The Title Formula

I think I have a pretty good way to estimate contenders at this point. Let's break it down.

Last year, I wrote a piece titled “no one can win the title.” The basic premise was to take each team remaining in the playoffs and find some metric on which at least 90% of previous champions had met a threshold on which that team failed. The conclusion was that while there were traits applying to each team, the team with the fewest failed “tests” had the best odds - namely that the Celtics should be expected to win. The main flaw with that approach I think is that it was not localized to any time period and also if you open up enough metrics, you’ll find things only a few title teams failed. That said, I remain convinced that finding historical parallels is the most effective way to understand what things provide concrete value to a team. Recently, I’ve set out the goal of overlaying enough of these “rules” together to create a formula that isolates 1-3 serious contenders per year and I believe I have succeeded. In this post, I’ll lay out the criteria, what it gives up, and why I think it works.

The formula itself has two components: a “contender” component that assesses a team’s performance and a “star” component that assesses if the team has a leader who can take them across the finish line. The star component is much harder to isolate and I’ll get into why later, but the contender component is rather straightforward. There are four criteria: seed, net rating, shooting, and defense. Since the merger, 47/48 title teams have had a top-3 seed and a top-10 net rating. The one exception to this rule was the 95 Rockets. I’ll discuss the seasons the formula misses below, but this season was anomalous for quite a number of reasons. I think it is fine to ignore it and refuse to take any team that does not meet this threshold seriously.

The third criterion is shooting. This is a more recent development and has not been true for a lot of NBA history, but in the modern world, a team’s ability to shoot is critical. Since 2010, every single title winner has been top-5 in effective field goal percentage. We can quibble about eFG vs true shooting, but my goal was to have as few criteria for this formula as possible and the 2020 Lakers were 9th in TS%. Prior to 2010, shooting remained important but not critical. Between 2000 and 2010 there were 3 title teams outside the top-10 in eFG and a fourth who was 6th. This I think, is largely attributable to an overall spike in 3-point shooting. While 3-pointers have been steadily rising since their inception, 2010 to 11 saw a gap of 2.6 3-point attempts per game between champions and the champion passed 20 attempts for the first time. In 2015, this completely exploded but 2010 seems to be the point where we passed the critical mass. This changed the math behind the game in a substantial way which left me feeling good about a “top-5 eFG” rule even though it is more recent and runs on a much smaller sample. A big note here is that the goal of the formula is more to predict future champions than to capture the previous ones.

The last team-based criterion is defense. Of the 45 title teams in the 3-point era, 44 have had a top-half defense. 41 have had a top-10 defense, laying out a slightly stronger emphasis, but I prioritized making sure all champions were included over cutting down more outside teams, trusting that overlaying these different criteria would limit the field sufficiently that any individual criterion could take a backseat. The 2001 Lakers are the team that did not meet this rule. Once again, this was a fairly anomalous season that I’ll talk about below, I feel fairly good about betting against this rule being broken again.

These four together produce our “contender index.” If a team meets all four, we can take them seriously as an elite basketball team. However, history suggests simply being elite is insufficient to get the job done. There have been 224 playoff teams since 2010 (where these rules are applicable due to the eFG caveat) and only 37 of them have satisfied the contender index. Notably, all 14 champions are included in that number. The main downside of ending the conversation here is that the field is still substantial. We include an average of almost 3 teams a year. While this is useful for paring down the field, it is limited in actual predictive value since for the most part I think people are able to pick 3 contenders a year and be right on pure instinct. What I will say is that I think for a team to beat my formula in the future, it is far more likely they will break the star index than the contender index, this one is far more ironclad.

