About Box Office Watch

Box Office Watch is a site built to make weekend box office results easy to follow. It tracks opening weekends, daily grosses, total box office performance, and how new releases compare to past films — in an easy-to-read, visual format.

If you spot a data error, have a suggestion, or just want to talk movies - send a message and we'll get back to you.

Get in Touch
Have feedback, a correction, or just want to talk movies? Send a message below.

Something went wrong. Please try again.
Thanks for reaching out! Your message has been sent.

About the prediction model

How the forecasts on this site are built — and how we know they work.

Every wide release on this site gets two numbers: what it opened to, and what it will finish at. The first is a fact. The second is a forecast — and it's the hard one, because it has to be made before the legs play out. Here's how that forecast gets built.

It all comes down to the multiplier

Divide a film's total domestic gross by its opening weekend and you get its multiplier — its "legs," or how many times over it earned its debut. Avengers: Endgame, about as front-loaded as blockbusters get, multiplied its opening just 2.4x. A frontloaded horror opener might manage 2.0x and be done. A Christmas family title can run past 6x on holiday tailwinds. The average wide release lands near 2.9x. Nail the multiplier and you've nailed the run — so the entire model exists to predict that one number as accurately as possible.

Seven signals, read together

No single stat predicts legs on its own. A great opening-night grade doesn't guarantee a great hold; a beloved film in the wrong month still fights the calendar. So the model weighs seven factors at once, each one nudging the multiplier up or down from that 2.9x baseline:

1
Audience reception
A composite of Rotten Tomatoes audience score, Letterboxd, and IMDB — read together to triangulate how audiences actually felt, instead of trusting any one platform that can run hot or cold.
2
CinemaScore
The opening-night exit grade from real ticket buyers. The earliest clean read on whether word of mouth will carry a film or quietly bury it.
3
Second-weekend hold
How steeply a film drops in weekend two. On its own, the single most powerful signal of legs in the entire dataset — a soft drop is a movie people are telling their friends about.
4
Seasonality
When it opens. A late-December release inherits the Christmas-to-New-Year corridor and its extended legs; a January release fights the calendar the whole way down.
5
Genre
Western animation legs like almost nothing else. Japanese anime front-loads and vanishes. Horror burns hot and fast. The model treats these as the genuinely different animals they are, rather than lumping them together.
6
Sequel & remake status
Built-in demand shows up on opening weekend and fades faster afterward. Sequels and live-action remakes carry a distinct, front-loaded legs signature that originals don't.
7
Release-day timing
A Wednesday launch, a Thursday-night preview, a holiday Monday — the opening "weekend" isn't always three days, and the model adjusts so debuts are compared apples to apples.
Why most forecasts double-count — and this one doesn't

Here's the part that's easy to get wrong: these signals overlap. A film with an A+ CinemaScore usually also has strong audience scores and a soft second-weekend drop — because all three are measuring the same underlying thing, that people love it. Add each one at full strength and you pay a good movie three times for a single fact, and your forecast sails past reality.

The model separates each factor's raw correlation from its true marginal contribution — the piece of the signal that's actually unique to that factor once the overlap with the others is stripped out. A great grade and a great hold don't get double-billed for the same quality. That decorrelation is the core of the model, and it's why the calls hold up on the films where the obvious read is wrong.
It's backtested, not guessed

The model isn't a gut feel with a coat of math on top. It's been backtested against every wide release from 2020 to 2026 — more than 280 films — and rebuilt version over version as the misses revealed where it was double-counting or blind. Anime got split from animation after lumping them cost real accuracy. Horror's harsh CinemaScore curve got its own handling. The holiday corridors got their own math. Each fix came from a film the previous version got wrong.

280+
Films backtested
2020–26
Release history
~14%
Median forecast error
7
Decorrelated factors
We publish the record

Every forecast on this site is made in public and left on the record. When a call is right, the weekend recap says so; when it's wrong, the recap says that too, and explains what the model missed. There's no paywall on the reasoning and no black box behind it. The whole point is that you can watch it work — weekend after weekend, film after film — and judge it for yourself.