Running Up That Score

Plus an interview with Mike Beuoy from Inpredictable

Howdy, folks. This week’s edition of The F5 contains an analysis of 10-0 runs in the NBA. Plus I did a Q&A with Mike Beuoy who runs the indispensable website inpredictable.com.

A monthly reminder: The F5 exists because readers like you have decided that this stuff is worth paying $5/month to support it. Paid subscribers receive an extra email every Friday which is one part charts and one part coding tutorial. In these emails I walk through the R code I used to generate the visualizations found in this newsletter. If one of your goals in 2025 is to improve your coding and/or up your data visualization game, consider a paid subscription for the low cost of a Modelo at happy hour.

---

Basketball is a game of runs. Or so I’ve been told.

A 10-0 scoring run can break a game open or bring a trailing team back from the dead. It can fill a team with confidence or demoralize an opponent. The impact of a scoring barrage seems even more powerful when the team that’s on the run is playing in front of an energized home crowd.

The Cleveland Cavaliers go on so many scoring runs that their fans have started to call those runs “Cavalanches.” Their runs, which often come in the form of a cascade of three-pointers, break their opponents’ competitive spirt.

The Indiana Pacers were the latest victim of a Cavalanche. Yesterday, Cleveland was leading Indiana 82-77 with 5:12 to go in the third quarter. In the span of about 90 seconds, the Cavaliers made a free throw and three 3s to create a 10-0 run and go up fifteen points.

After Evan Mobley put the Cavs up 15, Rick Carlisle called a timeout (he actually called two timeouts during this run) but the damage was already done. The Pacers never got within five points the rest of the game and ended up losing by ten — the same amount that was created by the Cavs’ initial scoring run.

TV broadcasters¹ love to mention when a team is on a scoring run. But surprisingly there isn’t much public data on scoring runs. Evan Miyakawa tracks them at the college level. And Ramiro Bentes has an app that records every time a scoring run occurs. But as far as I know, no one is tracking how often NBA teams go on scoring runs and how often they’re giving them up.

So I sifted through the NBA’s raw play-by-play data to create the chart below.

This chart shows which teams are best at creating scoring runs and keeping opponents from doing the same.

To keep things simple, I’ve defined scoring runs as any time a team scores ten or more unanswered points. So in this chart, the x-axis indicates how often teams have gone on runs of ten or more unanswered points. While the y-axis indicates how often teams have conceded runs of ten or more unanswered points. Note that I’ve flipped the direction of the y-axis so that the “good” teams are in the top right corner and the “bad” ones are in the bottom left.

It would be easy to look at this chart and think, “Great work, dude. Good teams go on scoring runs and bad teams don’t. That’s what makes good teams good and bad teams bad!”

And that’s not totally wrong.

Cleveland and Oklahoma City, the league’s top two teams, are in the top right corner. Meanwhile Utah and Washington, two of the leagues’s worst teams, are in the bottom left.

But I think there’s more we can take away from this chart.

Teams like San Antonio and Minnesota, who rank 18th and 19th in offense efficiency, are able to generate scoring runs at a high rate. If two teams have something in common it’s that they have both have supernova scorers (Victor Wembanyama and Anthony Edwards) that can get hot in a hurry and stingy defenses that can lock in for a few possessions in a row. These teams have“spurtability,” to steal an imaginary phrase from Charles Barkley.

Meanwhile, the New York Knicks have successfully been able to limit opponent scoring runs despite ranking 15th in defense. I’ve watched a lot of Knicks games up close the past few years and I think one reason they’ve conceded so few scoring runs this year is because Tom Thibodeau is aggressive with his timeouts. I’ve seen him call timeouts to try to stop a 4-0 run.

That’s the thing about scoring runs. They put pressure on opposing coaches to call timeouts. So in addition to tilting the scoring margin in your favor, scoring runs also influence the timeout battle between teams. Putting your opponents in position to choose between trying to stop a scoring run now or saving a timeout for a potential coach’s challenge later is all part of the game within the game coaches partake in.

I’ll wrap things up with a few stats that I think underscore the importance of scoring runs:

• More threes and better offenses are going to lead to more runs, which is why scoring runs are more common than even before. This season, scoring runs are an all-time high of about 1.5 per game.

  

• Teams that have created at least one scoring run have gone on to win about 62% of the time this season.

• Teams that have created multiple scoring runs have gone on to win about 76% of the time this season.

