Predicting the NBA MVP with Machine Learning

Thesis

Every season, basketball fans debate who deserves the MVP award. We built 3 machine learning models that attempt to answer that question using box score statistics. At the end of each season, this award is determined by a panel of voters.

Methodology

Each model is trained on every NBA season from 1974 to 2017. For each player season, it looks at nine statistics:

• Points, assists, blocks, defensive rebounds, and field goals per game the core production numbers
• Win Shares (WS): an estimate of how many wins a player contributed to their team
• Value Over Replacement Player (VORP): how much better a player is than a league average replacement
• Box Plus/Minus (BPM): a player's net impact per 100 possessions
• Usage Rate (USG%): what share of team plays run through that player

From those nine numbers, the model learns what a typical MVP season looks like versus a non MVP season, then applies that knowledge to current players. Each model outputs an independent probability that a given player wins MVP, not a share of a single pool, so the values do not sum to 1. Think of it as each player's individual odds.

Three Models, One Question

Rather than relying on a single approach, the system runs three different models and lets you compare:

Logistic Regression

The simplest of the three. It draws a straight line through the data, each statistic gets a weight, and a player's score is the weighted sum of their stats. It's easy to interpret (a higher coefficient means that stat matters more).

Win Shares (WS) is by far the most influential feature, with an absolute coefficient of ~1.85, nearly double the next most important feature. Box Plus/Minus (BPM) ranks second at ~1.0, followed by Defensive Rebounds per Game (DRBPG, ~0.85) and Assists per Game (ASTPG, ~0.70). VORP and Field Goals per Game (FGPG) contribute moderately at ~0.50. Blocks per Game (BLKPG), Points per Game (PTSPG), and Usage Rate (USG%) have minimal weight, all under 0.15.

Random Forest

Builds hundreds of decision trees, each one asking a series of "is this stat above or below X?" questions and averages their answers. It handles complex relationships between stats well and is less sensitive to any one unusual data point. Think of it as a large committee of simple rules voting together.

WS again dominates at ~0.31, accounting for roughly twice the importance of the next feature. VORP (~0.15) and BPM (~0.125) rank second and third. DRBPG (~0.10), PTSPG (~0.08), BLKPG (~0.07), FGPG (~0.065), and ASTPG (~0.06) contribute in a fairly tight mid-range band. USG% is the least important at ~0.05. Compared to logistic regression, the Random Forest spreads importance more evenly across features.

Gradient Boosting

Also uses decision trees, but builds them sequentially: each new tree focuses on correcting the mistakes the previous ones made.

This model is heavily concentrated on just two features: BPM (~0.47) and WS (~0.41) together account for roughly 88% of total feature importance. All remaining features, PTSPG, VORP, ASTPG, DRBPG, contribute ~0.02–0.03 each, and BLKPG, USG%, and FGPG are effectively unused (near zero). This suggests the gradient boosting model learned that BPM and WS alone are nearly sufficient to separate MVP candidates.

Historical Results

The models were trained on data through 2017, so every season from 2018 onward is a genuine out of sample test, the models have never seen these players or seasons before.

Top-1 accuracy: LR 5/8 · RF 5/8 · GB 6/8

Top-3 accuracy: LR 8/8 · RF 7/8 · GB 7/8

Top-3 accuracy: LR 8/8 · RF 7/8 · GB 7/8

For five straight seasons (2019–2022 + 2024), all three models agreed on the same #1 pick, and were right every time.

In 2023, every model ranked Nikola Jokić #1, and by the numbers, he arguably had the better season. Joel Embiid won the award anyway, the kind of outcome that may reflect voter narrative/fatigue and team performance rather than pure statistics. In 2025, Gradient Boosting ranked Shai Gilgeous-Alexander outside the top 500, while Logistic Regression and Random Forest had him at #3 and #2 respectively. I have no idea why GB did this. Likely a bug.

Future Direction

No model is perfect, and these have known blind spots. Team record is not included, MVP voters have historically punished players on losing teams regardless of individual stats. Injuries and narrative don't appear in a box score. And the training data skews toward an older era; the three point revolution and the rise of players like SGA have introduced statistical profiles the 1970s–1990s data doesn't fully capture.

Current Season Predictions (2025–26)

Two of the three models have Nikola Jokić as the frontrunner. Random Forest is the dissenter, putting Shai Gilgeous-Alexander ahead. Victor Wembanyama appears in all three top 3s in just his second season, which is notable. Before running the models, I expected him to be #1 for all of them considering the way the models use advanced stats.

Conclusion

Thank you for reading. I hope you found this interesting. Basketball reference also has their own model if you would like to see a different result. Please do not gamble on my models!

So much of what the algorithms surface is negative. For all of the reasons that mostly everyone's aware of at this point.

It's easy to get the general impression that times are dark without realizing it. I think sometimes it's good to intentionally offset algorithmic (and general human) negativity bias.

Lets do a positive news thread, I'll start:

Hungary votes out Orbán after 16 years

Perovskite solar cells hit 34.85%

Portugal hits 80.7% renewable electricity

Hidden drainage system found in human brain

First lab-grown oesophagus using hosts own cells (fully incorporated with muscles, nerves, arteries within 6 months)

And of course Artemis II! Why is space exploration somehow more positive than the sum of its parts?

Please post anything, it doesn't have to be "news". The full range of the humanities works too

I've been working on a playlist for a while of rock / pop songs with sax solos. The rules are:

• Must be from this millennium
• sax can't be a primary instrument
• pop or rock genres preferred

Here's what I've got so far: https://link.deezer.com/s/323YPvabsQgEuS8BOTCXj

This topic is part of a series. It is meant to be a place for users to discuss creative projects they have been working on.

