2c:null 2d:["$","title",null,{"children":"Project 82-0, Part 1 | Gideon Jeffrey | Observable"}] 32:T6e5,33:Tbd4,_Why_, you might ask, _are doing this? Doesn't this spoil the game?_ The people who created 82-0 may think so, that part of the game's success or appeal depends on the algorithm's secrecy, but I don't believe that's true at all. First of all, I'd argue that - in general - fully understanding how a game works typically makes someone _more_ likely to enjoy a game, and want to play it, rather than less likely. Tyrese Haliburton knowing _why_ his team is rated at 77-5 instead of 82-0 wasn't likely to ruin it for him (in fact, it might have delayed the rage quit!), and I'd expect the same could be said for many other people who shouted "NO WAY this team loses [_supposedly absurd number_] games!!" at their phones. When the point of a game is trying to get the highest possible score, it's nice to know how that score is calculated. (Also - for what it's worth - the "secret sauce" was hardly under lock and key as of the time of Almog's interview. As Chrome DevTools and the Wayback Machine can show you, it was at the time explicitly laid out in the source code for the site.) Second, in the specific case of 82-0, knowing the algorithm does very little to change the game experience because of the randomness baked into game play. I don't choose what eras or teams I select players from - the game chooses those for me, part of the fun/frustration is having to work within those constraints. There's nothing, really, that I can do to "game the system" as long as I play the game how it's designed. What I _can_ do, if I know the algorithm, is make more intelligent choices about which players to select. (For example - Bill Russell isn't even visible on my screen by default when the 1960s Celtics come up, but mayyyyyybe I should scroll down and pick him instead of John Havlicek.) But I'm not primarily interested in understanding the 82-0 algorithm to get better at playing 82-0. I come at this from the perspective of someone who's both a basketball fan and who works with data. I find it fascinating to dig deeper into how one group of people, the creators of 82-0, have approached a very complex basketball data problem: how _could_ you predict or project a hypothetical NBA team record from the few individual player statistics we have across all eras, when combining players from different eras, in spite of the dramatic evolution of basketball between the 1960s and today? Yes, 82-0 is just a game, and yes, an answer to this question that makes for a good _game_ might be very different than one which is maximally accurate! Shocking, I know. But understanding how 82-0 works is a great springboard to think a bit more seriously about various aspects of the problem: what data _should_ we use, and why? What can we do when we're missing data? _Can_ individual player statistics be used to predict anything meaningful about team success in the first place and, if so, how? More specific to 82-0 itself, knowing how to go 82-0 helps us do some really fun analysis on the complete set of all possible 82-0 teams.34:T7f7,The datapoints should be reasonably familiar to anyone who follows basketball, but if you want to see detailed definitions and notes for each one, toggle the heading below!

Definitions

id: a unique identifier for each record built from the player name, team, and era.
team: the three-letter abbreviation for the NBA franchise the player played for - for example, "ATL" for the Atlanta Hawks.
era: the timeframe in which the player accumulated statistics for a particular team. Equivalent to a specific decade, 1960s through 2020s.
positions: the positions for which a player is eligible to be selected (based on the positions at which a player actually played over the course of their career). Available options are PG (Point Guard), SG (Shooting Guard), SF (Small Forward), PF (Power Forward), and C (Center).
name: the name of the player.
ppg: Points Per Game, the average number of points a player scored per game for a given team in a given era.
rpg: Rebounds Per Game; same calculation as **ppg** but for rebounds.
apg: Assists Per Game; same as **ppg** but for assists.
spg: Steals Per Game; same as **ppg** but for steals.
bpg: Blocks Per Game; same as **ppg** but for blocks.

A note on **team** - if a player played for a franchise that is now defunct, or moved, they are assigned to the modern-day equivalent franchise. For example, Wilt Chamberlain shows up with "Golden State" (GSW) in the 60s rather than the Philadelphia Warriors.

A note on **averages** - in some cases it looks like the stats are calculated as weighted averages, taking into account number of games played; in others, however, it looks like they are simple arithmetic means of season summary stats. I won't explore that further here, but just to explain an apparent inconsistency in "average".

