About
Contents: Who I am · What Chromahall is · The Chromahall number · How to read a film's specs · Your ratings · Your profile · Break your ties · Recommendations · Data & credits · Personal & non-commercial · Contact
Who I am
I'm James. I like statistics a little too much. Chromahall started from a small itch: your 4 and my 4 rarely mean the same thing. I might hand out a 4 where you'd give a 3 — same number, different yardstick. So I spent a long time analyzing ratings to put them all on one comparable scale, and built the algorithm that does it. The site you're on is the next step: a place to refine that reading, live, as people rate.
What Chromahall is
Chromahall is an algorithm. It doesn't rate films — it reads how you rate them, measures whether you score high or low against the full distribution, and rescales your ratings so they're comparable with everyone else's. Your score isn't touched; it's placed on a common axis. That's computation, not opinion.
The Chromahall number
The number in the box, labelled CHROMAHALL, is a percentile on a 0-to-100 scale. To get it, every rating is first put on the same scale — each voter's habit of scoring high or low is levelled out — and films are then ordered by those calibrated ratings. The number says where a film sits in that order. It isn't an average, so it won't match the star averages you see elsewhere: it measures position, not the mean of the votes.
Beside it sits the RAW number: the same 0-to-100 position, taken from each film's plain average of the ratings exactly as they were cast, with no calibration. The movement between the two, in positions, is what calibration changes: how many places a film climbs or drops once every rating is read on the same scale. Not a verdict on the film, and not a claim about its fans: a measurement.
How to read a film's specs
Open any film with a CHROMAHALL number and tap Specs → to turn the card over. The back has four rows. Every one of them comes from analysing ratings with my own model. Three of them carry a small strip: the bars show how every film in the ranking spreads on that measure, and the line marks where this film sits.
Net Taste. Won over is the share of this film's raters who liked it more than they usually like a film — each by their own standard; let down, the share who liked it less. Net Taste is the difference: a positive number means it won over more people than it let down, a negative one the opposite. The strip runs from −100 to +100, with the gold line at zero.
Rater leniency. How generously this film's raters score in general, across everything they rate, in stars. Strict audiences sit to the left, lenient ones to the right. It describes the group, never any one person.
Audience depth. How many films its raters have typically rated. A high number means the film is mostly rated by people who watch and rate a lot.
Precision. How firm the CHROMAHALL number is: the margin it stays within in 95% of simulations. "±4 pts" means it could reasonably sit four points higher or lower; "under ±1 pt" means it barely moves.
Your ratings
There are three ways to bring your ratings in. Import the history you already have from Letterboxd or IMDb, using the export file each site gives you, from Build your profile. Use Build your profile to rate films picked for one job: telling your taste apart quickly. They are well-known films that divide opinion, so most people have seen them and every rating says a lot about where you stand. Anything you've already rated is left out, so after an import you only see what's new to your profile; skip any you haven't seen. Or rate any film straight from its card, in half-stars.
My Ratings lists everything you've rated. Next to each film it shows how far your rating sits from the crowd's, in stars. You can sort the list, filter it by stars, decade, year or genre, and export it as a file at any time.
Your profile
Rate enough films and your profile builds itself from your ratings. It gives you a type — how you rate and where you part from the crowd — and three readings around it.
Radar. Your taste by genre, set against the crowd's: the genres you defend, the ones you're harder on, and where you stand apart.
Your own read. The films where your rating, on the common scale, sits furthest above the crowd's, and the ones where it sits furthest below.
Crew. The people whose ratings move most like yours. You never see who they are; their agreement is what powers your recommendations.
Your profile also comes as a Profile card you can download and share.
Break your ties
When you've given several films the same rating, Break your ties shows them to you in pairs and asks which one you rate higher. You can call it even, or skip a film you don't remember well, with or without a clock. Each pick fine-tunes the order of your films within the same rating in My Ratings and sharpens your own read; your ratings stay exactly as they are, and nothing else on Chromahall changes. After each pick you also see which way the crowd leans. You'll find it in My Ratings whenever there are ties to break.
Recommendations
For You recommends from your Crew: it looks for the films the people who rate like you agree on and you haven't rated yet. It isn't a prediction of your score, and it isn't a sort by genre. You get one film at a time, with a short line on why it surfaced for you, in three cuts by how exposed a film is: For You, well-known films your Crew rates highly; Vault, films almost nobody has watched that your Crew still backs; and Top picks, the wider consensus, for when your Crew has nothing to say.
Data & credits
Ratings. The CHROMAHALL and RAW numbers are computed from the MovieLens 32M dataset, published by GroupLens Research at the University of Minnesota: ratings given on MovieLens between 1995 and 2023, which is why the catalogue currently stops at 2023. Chromahall shows aggregate statistics only; no individual MovieLens rating is shown or redistributed. Until enough people rate on Chromahall, some numbers reflect the voter base of that dataset.
Chromahall is not endorsed by the University of Minnesota or the GroupLens Research Group. Dataset reference: F. Maxwell Harper and Joseph A. Konstan. 2015. The MovieLens Datasets: History and Context. ACM Transactions on Interactive Intelligent Systems (TiiS) 5, 4: 19:1–19:19. https://doi.org/10.1145/2827872 Thanks to GroupLens for publishing this dataset openly.
Film details. Film details and posters come from the OMDb API (omdbapi.com), used under CC BY-NC 4.0. Posters load straight from the image servers OMDb links to. OMDb is not affiliated with Chromahall and does not endorse it. Posters remain the property of their respective owners.
Imports. Letterboxd and IMDb are named because you can import your history from them; neither is affiliated with Chromahall.
Personal & non-commercial
Chromahall is a personal project. One person, no company, no investors, no ads, nothing for sale.
It doesn't track you: no ads, no tracking cookies, nothing that follows you outside Chromahall, no profiling for anyone else. It keeps only what the product needs to work — your account, your ratings, your session, any message you send and short-lived server logs for security — and never sells or shares it for anyone else's use. When Chromahall fetches film details, it asks about the film, never about you. Fonts and icons are served by Chromahall itself; film posters are the exception, loaded by your browser straight from the image servers OMDb links to. The full detail is in Legal.
Contact
Found something off, or have an idea? Write through the contact page or to contact@chromahall.com — I read everything.
These numbers come from statistics. They show what the ratings add up to, not whether a film is for you — that part is your own call. Every taste is different, and that's what makes it worth reading. Personally, my favourite finds are the films nobody agrees on: that's where my own gems tend to hide. If you find yours, or something that doesn't add up, I'd like to hear about it.
— James