Projects can be personal, professional, physical, digital, or even just ideas.

If you have any creative projects that you have been working on or want to eventually work on, this is a place for discussing those.

I mean the blue light filter that's built-in in most phones and computers. on Windows, it's called "Night Light" and on most android phones, it's called "Reading mode".

so my question is, do I need not that at all and if the glasses accomplish the same thing? or if they're completely different things. I don't even know if the Anti-Reflective coating provides any protection from blue light.

also I read that blue light can disrupt sleep but I don't really have any problems sleeping, even though I don't use the night light/reading mode

My current phone is about 6 years old and has a lot of signal problems and I replaced the battery about 6 months ago and the new battery is even worse than the original one was and I think it's time to get a new phone.

Things I want in my phone:

• Android
• I would love to have a microSD card but that seems impossible
  • if there's not microSD then I need min 512GB of internal storage and I would prefer 1TB if that's not like +1k to the cost
• Headphone jack
• If there's AI, then I can disable it
• Excellent battery life
• On the smaller end

Things I don't really care about:

• Camera quality (I'm a shitty photographer, the camera wont help)

Things that might be nice:

• Having a stylus but not at the cost of the phone being enormous. Reason I want this is because it gets cold in the winter in Chicago

I hope this phone lasts another 6 years and will be pissed if it lasts under 4 years; at that lifespan and amount of use I get out of it I'm pretty price-agnostic. I live in the USA but expect to travel to Europe a few times next year so it should do decently well when traveling. Network is T-Mobile.

This is not SUPER urgent right now so if your advice is "wait til January because the XYZ phone releasing then is probably better than anything available right now" I would be willing to do that. But my current phone is really not doing great.

Happy to answer any other followup questions if anyone has any!

Thanks!

AD&D 1e. First session of the module I6 Ravenloft, adapted to my campaign setting. Two of my regulars will be unable to join us, and we recruited two new players from friends, for a total of five players. One of them has only had experience playing and running D&D 5e, while the other has only touched D&D briefly, back in the 3e days. Funnily enough, he played through 3e's Ravenloft.

Module spoilers will be found ahead. Given that it's expecting mid-high level characters (six to eight of levels 5-7), I allowed the new players to create their PCs with 55k total XP, which is ~11k less than the next highest among my regular players, and rolled for magic items per DMG Appendix P. The full party is:

• Rowan, Human MU 7
  • Henchman Freya, Human Fighter 5
  • Henchman Rikka, Human Fighter 5
• Iskandar, Human MU 8
  • Henchman Thorgisl, Human Fighter 2
• Bo'Bert, Human Cleric 7
• Elowyn, Elf Fighter 5 / Magus 5
• Nobar, Dwarf Fighter 5 / Thief 6

After introductions were made, we got right into it. They dismounted in the courtyard and their horses led to the stables in the back. They're led through the foyer, into the dining room, and begin to eat as they have a short conversation with the Count. After a few minutes, I kick the module proper off with the doors slamming shut, lights going out, and the Count disappearing.

A couple of them pull out glowing magic weapons while Bo'Bert removes an amulet of continual light from under his clothes. They set up a marching order and set out, heading up and discovering the accountant's office, where they have a rather befuddling, albeit cordial, conversation and push forward into the audience hall. After a little poking around, they find the secret door on the southern wall leading to a turret and a set of stairs, but they forgo these to move through some ornate doors into a large, 20'x70' hall.

As they move through the halls, they're set upon by a baker's dozen gargoyles, swooping in from above. Despite best efforts, several characters are whittled down, a spell is lost from damage, and general pandemonium is at hand. Iskandar uses knock to get the closest set of doors open, Elowyn tries to open with magic missile but a tail swipe knocks her off balance, Bo'Bert blesses the party, Nobar is attempting to train his crossbow on a gargoyle, and everyone else is fending them off with their weapons. On the second round, the party splits into two groups to try and mitigate attacks going towards the casters while Freya stands her ground and continues attacking. By the end of that round, Freya, Rikka, and Thorgisl have formed one group while everyone else formed a second group. The gargoyles ended up splitting between them in groups of 10 and 3, respectively. On the third round, Iskandar releases a monster summoning 1 as Elowyn successfully looses her magic missiles and Bo'Bert eases some of the damage taken by Rowan with a cure light wounds. Everyone else is doing their best to remain alive while in melee with these creatures. They managed to kill one of the thirteen gargoyles this round. On the fourth round, Rowan attempts to withdraw further away from his group while Rikka and Thorgisl do the same from theirs, both incredibly injured by this point. Iskandar looses his own magic missile and, with the efforts of others, kill another two gargoyles. At the end of the round, several manes demons form out of the shadows, and the gargoyles seem to break, choosing to flee from the fight.

Surprised to have made it through that endeavor without anyone dying, the party takes a few minutes to tend to their wounds and distribute a few magic items, before taking a set of stairs down to the chapel. After noting the age of the place and its apparent disuse, they strip the long-dead corpse of a cleric and Bo'Bert removes the Icon from the altar; being Neutral Good, he stands unsmote.

The party has decided to attempt to rest a few hours, just enough to get a couple of 1st level spells memorized. The general state of the chapel has led them to believe that this chamber is the closest thing they'll find to a... "safe"... room. Even if that simply manifests as a lower rate of encounters, it's better than nothing, in their eyes.

That's where we left off for the evening. I've not yet rolled the encounter checks to see if they make it through unbothered.