35:T424,If you're not fluent in JavaScript, the code might obfuscate a pretty straightforward process. The function goes through these steps: 1. Create two lists - one of the **spg** values in your roster that are greater than 0, and another of the same kind for **bpg**. 2. Calculate the length of each list. (Those happen always to be the same; every player in the source data missing **spg** is also missing **bpg**, and vice versa). 3. For each list: add all the numbers together, multiply by 5, and divide by the number of values in the list. Those two resulting numbers are your adjusted team SPG and BPG, respectively, that get used in the team score calculation. (If _every_ player has 0 or ```null``` for **bpg** and **spg**, you instead multiply the number of values in each list - 0, since they're empty - by 1, meaning that your adjusted team SPG and BPG are both 1 * 0 = 0. Yes, that's weird - a bug or mistake, even - and unfortunate in the extremely improbable event that you draft an entire 1960s team! But in the typical case that's not what happens.)36:Tb43,Plot.plot({ marks: [ Plot.line( d3.range(0, 145, 0.1), { x: (x) => x, y: (x) => calculateWins(x, 110), stroke: "black", strokeWidth: 1, tip: true } ), //Perfect Plot.areaY( d3.range(107.1, 145, 0.1), { x: (x) => x, y: (x) => calculateWins(x, 110), fill: teamTiers.perfect.color, stroke: teamTiers.perfect.color, opacity: 0.3 } ), Plot.ruleX([109.5], { stroke: teamTiers.perfect.color, strokeWidth: 2 }), //Historic Plot.areaY( d3.range(97.7, 107.1, 0.1), { x: (x) => x, y: (x) => calculateWins(x, 110), fill: teamTiers.historic.color, stroke: teamTiers.historic.color, opacity: 0.5 } ), //Dynasty Plot.areaY( d3.range(86.9, 97.7, 0.1), { x: (x) => x, y: (x) => calculateWins(x, 110), fill: teamTiers.dynasty.color, stroke: teamTiers.dynasty.color, opacity: 0.5 } ), //Contender Plot.areaY( d3.range(79.6, 86.9, 0.1), { x: (x) => x, y: (x) => calculateWins(x, 110), fill: teamTiers.contender.color, stroke: teamTiers.contender.color, opacity: 0.3 } ), //Playoff Plot.areaY( d3.range(71.0, 79.6, 0.1), { x: (x) => x, y: (x) => calculateWins(x, 110), fill: teamTiers.playoff.color, stroke: teamTiers.playoff.color, opacity: 0.3 } ), //Lottery Plot.areaY( d3.range(58.3, 71.0, 0.1), { x: (x) => x, y: (x) => calculateWins(x, 110), fill: teamTiers.lottery.color, stroke: teamTiers.lottery.color, opacity: 0.3 } ), //Tanking Plot.areaY( d3.range(0, 58.3, 0.1), { x: (x) => x, y: (x) => calculateWins(x, 110), fill: teamTiers.tanking.color, stroke: teamTiers.tanking.color, opacity: 0.3 } ), Plot.ruleX([0]), Plot.ruleY([0]), Plot.axisX({ fontSize: 12, labelAnchor: "center" }), Plot.axisY({ fontSize: 12, labelArrow: "none" }), Plot.tip( [`In the Version 110 wins\ncalculation, a team score\nof 109.5 is the lowest\npossible score that will\nget you to to 82-0.`], {x: 109.5, y: 70, dy: 8, dx: 8, anchor: "left" } ), ], x: { label: "Team Score" }, y: { label: "Win Total", grid: true }, marginBottom: 40, marginTop: 40, color: { legend: true, type: "categorical", domain: Object.keys(teamTiers).map(key => teamTiers[key].grade), range: Object.keys(teamTiers).map(key => teamTiers[key].color), fontSize: 12 }, subtitle: "Calculating wins from team scores" })37:T61e,Plot.plot({ marks: [ Plot.line( d3.range(70, 125, 0.1), { x: (x) => x, y: (x) => Math.min(Math.round(((x / 110) ** 1.15) * 82, 1), 82), stroke: "lightgrey", strokeWidth: 2 } ), Plot.line( d3.range(70, 125, 0.1), { x: (x) => x, y: (x) => Math.min(Math.round(((x / 116) ** 1.15) * 82, 1), 82), stroke: "steelblue", strokeWidth: 1.5 } ), Plot.dot(version_116_scores_and_wins, { x: "score", y: "wins", stroke: "red", fill: "red", r: 3, tip: true }), Plot.tip( [`Each red dot represents\none team score / team\nwins combination from the\nsample games I played.\nThis dot corresponds to\nthe team pictured above.`], {x: 98.3, y: 68, anchor: "top-left" } ), Plot.tip( [`In Version 116,\nyou can't go\n82-0 without a\nteam score of at\nleast 115.4.