how betting markets are reacting to the back-to-backs. So using statfox, I went through spread and totals data.
This part gets a little confusing and conceptual, so let’s start with a real example…
On January 7th, Merrimack played Sacred Heart. Before the game, the betting total (aka the over/under) was 128 points.
The actual game went way over. Merrimack defeated Sacred Heart 97-90 (187 points). The final number was inflated by the game going into overtime, but there were still 162 points scored at the end of regulation.
So for the January 8th rematch (remember: the second leg of a back-to-back), how do you think the market responded?
In this case, the second game’s projected total increased to 137.5. Because of the game one day prior, handicappers expected more points.
The graph below plots how the amount of points scored in the first game has affected the second game’s total for every single back-to-back this season.
The Merrimack-Sacred Heart data point is labelled in the top right.
The graph has a pretty strong correlation, meaning game one results actually do influence betting markets.
However, notice the scale of the y-axis. For every 20 points that game one goes above the expected total, the game two total moves by about 2.5 points.
Now let’s go back to Merrimack-Sacred Heart…
As mentioned above, the betting markets adjusted for game two — expecting 137.5 points.
In reality, the under hit. Sacred Heart won the rematch 68-62 for 130 total points.
This example is part of a larger trend.
When the game two total INCREASES from game one (just like Merrimack-Sacred Heart), the UNDER has hit 60% of the time (63-42-1).
When the game two total DECREASES from game one, the OVER has hit 57% of the time (78-60-4).
That’s just by betting against the total movement — without any knowledge of who the two teams are.
If you find a very simplistic betting trend that feels too good to be true, well, it’s probably too good to be true. It’s very hard to beat high-volume betting markets where the lines are being sharpened by action.
Also, I wonder if the over-adjusting plays a bigger role early in the season. In our Merrimack example, that was their very first game of the season. As we move later in the season, totals should be less dependent on single game results.
Links from around the internet
I love the idea of Jim Boeheim breaking news on Cameo of all places
David Worlock (NCAA Director of Media Coordination/Statistics) said he’s a “big fan” of wins above bubble on Jeff Goodman’s podcast — an encouraging sign for modernizing the selection process. Check out the 35-minute mark of the podcast
Scott Drew stealing a Bill Self special
Ed Cooley tight flex
Mark Titus needed better background music
Pat Kelsey false action for a backdoor
Juwan Howard sequential playcalling
Some more thoughts on helping one pass away
Reason #1,839 head-to-head should not be used to rank teams
Thanks for reading the Hoop Vision Weekly!
Please consider joining as an HV+ subscriber today ($10/month or $100/year) to receive more Hoop Vision content — including The Starting Five newsletter — throughout the college basketball season.
Your support helps us keep Hoop Vision independent and ad-free!