The star index is more breakable because it is so much harder to build. The starting point for me was 39/45 3-point era teams have had an MVP on the roster. While this is a number that approaches “maybe weird things are happening the other years” it is also wide enough that we’re going to miss a decent chunk of years if we take it at face value. Compounding this problem is that 2 of the 6 exceptions have come in the last 6 years; a substantial increase in percentage from before. If a third exception comes this year (such as if the Celtics or Cavaliers win, or the Thunder win without Shai winning MVP), that will leave us with a 9% chance that this is due to random variance rather than the change in year being a causal variable. I’ve been vocal that I think the MVP criteria changed radically in 2017, potentially presenting one such explanation for this shift. Another explanation could be that the game itself has changed and supporting cast has become substantially more important recently. In any case, the predictive value of having an MVP would drop substantially. That said, I think most people recognize a massive distinction between the 2019 Raptors and 2024 Celtics led by Kawhi Leonard and Jayson Tatum and say, the 2021 Jazz or 2015 Hawks led by Rudy Gobert and Al Horford. I’ll leave the door open for evaluation of star quality down the line, but what I ultimately determined is that for now, the team stats usually require star power to achieve and talent is not necessary to measure. As a result, the star index only measures experience rather than talent.

43/45 teams in the 3-point era have been led by a player in at least their 6th season and 44/45 have been led by a player with a prior Conference Finals appearance. The missing seasons for years of experience are 1981 and 1984 - both led by Larry Bird in his first 5 seasons. It’s also worth noting these happened right at the start of the modern era and seemingly many of the previous stars had retired, aged out of relevance, or gotten injured. The current era sees stars retain dominance for longer and so these factors are less likely to allow this rule to be broken. 35-year-olds are regularly making the All-NBA teams, there are simply more candidates to lead teams than in previous years. The missing team for the prior Conference Finals appearance is from 2015, a very strange season generally which I will address later.

With only these criteria plus the previous contender index, we limit our pool of 224 teams to 28 teams, roughly 2 per season. This result is functional but still feels a little unsatisfying. Conventional wisdom tells us that a team needs a superstar to lead them to a title. I do think most of this is contained in team stats, but there are teams floating around that I think could be excluded with slightly stricter criteria. However, couldn’t find an objective definition of superstars that feels elegant while at the same time including every actual title team. In our 14-year sample, the 19 Raptors and 24 Celtics present the biggest problems. Without those two seasons, the criteria could be “an MVP who has made 1st team in the last 3 seasons” and the field would be limited to 18 teams across 12 seasons, a clean 1.5 teams per season with only the 15 Warriors unaccounted for among title teams. However, that’s nearly 15% of title teams missed in this window so perhaps limited value as a predictive tool.

The other disadvantage of this formula is that you lose more of history. Even if you ignore eFG% for teams from 2010 and prior, you still lose all teams previously mentioned (84 Celtics, 95 Rockets, 01 Lakers) but additionally the 89, 90, and 04 Pistons. The trouble is finding a set of criteria that encapsulates the stars from all 5 teams that fail it. We could relax the MVP requirement to an MVP, 2 DPOYs, or having made 3 total 1st teams and remove the requirement to have made a 1st team in the last 3 years, but this ends up making the field too large. It ultimately doesn’t limit the field much more than the simple contender index would. If we dismiss the old championships and focus only on 2011-2024, it is still somewhat difficult to find criteria to overlap Tatum and Kawhi. This would become very simple if say, Kawhi had won MVP in 2017 and Tatum in 2024, but making arguments for those awards is beyond the scope of this post.

The cleanest cut I was able to find is to keep the MVP and recent 1st team requirement but to have an exception for teams with a player who has made 2 first teams in the last 4 years. I separately considered making it 2 first teams in 3 years with a 4th added if they played under 50 games one year (so that Kawhi could still make it), but as it turned out that did not shave any additional teams out of the pool. One of my primary goals has been to keep the criteria as simple as possible so adding this if statement to the calculation is without benefit would fly against the goals. Even without it, this 2 in 4 criteria felt a little forced, and frankly, felt low for defining “superstar.” Within the 14-year sample, cutting the accolades requirement altogether adds 5 teams: the 11 Spurs, 12 Spurs, 21 Suns, 22 Suns, and 22 Heat. These teams have a combined win percentage of 72.66% - a 60-win pace over an 82-game season. 2 went to a Conference Finals and a 3rd made the Finals. The point here is that these were still very good teams. I think there’s a reasonable argument to be made, particularly with the Suns and Heat teams, that the accolades are meant to keep teams like them out, but for now, I will prioritize the formula being less convoluted and say they pass. I’ll potentially return to it later depending on results this year or if I find a convincing way to define “should’ve been MVP” that isn’t simply my subjective belief.