`], {x: 115.4, y: 82, anchor: "top-left" } ), Plot.ruleX([0]), Plot.ruleY([0]), Plot.axisX({ fontSize: 12, labelAnchor: "center" }), Plot.axisY({ fontSize: 12, labelArrow: "none" }) ], x: { label: "Team Score", domain: [70, 125] }, y: { label: "Win Total", grid: true, domain: [45, 85] }, marginBottom: 40, marginTop: 40, color: { legend: true, type: "categorical", range: ["lightgrey", "steelblue"], domain: ["Version 110 Team Wins", "Version 116 Team Wins"], fontSize: 12 } })38:[{"id":"feec9dd33f899fbc13c3288d81ff6b7b06ea53a5fc1527f1f87c94cc82154a2aadd9114ec6c5542d36e1cc2f771a12924272867997d627f146e2185f85cd0b97","href":"https://static.observableusercontent.com/files/feec9dd33f899fbc13c3288d81ff6b7b06ea53a5fc1527f1f87c94cc82154a2aadd9114ec6c5542d36e1cc2f771a12924272867997d627f146e2185f85cd0b97","name":"82_0_player_data_6_22.json","type":"application/json","lastModified":1790776384000,"size":1757569},{"id":"57a895f1d4010d7036a45e3d64adf66979494ad172e449794d237210656e4ac155fcafd47bdfc5fb6b2427250ee1cb566423e1ceb5301743254e1c5abaa910d3","href":"https://static.observableusercontent.com/files/57a895f1d4010d7036a45e3d64adf66979494ad172e449794d237210656e4ac155fcafd47bdfc5fb6b2427250ee1cb566423e1ceb5301743254e1c5abaa910d3","name":"68-14 result 10-1-26.png","type":"image/png","lastModified":1790871803000,"size":254763}] 30:{"id":"7097cad0ba1f9797","slug":"project-82-0-part-1","title":"Project 82-0, Part 1","thumbnail":"49e3b8e7b5c25c76853ac31b4ed41e895e2facb91573182f4e68e995d0ccb899","access_level":"public","create_time":"$D2026-09-23T18:15:15.972Z","update_time":"$D2026-10-09T16:19:30.007Z","view_time":null,"first_public_time":"$D2026-10-09T02:27:35.730Z","has_unpublished_edits":null,"fork":false,"stars":0,"starred":false,"owner":{"type":"team","id":"030c90355475f8d1","login":"gideonkylejeffrey","name":"Gideon Jeffrey","avatar_url":null,"tier":"starter_2024","bio":""},"version":1428,"publish_version":1428,"role":null,"authored":false,"authors":[{"type":"individual","id":"55b53f565c15d32f","login":"gideonkylejeffrey","name":"Gideon Jeffrey","avatar_url":null,"tier":null,"bio":""}],"members":[{"type":"team","id":"030c90355475f8d1","login":"gideonkylejeffrey","name":"Gideon Jeffrey","avatar_url":null,"tier":"starter_2024","role":"editor"}],"comments":[],"custom_thumbnail":false,"trashed":false,"lapsed":null,"subscribed":null,"forks":0,"collection":null,"type":"notebook","body":{"titling":"auto","title":"Project 82-0, Part 1","cells":[{"id":1,"value":"# Project 82-0, Part 1\n\n#### *A recipe for the \"secret sauce\" powering [the viral online game 82-0](https://www.82-0.com).*\n\n
\n\n_October 1st, 2026_","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":82,"value":"toc()","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":19,"value":"## Why this project?\nEarlier this year, late at night and recovering from jet lag, I was scrolling on my phone and saw some chatter about about a new basketball-themed game called 82-0. By the time I became aware of it, Indiana Pacers guard Tyrese Haliburton had already rage quit:","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":22,"value":"$32","pinned":false,"hidden":false,"mode":"html","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":25,"value":"I decided to check it out for myself, and immediately lost an hour of sleep to this distressingly addicting game.\n\n","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":29,"value":"For anyone who didn't get sucked in like I did, the easiest way to understand 82-0 is to play it - either [the current version](https://www.82-0.com) or [the original that first went viral](https://web.archive.org/web/20260602215953/https://www.82-0.com/), still available thanks to the Internet Archive's Wayback Machine! But here's how it works: you participate in a five-round draft to try to create the best team of five NBA players, past or present, that you can. In each round you must pick one player from a specific, randomly chosen franchise and decade. Once you've chosen a complete lineup, one player per court position (point guard, shooting guard, small forward, power forward, and center), the game calculates an overall score for your team and translates that score into a win-loss record. Maybe, just maybe, the game will decide that your chosen players are good enough to achieve a perfect 82-0 regular NBA season.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":28,"value":"There are many reasons why this game was such a hit. One was the game design itself: beautifully simple and fast to play, shareable lineups so you could compete with friends for the best records, and just the right level of difficulty - very challenging but clearly (given that _some_ braggadocious people were going 82-0!) not impossible. It was also a great opportunity to \"remember some guys\" or discover ones you didn't know about in the first place - _Truck Robinson! World B. Free! **FAT LEVER**!_ But more than anything else, it provided a shared framework for NBA fans to engage in their favorite what-ifs - _imagine Wilt and LeBron on the same team!