• Mark Fertel, who analyzed scoring runs last season, showed that 15 out of last year’s 16 playoff teams each registered a positive scoring run margin (i.e., going on more scoring runs than you give up). That could spell trouble for the Pacers, Lakers, Hawks, Mavericks, and Sixers. All of them have playoff aspirations but each are allowing more scoring runs than they are creating at the moment.

This analysis just scratches the surface. There’s nothing special about a 10-0 run. It’s an arbitrary threshold. Future work on this topic could look at different cutoff points for scoring runs, filtering out runs that occur in garbage time, adjusting for opponent strength, and examining the difference in the effects of scoring runs when playing at home versus on the road.

---

A Refreshing Interview with Mike Beuoy from inpredictable.com

One of the first things I’d do if I were running the NBA’s research and innovation lab is write Mike Beuoy a blank check and tell him to work on whatever he wants.

Beuoy is a physicist by training, a healthcare actuary by trade, and all around mensch. His website, inpredictable.com, is one of the best resources for novel and useful stats about the NBA. I use his preCap feature regularly to help decide what game to watch if I’ve missed a night of basketball.

He has written extensively about clutch performance, shooting mechanics, and betting market derived team rankings. How he finds the time outside of his day job is beyond me, but the NBA analytics community is better because of it.

Beuoy grew up in Indiana and still roots for his hometown Pacers even though he now lives in California. I talked to him about his career path, win probability charts, and what he thinks about the state of NBA analytics blogging.

---

This Q&A has been edited for length and clarity

F5: How did you first get into sports analytics? Was there a specific moment or sport you were interested in? Or did it all flow from your interest in betting?

MB: I would say it started when I began transitioning into a management role in my actuarial career. I always liked the coding and data aspects of the job, but as my role changed I was doing less of that. So I started getting into sports analytics because it gave me an outlet to continue to build models and play with data.

My first forays were in basketball and football as those were the two sports I followed most as a fan. The betting stuff was more on the periphery. I did harbor hopes at one time of building some brilliant model that would reliably beat Vegas but I quickly found out that was not easy. What’s been more fruitful for me is seeing how you can leverage betting market information to learn things, like I do with my betting market team rankings. Using point spreads to build an NFL team ranking was one of my first "a-ha!" moments in sports analytics, and it just kind of snowballed from there.

You went to school for physics. How did you end up as an actuary?

After I completed my masters in physics, I realized that most of my career prospects were in academia, which didn't interest me at the time. If I had been any good in the lab, I may have had more opportunities, but my focus and strengths were all on theory. My dad worked in insurance, so I always knew the actuarial field was an option for me, so I started taking actuarial exams and got a job with a company in Chicago in the healthcare space.

I thought I would be doing advanced calculus and solving complex statistical problems all day because that's what the early exams covered. The reality was much more mundane, but the job eventually became fun and challenging once I better understood what we were trying to do and how actuaries can contribute.

What’s the deal with so many actuaries also being interested in sports analytics. Are the skill sets similar or do they just attract similar types of people?

I think it's a little bit of both, but probably more due to the similarity of skill sets. A lot of what I learned on the job as an actuary translated pretty well to sports analytics: treating your data with a skeptical eye, keeping your models simple and explainable, complementing your numeric analysis with business/subject matter knowledge, and developing a good "spidey sense" for discerning emerging trends from statistical noise.

I get the sense that your day job is very fulfilling. But did you ever consider a career in sports analytics?

I've been very fortunate to do a lot of interesting and meaningful work during my career, and don't have any regrets on that front. But I definitely considered whether I could shift gears into a sports analytics role. But I feel like I'm about 10 years too old to have made a career out of it.

By the time I got into sports analytics, I was pretty established in my career, and starting from scratch would have been asking a lot of my family. And it is very nice to have the sports stuff as a hobby that I can pick up when I have the time, and put on the back burner when needed.

It's also nice to decide what I want to dive into. A lot of the work I do stems from answering questions that I have as a fan, even if I know it's not necessarily an "important" question.

The NFL power rankings that Ben Baldwin tweets every week is one my favorite reoccurring bits. It’s derived from your betting market rankings for the NFL. Surely you’ve read some of the replies to Ben’s tweets. What’s your reaction to people that angrily disagree with it. Do you feel the need to explain or do you just laugh it off?

I have almost an anthropological fascination with how people react to team rankings. To my actuary-pilled brain, I just view a ranking system (the good ones) as information, and "all facts are friendly", as they say. But that's not how a lot of people react, particularly on social media.