This leaves the formula including 6 data points: top-3 seed, top-10 net rating, top half defense, top-5 eFG%, and a star that has played in a Conference Finals and is in at least their 6th year. I’ll start by going through all the years in the 16 playoff team era (1984-present) where this does not pick up the correct NBA champion (with eFG excluded for all years prior to 2011) as a possibility.

First, 1984. The Celtics, led by Larry Bird, won this title. They met 4 of the 5 live data points this year, but Bird was only in his 5th season. As noted, this was a transitional time. Magic Johnson and Larry Bird were quickly overtaking the older generation and the addition of the 3-point line was beginning to change the geometry of the court. In 84, 6 teams passed the contender index and only 2 of those had a leader with the requisite experience: the Lakers and the 76ers (conveniently the Finalists from 1983). Two of the other three teams that passed the contender index - the Blazers and the Jazz - lacked a player with Conference Finals experience but also they and the Bucks lacked an MVP. While this isn’t a mandatory criterion, it does seem to present advantages. The Celtics had an MVP and their numbers this year were better than either team who passed everything - they had both a better offense and a better defense than either of them. The 76ers ended up upset in the first round by a 7th seed (that somehow had a very similar net rating) and the Lakers made the Finals. I’ll say that while the formula misses here but was 1 year of Larry Bird's experience from passing despite being optimized for a much different era.

The next team to miss is the 95 Rockets. This is perhaps the most anomalous champion in NBA history. They were a 6-seed with the 11th-best net rating, though they cleared every other threshold and even had an MVP. While a number of teams passed the contender index this season, the only team that cleared every threshold was the Jazz. Funny enough, they were eliminated in the first round by this Rockets group. I don’t have a great explanation for this Rockets team and likely never will, but I’ll note a few things that gave them a leg up. If the MVP criterion were introduced, this season would have no one who passes all the data points - Malone does not win MVP until 97. Hakeem had won in 94, suggesting they at baseline had a superior superstar. Additionally, the Rockets acquired Clyde Drexler, who had made the All-NBA 1st team in 1994, at the trade deadline. While they didn’t see a considerable leap in regular season performance after doing so, that at least gives some room to believe their regular season numbers were not representative of true ability. With that in mind, I defy anyone to develop a coherent formula that encapsulates this team. They were simply too poor in the regular season to have any formula take them seriously without exploding the available teams.

Next up is the 01 Lakers. 01 saw 0 teams who satisfied the formula. 3 teams passed the contender index but didn’t have an experienced enough leader. The Spurs and the 76ers had stars who hadn’t played long enough and Chris Webber of the Kings had not made a Conference Finals. The Lakers' major flaw was they had the 21st-ranked defense. Of note here is that the Lakers faced considerable health problems this year, likely disrupting their chemistry. In a year with no one passing the formula, it’s hard to say which element is the easiest to make do without. If we look at MVPs, the Kings seemed like an unlikely candidate. The Spurs still had David Robinson on the roster, but he hadn’t been relevant in several years and wouldn’t have satisfied the “making a 1st team within the last 3 years” requirement. Perhaps the 76ers should have won that year - their team numbers were better as a whole and Iverson was the MVP. The Lakers had a much better offense and better experience but there isn’t enough data on seasons with 0 teams to really extrapolate from. It’s anyone's guess what gave the Lakers the edge here and I think calling it a toss-up would have been reasonable.