_ - while at the same time reigniting perennial debates about who the best players actually are, or how NBA greatness should be measured in the first place.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":43,"value":"And on that topic - a big part of the game's intrigue was that its algorithm wasn't obvious. Your team score and win-loss record definitely depended on individual player statistics, but exactly how wasn't clear. Bar Almog, CEO of the company platforming the game and friend of the game's creator Roy Saar, was quite cagey [when asked about it](https://allphly.com/76ers/82-0-nba-game-starting-lineup-viral-interview/):","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":30,"value":">\n> \"That’s the secret sauce. We’re always working to improve it and add more statistics and then make sure that the data that sits behind it is the most accurate and relevant, but that’s definitely our secret sauce.\"\n>","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":31,"value":"As it turns out, however, the sauce isn't that secret (yet). In this project, I'll start by giving you the recipe - in fact, two versions of it! True to Almog's word, the algorithm _has_ changed over time, and may have changed again by the time you read this - but I can reproduce at least two versions that I know to have been actually used, one which I'll call \"Version 110\" (dating back to at least June 2nd, when the game went viral) and another which I'll call \"Version 116\" (which is live as of writing on October 1st, 2026). The reason for the names will become obvious later.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":33,"value":"$33","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":36,"value":"More on those things in future posts, though. For now, my goal is _only_ to explain how, in fact, two actual versions of the 82-0 secret sauce get made, though you'll find that lots of interesting questions come up about both the ingredients and the recipe. Thinking about what alternative recipes might look like is subject matter for a different time.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":24,"value":"## Ingredients","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":92,"value":"Let's start with the data.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":50,"value":"Every datapoint that 82-0's algorithm uses to generate scores and win-loss records is contained in a JSON file loaded by the site. That file contains individual player information and statistics for 10,626 unique combinations of *player*, a *team* that the player played on, and an *era* during which they played for that team (a decade between the 1960s and 2020s). I've replicated that JSON file and added it as a source to this notebook. Here's what the first record in the dataset looks like, for Al Ferrari's tenure with Atlanta in the 1960s:","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":45,"value":"playerData = FileAttachment(\"82_0_player_data_6_22.json\").json()","pinned":false,"hidden":true,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":51,"value":"al_ferrari_atl_1960s = playerData.players[0]","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":55,"value":"$34","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":60,"value":"You can explore the full dataset using the table below. One thing you'll see right away is that any record from the 1960s, and some from the 1970s, doesn't have any values for either **bpg** or **spg**. This is thanks to the fact that the NBA didn't officially begin recording blocks and steals [until the 1973-74 season](https://www.nba.com/stats/help/faq). As we'll see, the team score calculation has an \"estimation\" method of sorts to handle missing data.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":49,"value":"playerDataFlattened = playerData.players.map(({ stats, ...otherColumns }) => ({\n ...otherColumns,\n ppg: stats.ppg,\n rpg: stats.rpg,\n apg: stats.apg,\n spg: stats.spg,\n bpg: stats.bpg\n}))","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":47,"value":"viewof playerDataTable = Inputs.table(playerDataFlattened, {\n layout: \"auto\"\n})","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":61,"value":"For the purpose of the team score calculation, only the five statistics - **ppg**, **rpg**, **apg**, **spg**, and **bpg** - matter. **era** and **position** constrain which players you can select, but don't influence team score at all, as we'll see below.