Ben offered at one point to untag me from his weekly tweet to spare my mentions, but I find the ensuing slap fights entertaining sometimes. If I see someone who has good faith questions or feedback on the rankings, I do try to engage with them. And despite knowing better, I do occasionally get baited into replying to someone who is particularly loud and wrong, but that is rarely productive.

One of my favorite tools you've made is your NBA win probability calculator. Last summer, I was watching a playoff game with Canzhi Ye and we were trying to guess what the win probability was for a team that was trailing with less than a minute to go. Canzhi still laughs at me for being off by a factor of 10.

When you’re watching games, do you consciously or subconsciously find yourself thinking in terms of win probability?

I find I can still mostly enjoy a game for what it is. If I've seen a particularly crazy game or a huge comeback, I do tend to try to get the win probability chart uploaded as soon as the data is available so I can see what the graph looks like. I actually find myself thinking about win probability the most during NFL games, just because there are so many decisions coaches make that can be informed by win probability. There are fewer opportunities for a win probability model to drive decision making in the NBA. The best use case I've come up with is determining when to start fouling, and the optimal strategy only moves your win chances by a few tenths of a point at best.

Pretend a general manager comes to you and says they want you to run their G-League team and treat it like a laboratory for ideas. What's the first thing you'd do?

I don't know if I have any brilliant, upending ideas when it comes to basketball strategy, but it would definitely be fun to try. Off the top of my head, I would love to see what can be done with rebound strategy. If you asked me this 5-10 years ago, I definitely would have said "pushing pace", as that was the low hanging fruit in the NBA for quite some time, particularly pushing pace after an opponent made shot (this was Mike D'Antoni's secret sauce). But I think the NBA may have reached an efficient frontier on that. On rebound strategy, I would love to try out different defensive schemes and rebound "mindsets" (e.g. go all out for rebounds vs. concede and focus on getting back on defense, etc.), and then test rigorously how it impacts rebound success and subsequent offensive/defensive efficiency.

I would also try to keep in mind the impact of inertia. You have coaches and players who have done things a certain way their whole career, you can't pretend you're working with a blank slate. In other words, what are changes in strategy that have a good chance of getting your team to pivot, versus those that will result in heel-digging.

There was a great article in the New York Times recently on Jim Crutchfield, the Division II coach of Nova Southeastern. He's achieved a ton of success with some pretty innovative schemes, but what jumped out to me was how much time and focus there is on getting the players bought in and trained on his system - so much so that he does very little in-game coaching because the team is so well prepared.

It seems to me that big leads are less safe this season in NBA. The three-point rate is an all time high. Offenses are more efficient than ever before. Going on a 10-0 run is more common than ever. How does your win probability model account for this, if at all?

It is definitely the case that big leads are less safe. You can see this in the data, and it really starts to pick up around the 2016-17 season. I track how often a team trailing by 15 or more points at any point in the game eventually comes back to win. It was in the 8-9% range for most of the 1990s and 2000's. It's now around 13-14%. In addition to increased three point rate, the increase in pace is also a factor - more possessions equals more opportunities to come back.

Unfortunately, my win probability model currently does not account for this. It was trained on data from 2000-2012, which is right before the big explosion in three point attempts and faster pace. It's been on my to-do list to update the model for the modern NBA, but other priorities have gotten in the way. My hope is to future proof it by adding projected pace and three point attempts as an input into the model, so if we see a shift in NBA strategy, or perhaps rule changes, the model would organically adjust for it.

Does it feel like there's less public NBA analytics work today then there was ten years ago? We've both written for FiveThirtyEight and Deadspin, neither of which exist anymore in the same way they used to. Meanwhile, people who publish quality work get hired by teams or organizations and are never heard from again publicly. Do you feel like there's a difference today then there was when you first started blogging on your site back in 2011??

It definitely feels that way, although some of it might just be nostalgia. There is still plenty of really interesting publicly available work out there. Dunksandthrees and craftedNBA come to mind. Neil Paine has a good substack. Caitlin Cooper and Nekias Duncan do really great X's and O's breakdowns. And, of course, The F5.

But people, myself included, just seem to read less these days. I do miss where there were a lot of outlets where you could read medium and longform content on analytics topics. One of the reasons I started inpredictable was because I enjoyed writing (I started college as an English major before pivoting to physics). Now, a lot needs to be boiled down into social media-sized snippets, which can only go so deep.

What's one thing you cant live without during the NBA season?

I always love seeing what Todd Whitehead comes up with next. It's rare when someone can create work that is entertaining, creative, and informative.

1. While I was writing this I came across a midly amusing post on a forum for sportscasaters asking for advice on how to keep track of scoring runs. I miss forums.