Finally, the 15 Warriors. This is the only team in the main 14-year sample I looked at (which includes eFG% now) to fail the formula. In fact, no team in 2015 passed the formula. The Hawks and the Warriors passed the contender index but both lacked a star who had been to the Conference Finals. Of the other top seeds, the Rockets were outside the top 5 in shooting, the Bulls were outside the top 10 in net rating (and 21st in shooting), and the Clippers and Cavaliers missed on defense. Based on 01, the Cavaliers probably would have been the most logical pick with Lebron as a stand-in for Shaquille O’Neal and Stephen Curry as a stand-in for Iverson. I’ll note that Curry had a year of experience on Iverson which helps but even if you take it as a straight analog there simply isn’t a large enough sample to predict a repeat outcome. I think the most important thing the Warriors had is that they altered the way basketball was played that season. As previously noted, 3s had been steadily going up for years before hitting a critical point in 2011. This took off to absurd levels in 2015. The 15 Warriors took the lid off of the pot, shooting 5.6 more 3s than the 14 Spurs and 4.9 more than the 13 Heat (the previous record holder among champions). The 15 Warriors took the fewest 3s per game of any champion since then, signaling a clear shift in approach. I think understanding the advantage the Warriors created as subsuming everything else is the clearest way to understand this season. There’s a reasonable argument to be made that it wouldn’t have been enough if the Cavaliers stayed healthy, but I think “fundamentally altering the way basketball is played breaks the math” is a more persuasive argument to me personally.

That’s the formula in full, it catches all other champions and eliminates almost every single other team in NBA history, usually leaving 1 challenger. In the last 3 years, it has given the Celtics standing alone in 24, the Celtics and the Nuggets in 23, and the Warriors, Suns, and Heat in 22. These all feel roughly indicative of how strong teams felt relative to the field in their respective seasons which I’ll count further in favor. I’ll go through some notable successes as well. The 16 Spurs are a top-5 regular season team in history and this formula says they weren’t a contender because they couldn’t shoot well enough. They ended up eliminated in the second round by the Thunder, who the formula says was a contender that year despite much worse overall numbers. In 8 of the 14 seasons, it picked up either only 1 team (who would win the title) or both Finalists. This would become 9 had Jayson Tatum not sprained his ankle in 2023. 2018 it only picked up 2 teams, but many considered the WCF between the Rockets and the Warriors (the 2 teams) the “true” Finals, somewhat validating it further. Ignoring the 22 Celtics and the 12 Thunder for lack of experience also appears apt - despite the deep runs they made both were thoroughly outclassed by a far more experienced opponent in the Finals. Overall, I find the results quite satisfactory.

The important remaining question is what it says about this season. As of right now, we are getting another 0-contender season but I do not believe this will hold through game 82. I’ll go through each of the top 3 seeds in each conference, what is excluding each of them, and if it is reparable.

In the West, the Thunder are on a historic pace with their overall numbers, leading the league in record, net rating, and defense. However, they are 11th in shooting and their star has not been to a Conference Finals. Their most persuasive argument is that their turnover rate is so historic that they are in essence changing what is possible on the court, possibly moving the game to a de-emphasis on 3-point shooting. I’m not convinced by this personally as teams have held similar advantages in turnover rate before but could not escape the rest of the data come playoff time - as opposed to the 15 Warriors who legitimately broke 3-point shooting. I do think lack of experience may also be easier to overcome if the rest of the field is weak enough, but certainly, it leaves a team vulnerable to be able to lose virtually any series. This makes the luck factor needed to win 4 straight higher. It is that luck factor that I think is their biggest weakness; they simply don’t shoot very well. Any team that gets hot against them will have very reasonable odds to knock them out. This is basically what happened to the Celtics in 23 and for that reason, I think they are hard to trust. They can’t fix Shai’s inexperience and I think it is unlikely the shooting gets hot enough to fix their standing there.

The Grizzlies are a great-looking team across a lot of metrics. In fact, the only criteria they miss is that Ja has not been to a Conference Finals. Perhaps this should be enough to get them credit. However, I just don’t see the way they play as revolutionary enough to be a proxy for the 15 Warriors. Notably, the 15 Warriors had unmitigated dominance in the regular season, while the Grizzlies have simply been pretty good. I also think there is a reasonable argument that Ja is the worst or second-worst player of all these top-6 teams, hurting their case if we think the independent quality of the star (separate from how they impact team stats) has outsized importance (such as with the MVP criteria). Honestly though, if no one hits their thresholds, there is no concrete reason it shouldn’t be them, just that I (and seemingly no one else) don’t see it happening. They can probably hold the thresholds they have but cannot fix Ja’s inexperience.