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":39,"value":"## Step 1: Calculate team score","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":62,"value":"The team score calculation works the same for both Version 110 and Version 116. Let's use Tyrese Haliburton's viral lineup to illustrate it:","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":65,"value":"teamTyrese = [\n \"lebron_james_cle_2000s\",\n \"kobe_bryant_lal_2010s\",\n \"michael_jordan_chi_1990s\",\n \"larry_bird_bos_1980s\",\n \"moses_malone_hou_1980s\"\n]","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":64,"value":"viewof teamTyreseTable = Inputs.table(playerDataFlattened.filter(d => teamTyrese.includes(d.id)), {\n layout: \"auto\",\n columns: [\"name\", \"team\", \"era\", \"ppg\", \"rpg\", \"apg\", \"spg\", \"bpg\"]\n})","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":63,"value":"The first step, assuming all stats exist for all five players (we'll talk about cases of missing blocks and steals in a minute), is to simply add them all together. Here's how the totals end up looking for Team Tyrese:","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":71,"value":"teamTyreseStats = ({\n ppg: Math.round(d3.sum(playerDataFlattened.filter(d => teamTyrese.includes(d.id)), e => e.ppg) * 100) / 100,\n rpg: Math.round(d3.sum(playerDataFlattened.filter(d => teamTyrese.includes(d.id)), e => e.rpg) * 100) / 100,\n apg: Math.round(d3.sum(playerDataFlattened.filter(d => teamTyrese.includes(d.id)), e => e.apg) * 100) / 100,\n spg: Math.round(d3.sum(playerDataFlattened.filter(d => teamTyrese.includes(d.id)), e => e.spg) * 100) / 100,\n bpg: Math.round(d3.sum(playerDataFlattened.filter(d => teamTyrese.includes(d.id)), e => e.bpg) * 100) / 100\n})","pinned":false,"hidden":true,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":80,"value":"viewof teamTyreseAggTable = Inputs.table([teamTyreseStats], {\n layout: \"auto\",\n width: 450,\n header: {\n \"ppg\": \"Team PPG\",\n \"rpg\": \"Team RPG\",\n \"apg\": \"Team APG\",\n \"spg\": \"Team SPG\",\n \"bpg\": \"Team BPG\",\n }\n})","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":68,"value":"Only two additional components are required to create the team score formula. First, each stat is divided by a *baseline* value specific to that stat. Second, the resulting value is multiplied by a particular *weight* out of 100% that determines how much it contributes to the overall team score. Here are the baselines and weights used by both versions of 82-0 discussed here:","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":73,"value":"Inputs.table(\n [\n {\"Metric\": \"ppg\", \"Baseline\": 133.4, \"Weight\": 0.46},\n {\"Metric\": \"rpg\", \"Baseline\": 39.7, \"Weight\": 0.25},\n {\"Metric\": \"apg\", \"Baseline\": 29.3, \"Weight\": 0.18},\n {\"Metric\": \"spg\", \"Baseline\": 6.1, \"Weight\": 0.07},\n {\"Metric\": \"bpg\", \"Baseline\": 3.2, \"Weight\": 0.04}\n ], {\n width: 400,\n format: {\n // Replace \"PercentColumn\" with the exact name of your data key/column\n Weight: d3.format(\".0%\")\n }\n }\n)","pinned":false,"hidden":false,"mode":"ts","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":84,"value":"Using these baseline values and weights, we can now reproduce 82-0's team score formula:","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":86,"value":"\\text{Team Score} = 0.46(\\frac{\\text{Team PPG}}{133.4})\n\\\\~\\\\\n+ 0.25(\\frac{\\text{Team RPG}}{39.7})\n\\\\~\\\\\n+ 0.18(\\frac{\\text{Team APG}}{29.3})\n\\\\~\\\\\n+ 0.07(\\frac{\\text{Team SPG}}{6.1})\n\\\\~\\\\\n+ 0.04(\\frac{\\text{Team BPG}}{3.2})","pinned":false,"hidden":false,"mode":"tex","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":87,"value":"In the old source code for 82-0, the result is rounded to one decimal, and so a JavaScript implementation of the function would look something like this:","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":75,"value":"function calculateTeamScore(p, r, a, s, b) {\n /*\n p = team ppg; r = team rpg; a = team apg; s = team spg; b = team bpg\n */\n return Math.round(100 * (p / 133.4 * 0.46 + r / 39.7 * 0.25 + a / 29.3 * 0.18 + s / 6.1 * 0.07 + b / 3.2 * 0.04) * 10) / 10\n}","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":88,"value":"At long last, we can calculate the team score for Team Tyrese and confirm that it matches what he shared on X.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":89,"value":"teamTyreseScore = calculateTeamScore(teamTyreseStats.ppg, teamTyreseStats.rpg, teamTyreseStats.apg, teamTyreseStats.spg, teamTyreseStats.bpg)","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":91,"value":"The \"share\" view used by 82-0 rounds team score to the nearest integer - in this case 103.8 rounds to 104, exactly what we see in the screenshot.