Hoops Auction About

About this site

Hoops Auction is a free basketball auction game that runs in a browser. Two people, or a league of up to eight, bid against each other for real players, one at a time, and the rosters they buy play a simulated game, a round robin or an 82-game season that settles it. There is no sign-up, no download and no account.

Start an auction

Who makes it

It is built and maintained by Armaan Ebrahim, working alone. It is not a company, a team or a studio, and there is nobody else to escalate to — if something on this site is wrong, the person who can fix it is the person who wrote it. The email address on the contact page reaches me directly.

It is an independent project with no affiliation to, endorsement from, or relationship with the National Basketball Association, Basketball Reference, or any team, player or league. All trademarks belong to their owners. Nothing here is official.

Why it exists

Fantasy drafts and salary-cap modes are usually a side dish inside a much larger game, and they almost always assign player values by hand. That makes the argument afterwards unwinnable in a boring way: if a rating is somebody's opinion, disagreeing with it is just two opinions.

So this is the auction on its own, free, in a browser, with the ratings derived from a published statistical method instead of assigned. When the game says a player is worth eighteen million, that number came out of a pipeline you can read about. You are welcome to think it is wrong — but you can argue with the method rather than with my taste, and that is a far better argument.

Where the numbers come from

Every rating in the game is computed from public per-game statistics, not entered by hand. The short version:

The full method, including what it gets wrong and why the all-time list needs an accolades component at all, is written up in how the ratings work. What each player is then worth at auction is a separate calculation, set out in full in how players are priced.

Data and images

Player statistics are derived from a publicly available dataset compiled from Basketball Reference and processed before the game ships. Only computed per-game figures are published here; the source files are not redistributed.

The game itself shows no photographs of players. A Creative Commons licence clears a photographer's copyright, but not the player's right to his own image, and a real athlete's likeness used as a piece in a sports game is the use courts have been least willing to excuse. So a lot card identifies a player by his name and his numbers, with his initials where a photograph might go. Basketball GM ships its real-player leagues the same way.

The one place portraits appear is the all-time list, an editorial ranking, where 37 Creative Commons photographs from Wikimedia Commons show who is being ranked. Each is credited with its author, a link to the original and its licence. No NBA or Basketball Reference imagery is used anywhere on this site.

What this site will not claim

The project is new and small. There are no user numbers to boast about, no testimonials, and no awards. If you ever see this site claim otherwise, it is a mistake and I would like to know.

Equally, nothing here is a prediction. The simulated game is a model built from rating and fit; it settles an argument between two rosters and says nothing about what would happen on a real court. It is not betting advice and there is nothing to stake.

How it is paid for

The game is free and intended to stay that way. Hosting runs on a free tier, and the site is set up to carry advertising to cover what it costs. There is no premium tier, no subscription and nothing to buy.

There are no accounts, and the only thing a player types is a name for their seat, which is deleted with the room. What is and is not collected — including what advertising and analytics see — is set out in the privacy policy.

Corrections

If a player's rating looks wrong to you, that is a legitimate thing to write in about, and the useful version of that email names the player, the scale you were on, and what you expected instead. Tuning happens in one published configuration file where the reasoning is visible, and changes are checked against a traditional-consensus test rather than nudged until they look right.

The quickest way to judge any of this is to play a sale and see whether the prices feel defensible.

Start an auction

How the ratings work How players are priced How to play The all-time list FAQ Contact Privacy Terms

Start an auction