The Nuggets are one of the closest teams to meeting every criterion. After a slow start, they have raged forward and now have everything they need except that they are currently the 16th-ranked defense. I think if their recent heater is sustained at least somewhat, that defense will climb into the top half and they’ll satisfy the formula. If this happens, they’ll also have the benefit of one of the few true MVPs among serious teams - historically a powerful force even if not totally determinative. Their big thing is that we are assessing numbers at the height of a hot streak right now and it’s not clear where they’ll settle. Personally, I feel good about taking them seriously no matter what. Even if the defense doesn’t quite clear the threshold, the 01 Lakers and 15 Cavaliers are solid arguments that defense is one factor you can at least somewhat disregard if no one hits everything. They should be seen as a serious threat to win it all.

In the East, the Cavaliers have dominated. They meet every threshold except Donovan Mitchell’s playoff experience, something that is unfixable. Their pull-up shooting has been historic, perhaps in a way that can be argued is analogous to the 15 Warriors. However, I’m not sure I believe “simply hit every shot” is as revolutionary as “let’s change the way we think about shots.” That said, its historic levels make it more believable. I’m also a big Donovan Mitchell believer so I may be higher on this team than I ought to be. That said, Donovan Mitchell is cleanly in the bottom half of the stars on this list so once again, a knock on them. Like Ja, he has not even made a 1st team yet. While I think he ought to this year, he may not. If we trust the voters’ talent assessment, this casts serious doubt on this team. Like the Grizzlies, I’ll leave it as there isn’t really a reason to doubt them if no one hits their thresholds, and I think they’ll cause problems for anyone they see, but not sure I see them winning anything.

The Celtics are also very close to meeting their thresholds. All they lack currently is shooting and they currently sit 7th in eFG%. They currently have the Pacers, Grizzlies, and Bucks in front of them. None of those teams stand out as shooting juggernauts who Boston can’t catch. Boston was also 2nd last season with basically the same roster and has the reverse situation of the Nuggets; they’re being evaluated as they come out of a slump, perhaps at a time when their numbers should be expected to rise. The main reason to doubt they can hit this threshold is that Al Horford and Jrue Holiday, their oldest rotation players, are shooting far worse than last year. It seems quite plausible this is due to the miles on their legs more than a slump and so will not simply bounce back over time. While the Celtics may still have the shooting talent to get into the top 5, this would make it a lot harder. The other argument against them is that some would argue Tatum is only the 3rd best player of these teams. However, it’s hard to credit this argument substantially as he has already proven that he is sufficient to win a title. Additionally, the two arguable players ahead of him are both in the West and so may have to fight each other to even see him. The Celtics overall look like a good bet to clear their thresholds and as good as any to win without them if no one does. Perhaps the best counterargument is there isn’t a great precedent for lack of shooting being something you can overcome in the 3-point-dominated world.

Finally, the Knicks. The Knicks are clearly the weakest team on this list. Their stars are mediocre, lack experience and the team doesn’t guard. Frankly, a team anchored by KAT with Brunson out front is never going to have a title-level defense and I think their current 18th feels like as high as it goes. I don’t care to spend a ton of time with the Knicks. They are shooting well enough now, but the team sucks and is more likely to be a 1st round exit than make the Conference Finals. I won’t insult anyone’s intelligence by making their case on page 15 of this post.

So that’s the conclusion. I feel very good about the formula and the current assessment for this year is that the Nuggets and Celtics seem like the frontrunners once again but the field may be wide open. What do you think? Are their other aspects I should have factored in? Things I overvalued? Please let me know, I’m very interested in refining this formula and continuing to increase precision through very simple historical analogs. Thanks for reading!