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":93,"value":"## Handling missing blocks and steals","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":94,"value":"The formula above works perfectly so long as we have all datapoints for all players - but of course, that's not always true. What should we do if our lineup includes early-era players who are missing **spg** and **bpg**? According to the explanatory text in the old version of the 82-0 site, \"missing defensive stats from older eras are estimated fairly\". Well, here's what that fair estimation amounts to.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":107,"value":"There’s a function, ```adjustSpgBpg```, in the old source code that I've reproduced below and which we can use to calculate the team adjusted SPG and BPG:","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":96,"value":"function adjustSpgBpg(team) {\n let validSteals = team.filter(e => e.spg > 0).map(e => e.spg)\n , validBlocks = team.filter(e => e.bpg > 0).map(e => e.bpg)\n , slength = validSteals.length\n , blength = validBlocks.length;\n return {\n adjustedSpg: validSteals.reduce( (e, t) => e + t, 0) * (slength > 0 ? 5 / slength : 1),\n adjustedBpg: validBlocks.reduce( (e, t) => e + t, 0) * (blength > 0 ? 5 / blength : 1)\n }\n}","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":108,"value":"$35","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":110,"value":"Here's what that looks like in a specific case, using a sample roster from the team [in this Tweet](https://x.com/SleeperOrlando/status/2062957130575929814?s=20):","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":103,"value":"teamSleeperMagic = [\n \"oscar_robertson_sac_1970s\",\n \"michael_jordan_chi_1980s\",\n \"kevin_durant_phx_2020s\",\n \"gerald_wallace_cha_2000s\",\n \"wilt_chamberlain_gsw_1960s\"\n]","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":104,"value":"viewof teamSleeperMagicTable = Inputs.table(playerDataFlattened.filter(d => teamSleeperMagic.includes(d.id)), {\n layout: \"auto\",\n columns: [\"name\", \"team\", \"era\", \"ppg\", \"rpg\", \"apg\", \"spg\", \"bpg\"]\n})","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":106,"value":"teamSleeperMagicAdjustedSPGBPG = adjustSpgBpg(playerDataFlattened.filter(d => teamSleeperMagic.includes(d.id)))","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":111,"value":"To understand what this function does in another way: it effectively sets **spg** and **bpg** for players _without_ them to the average value of players _with_ them. I'll let you check the math in our example above to see that we get the same values by setting **spg** and **bpg** for both Wilt and Oscar to the averages (1.97 + 2.81 + 0.8)/3 and (1.21 + 1.18 + 1.3)/3, respectively, and then adding all of the values up like we did in our earlier formulas for Team SPG and Team BPG.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":102,"value":"Just to check that our team score calculation will match what's in the tweet (rounded to 114), let's run the numbers:","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":116,"value":"teamSleeperMagicStats = ({\n ppg: Math.round(d3.sum(playerDataFlattened.filter(d => teamSleeperMagic.includes(d.id)), e => e.ppg) * 100) / 100,\n rpg: Math.round(d3.sum(playerDataFlattened.filter(d => teamSleeperMagic.includes(d.id)), e => e.rpg) * 100) / 100,\n apg: Math.round(d3.sum(playerDataFlattened.filter(d => teamSleeperMagic.includes(d.id)), e => e.apg) * 100) / 100,\n spg: teamSleeperMagicAdjustedSPGBPG.adjustedSpg,\n bpg: teamSleeperMagicAdjustedSPGBPG.adjustedBpg,\n})","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":115,"value":"calculateTeamScore(teamSleeperMagicStats.ppg, teamSleeperMagicStats.rpg, teamSleeperMagicStats.apg, teamSleeperMagicStats.spg, teamSleeperMagicStats.bpg)","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":95,"value":"Before moving on: I already mentioned the unfortunate consequence of this method that if you draft an all-early-era team, congratulations! Everyone's steals and blocks will remain at 0. Beyond that, there are two other... interesting, I'll say? features of this method worth highlighting. First, position isn't taken into account at all; Wilt and Oscar are \"equally good\" at getting steals and blocks when in the same lineup. Second, **spg** and **bpg** for early-era players are completely dependent on who else gets drafted, and so they will fluctuate dramatically from lineup to lineup.