oluwatobi oyinloye
Back to all work

Do Higher-Rated Movies Make More Money?

Barbie was rated 6.9, while Oppenheimer was rated 8.3. If ratings drove revenue, you'd know which one won the box office. For this project, I examined whether rating or budget led to higher box office earnings, using a random sample of 1,000 modern films.

Individual project. CM561 (Entertainment Media) analytics project, Boston University, April 2026.

Built with

Language
Python
Analysis
pandas and NumPy
Charts
matplotlib
Data
TMDB, via Kaggle

Did it Make More or Less?

Before I show you what I found, try it yourself! Using each film's rating and budget, guess whether the film on the right made more or less at the box office than the one on the left, and keep going until you miss.

Streak0

Over the same films, here's how far each strategy would have got.

That's the whole project in one game. Across my 1,000 films, budget explained 56% of the variation in box office revenue. Rating explained 8%.

This game was created using my own data, a sample of 1,000 films from the TMDB dataset, prepared in Python with the help of Claude. It uses the 200 best-known films in that sample. Box office is worldwide revenue as listed by TMDB.

Background

The assumption

Audience ratings often get used as a signal of commercial potential because the idea is that 'better films' attract bigger crowds, which means more money. But sometimes it's not as straightforward as that.

The question

Industry analysts have found that budget is one of the strongest predictors of box office gross, far outpacing audience scores or critic reviews (Follows, 2023).

The stakes

If ratings and revenue are only weakly linked, studios and streamers may be misallocating resources by over-indexing on quality signals instead of scale.

Research questionTo what extent are movie ratings associated with box office revenue, and how does budget compare?

Methods

I took a random sample of 1,000 movies from a dataset of over 1 million TMDB films, keeping only those released since 2000, with a budget and revenue above zero and more than 100 votes.

Revenue is heavily right-skewed. A handful of blockbusters pull the mean up to $112.3M while the median is $38.8M, so I applied a log10 transformation to stop those outliers from bending the correlation.

Measured how strongly rating, budget and revenue move together, on both raw and transformed revenue.

Split films into low, medium and high rating groups and compared median revenue, since the average gets pulled around by hits.

What the Sample Looks Like

Log transformRescaling numbers so each step is ten times bigger than the last. It stops a few enormous values from dominating a chart or a correlation, which is why the right-hand charts below look so much tidier.

The data came from TMDB via Kaggle (Asaniczka, 2024), which contains audience scores. After filtering, my random sample of 1,000 films had six columns: title, release date, rating, number of votes, revenue and budget.

  • Ratings averaged 6.5 out of 10, ranging from 3.2 to 8.5.
  • Revenue averaged $112.3 million, but the median was just $38.8 million, because a few blockbusters pull the average up. That's why I also used a log transform on revenue, shown in the right-hand charts below.
  • Budgets averaged $39.0 million, with a median of $21.7 million.
Four scatter plots of rating, budget and revenue
Rating and budget against revenue, raw and log-transformed. Budget tracks revenue far more tightly than rating does.

Budget Wins!

Higher rated films do earn more at the median, but the correlation is weak. Squaring each Pearson correlation (r²) shows that rating explains about 8% of the variation in revenue, while budget explains 56%, seven times as much. And spending more doesn't buy better reviews either, since budget and rating barely correlate (r = 0.130).

Correlation (r)How closely two things move together, from -1 to 1. Squaring it, r², gives the share of the variation one explains in the other. That's where 56% and 8% come from.

$20.9MRated6 or below262 films$44.2MRated6.1 to 7.5636 films$62.3MRatedabove 7.5102 films
Budget and revenue0.751Strong
Budget and log revenue0.580Moderate
Rating and log revenue0.283Weak
Rating and revenue0.276Weak
Rating and budget0.130Very weak

If you Greenlight Films for a Living

Stop using ratings as a revenue estimator. Ratings explain only ~8% of box office variance but factors like distribution scale, marketing spend, and franchise brand equity are stronger drivers. This is why revenue projections shouldn't be anchored to test screening scores or early audience reactions.For studios and distributors

Invest in distribution alongside production. The budget-revenue correlation reflects the entire investment ecosystem into: P&A (prints & advertising), wide release windows, and pre-release marketing. A well-made film with a release that's not scaled is likely to be disadvantaged.For producers and financiers

Evaluate quality and commercial success separately. High revenue + moderate rating is not a failure and low revenue + high rating is not a flop - it may be a niche success. Conflating (confusing) these metrics leads to misreading both audience needs and market dynamics.For streaming platforms

Further Research

To take this research further, I'd:

  • Add release timing, genre and franchise to see what else moves the box office.
  • Track how these relationships have shifted over the years.
  • Look at streaming, where revenue works very differently.

There's a bigger point here too. If investment drives revenue, smaller-budget films, which are more likely to feature diverse stories and people, start at a disadvantage. That shapes what kind of movies get told, and who gets to tell them.

Loved ≠ Lucrative

A film can be loved and still underperform at the box office, and vice versa. They're not measuring the same thing.

A film's commercial success turns out to be less about how good it is, and more about how much was put into making, marketing and distributing it.

Next project
From Emergency to Aftermath: Tracking Vaccine Hesitancy and Anti-Vax Discourse on X/Twitter