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":120,"value":"It's reasonable to wonder whether there's a better way to handle these cases. Let's set that aside for now, though, so we can get to the final step: turning team score into a win-loss record.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":41,"value":"## Step 2: Derive win-loss record (Version 110)","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":122,"value":"82-0's \"How to Play\" page hints that \"the relationship between stats and wins is not linear.\" Here's the \"Version 110\" edition of that non-linear function for converting team score into wins, used on the site as of June 2nd of 2026 and for at least a couple months after that.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":123,"value":"\\text{Team Wins}_{v110} = 82 * (\\min(\\frac{\\text{Team Score}}{110}, 1)^{1.15})","pinned":false,"hidden":false,"mode":"tex","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":124,"value":"So - step by step: first, divide team score by a baseline of 110 (and yes, that's where the very creative name \"Version 110\" comes from). Take whichever is smallest, the result or the number 1, and raise that to the power 1.15. Finally, multiply by 82, and there we have your lineup's projected total wins. It should basically go without saying, but of course Team Losses is just 82 minus Team Wins, and we combine the two to get the projected record. In practice, as in the JavaScript implementation of the function below, we'll want to round Team Wins to the nearest integer because \"partial wins\" aren't a thing:","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":16,"value":"function calculateWins(s, b) {\n return Math.min(Math.round(((s / b) ** 1.15) * 82, 1), 82)\n}\n","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":125,"value":"And now we can confirm that, yes, Team Tyrese was destined for 77-5:","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":27,"value":"teamTyreseVersion110Wins = calculateWins(teamTyreseScore, 110)","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":126,"value":"I think it's helpful to visualize the Version 110 team wins formula to get a better sense of the relationship between team score and team wins. You'll see that I've color coded different bands according to the \"grade\" you'd receive for a given team in Version 110.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":2,"value":"$36","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":3,"value":"## Step 2, Revised: Derive win-loss record (Version 116)","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":127,"value":"As I began to play 82-0 again recently to work on this project, I noticed something fascinating: the game seemed significantly harder than it had been previously. Take this example game I played on October 1st:","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":141,"value":"FileAttachment(\"68-14 result 10-1-26.png\").image({\n height: 400,\n alt: \"A roster of 20s Portland Dame, 20s Dallas Luka, 90s Washington Bernard King, 90s Spurs Duncan, and 80s Philly Moses goes 68-14.\"\n})","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":142,"value":"The exact team score was 98.3, which - using the version 110 formula - ought to produce a record of 72-10, not 68-14!","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":128,"value":"calculateWins(98.3, 110)","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":131,"value":"So the game creators had clearly changed something - but what, exactly? The easiest way to figure this out, I decided, was to play a range of games, look at the resulting team score and win totals, and plot those values to try to figure out what the change had been. I also assumed to start that the change was probably simple - either the baseline 110 was different, or the exponent 1.15 was different. It turns out that assumption was correct: all evidence points to the fact that the baseline changed from 110 to 116.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":132,"value":"\\text{Team Wins}_{v116} = 82 * (\\min(\\frac{\\text{Team Score}}{116}, 1)^{1.15})","pinned":false,"hidden":false,"mode":"tex","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":133,"value":"The actual formula for team wins is not in the public-facing 82-0 code anymore, so I can't confirm this directly. But I think playing the game repeatedly gives us proof enough that 116 is the baseline as of October 1st, 2026. The plot below shows a line for the old Version 110 calculation, a line for the hypothesized Version 116 calculation, and the results of the 20 games I played. I think it's pretty compelling:","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":15,"value":"version_116_scores_and_wins = [\n {\"score\": 106.9, \"wins\": 75},\n {\"score\": 100.4, \"wins\": 69},\n {\"score\": 98.3, \"wins\": 68},\n {\"score\": 61.4, \"wins\": 39},\n {\"score\": 72, \"wins\": 47},\n {\"score\": 84.9, \"wins\": 57},\n {\"score\": 87.1, \"wins\": 59},\n {\"score\": 84.2, \"wins\": 57},\n {\"score\": 77.4, \"wins\": 51},\n {\"score\": 103.9, \"wins\": 72},\n {\"score\": 89.9, \"wins\": 61},\n {\"score\": 96.1, \"wins\": 66},\n {\"score\": 96.4, \"wins\": 66},\n {\"score\": 100.8, \"wins\": 70},\n {\"score\": 103.2, \"wins\": 72},\n {\"score\": 89, \"wins\": 60},\n {\"score\": 87.7, \"wins\": 59},\n {\"score\": 90.6, \"wins\": 62},\n {\"score\": 108.3, \"wins\": 76},\n {\"score\": 93.7, \"wins\": 64},\n]","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":9,"value":"$37","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":134,"value":"So the last step of the recipe, at least, _is_ different than it used to be. And it is, on October 1st, _significantly_ harder to go 82-0 than it was in early June when the game went viral; you now need a very high team score of at least 115.4 to do so. It's hard to know what motivated the game's creators to make this change; perhaps they thought that we all just had it too easy before! Tyrese Haliburton, whose lineup would lose *ten* games instead of five if he created it on October 1st, would probably disagree.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":136,"value":"teamTyreseVersion116Wins = calculateWins(teamTyreseScore, 116)","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":145,"value":"_**Update October 8th, 2026** - it appears the algorithm has now changed **again** to a \"Version 120\" - team score baseline of 120, rather than 116, used in the wins calculation. As before, this makes it significantly harder to actually go 82-0. As before, hard to know exactly what's motivating the change!_","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":137,"value":"## Try for Yourself","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":143,"value":"So there you have it - 82-0's secret sauce demystified. Perhaps, though, you're not yet convinced. Maybe you'd like to run one of your own actual lineups through the team score and total wins calculations. Maybe you're curious how _any_ hypothetical lineup would fare, on either Version 110 or Version 116, without having to play 82-0 thousands of times! If you fall into any of those categories, check out my 82-0 simulation notebook. It will let you build whatever lineup you'd like and instantly see what its team score and record would be, according to both the Version 110 or Version 116 calculations. I hope you enjoy exploring the quadrillions of possibilities!\n\n*****","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":144,"value":"
\n
\n\n#### Additional code and configuration","pinned":true,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":14,"value":"teamTiers = ({\n perfect: {\n minWins: 80,\n color: \"#a855f7\",\n grade: \"S\"\n },\n historic: {\n minWins: 72,\n color: \"#22c55e\",\n grade: \"A+\"\n },\n dynasty: {\n minWins: 62,\n color: \"#85eaaa\",\n grade: \"A\"\n },\n contender: {\n minWins: 57,\n color: \"#3b82f6\",\n grade: \"B\"\n },\n playoff: {\n minWins: 50,\n color: \"#f59e0b\",\n grade: \"C\"\n },\n lottery: {\n minWins: 40,\n color: \"#64748b\",\n grade: \"D\"\n },\n tanking: {\n minWins: 0,\n color: \"#ef4444\",\n grade: \"F\"\n }\n})","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":83,"value":"function toc(selector = \"h2\", heading = html`Table of Contents`) {\n //modified from Mike Bostock's original at https://observablehq.com/@mbostock/toc\n return Generators.observe(notify => {\n let headings = [];\n\n function observed() {\n const h = Array.from(document.querySelectorAll(selector));\n if (h.length !== headings.length || h.some((h, i) => headings[i] !== h)) {\n notify(html`${heading}

${DOM.text(h.textContent)}`\n );\n })}`);\n }\n }\n\n const observer = new MutationObserver(observed);\n observer.observe(document.body, {childList: true, subtree: true});\n observed();\n return () => observer.disconnect();\n });\n}","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"}],"files":"$W38","stdlib":"2"}} 31:"$30:body"