All the movies and series I watched in 2025

And how to calculate the Mutenge score

100DaysOfPolars

Author

Joram Mutenge

Published

2025-12-31

If someone asks about your hobbies, do you tell them something that makes you seem smart or cool, or do you tell the truth? If a hobby is what you spend most of your free time doing, then many people would be embarrassed to say their true hobbies out loud. But numbers do not lie, and the evidence they provide is always beautiful. I spend most of my free time watching movies and TV series. That is my hobby. The good news is that I have been recording every movie or series I have watched since 2022.

About the dataset

Below is a snapshot of my movies and series dataset. I’ve filtered for the year 2025 to only show the movies I watched this year.

shape: (3, 7)

Type	Title	Season	Release_Year	Person	Quote	Watch_Year
str	str	i64	i64	str	str	i16
"Series"	"Hotel Babylon"	4	2006	"Rachael Stirling \| Natalie Men…	"Imagine your worst day a day w…	2025
"Movie"	"You Hurt My Feelings"	null	2023	"Sunita Mani \| Erica Matlin"	"No man wants to have a V-neck.…	2025
"Series"	"Bad Sisters"	2	2022	"Eve Hewson \| Sharon Horgan \| S…	"We do things for those we love…	2025

Most of the columns are self-explanatory, but two require some explaing:

Person: contains a list of people I enjoyed watching. Names are separated by |, so “Eve Hewson | Sharon Horgan” contains two people Eve Hewson and Sharon Horgan.
Quote: contains lines that were good enough to be quotable. Lines are saprated by |.

Some analysis

Since the dataset is proprietary, I’ll show you a few analyses that I think you may be eager to know the answers to.

How much TV did I watch in 2025?

Let me answer the question you are dying to have answered. As I said earlier, I watch a lot of TV. It is my hobby. Let us find out how much with the help of some Polars code:

(df
 .select(Titles=pl.col('Title').count())
 )

shape: (1, 1)

Titles
u32
32

Watching 32 titles in a year may not seem like a lot, but remember that some of these titles are series with 40 episodes spread across multiple seasons. That’s a lot of watch time!

Note

I have watched all the series and movies on this list from beginning to end. If it is a series, then I have watched all the seasons and episodes released under that title.

What’s the breakdown between movies and series

Let us get more granular and see how many of these 32 titles were movies and how many were series:

(df
 .group_by('Type').len()
 )

shape: (2, 2)

Type	len
str	u32
"Series"	15
"Movie"	17

The Mutenge score

You may be wondering why I go through the trouble of building this dataset. The IMDB ratings assigned to movies and series do not mean much to me. They do not truly capture my taste. To capture my true taste in movies and series, I created a metric known as the Mutenge score (yes, it is named after yours truly) to quantify how much I like a movie or series. At first glance, the formula may seem complicated, but after explaining it, I promise it will make a lot of sense.

Formula

\[ \text{Mutenge Score} = 4 \cdot \tanh\left( \frac{ \dfrac{ \begin{cases} 2P + Q & \text{Movie } (S = 0) \\ 4P + Q & \text{Series } (S > 0) \end{cases} }{1 + 0.5(8S - 1)} - \begin{cases} 3.5 & \text{Movie} \\ 2.0 & \text{Series} \end{cases} }{1.5} \right) \]

Explanation

The Mutenge score ranges from -4 to +4. A score of -4 signifies OK, a score of 0 signifies Good, and a score of +4 signifies Excellent. Each movie or series is assigned a single score, which can be an integer or a decimal value.

Unlike most scales that range from most disliked to most liked, the Mutenge score is divided into two territories. The OK-to-Good territory encompasses the scale between -4 and 0 (the midpoint), while the Good-to-Excellent territory encompasses the scale between 0 and +4. I liked all the movies and series that ended up in the dataset; if I do not, I simply stop watching and move on to something else. When it comes to movies and series, we are spoiled for choice. There is no need to punish yourself by watching something you are not enjoying. The Mutenge score allows me to quantify how much I liked and enjoyed a movie or series.

Parameters and their meaning

P — Person count

Number of unique persons in the title
Weighted more heavily because characters matter more to me
Person count weight is not the same. Movies = 2, Series = 4

Q — Quote count

Number of unique quotes
Still important, but secondary to people

More people matters more than more quotes.

S — Season count

S = 0 → Movie
S ≥ 1 → Series

Episodes per season = 8

A fixed storytelling-length assumption used for normalization.

\(α\) (alpha) — exposure softening factor

\[ \alpha = 0.5 \]

Controls how strongly long series are penalized.

α = 1.0 → full linear penalty (too harsh)
α = 0.5 → diminishing penalty (chosen)
α → 0 → almost no length penalty

Exposure

\[ \text{Exposure} = 1 + \alpha \cdot (8S - 1) \]

Movies → exposure = 1
Series → grows sublinearly with seasons
Prevents series from collapsing toward −4

D₀(S) — Neutral density

\[ D_0(S) = \begin{cases} 3.5 & \text{Movie} \\ 2.0 & \text{Series} \end{cases} \]

Represents “Good” (neutral enjoyment) per unit exposure.

k — Sensitivity

\[ k = 1.5 \]

Controls how quickly scores approach ±4.

Smaller → more extreme scores
Larger → flatter distribution

tanh compression

Smoothly bounds the score to [−4, +4]
Preserves relative differences without hard clipping

Table 1 illustrates a summarized version of the parameters of the formula.

Table 1: Parameters for Mutenge score formula.

Parameter	Value	Meaning
\(P\)	number of persons	unique persons in the title
\(Q\)	number of quotes	unique quotes in the title
\(S\)	number of seasons	0 = movie, ≥1 = series
\(W(S)\)	person weight	2 for movies, 4 for series
Episodes per season	8	exposure constant
\(α\)	0.5	exposure softening factor
\(D\)₀\((S)\)	neutral density	3.5 movies, 2.0 series
k	1.5	sensitivity for `tanh` compression
Score range	[-4, +4]	bounded via `tanh`

Apply the Mutenge score

Now that I’ve explained how to calculate the Mutenge score, let’s use Polars to assign a score to each movie/series.

(df
 .with_columns(pl.col('Season').fill_null(0))
 .with_columns(pl.col('Person', 'Quote').str.split(' | ').list.n_unique())

 # ---- constants ----
 .with_columns(_episodes_per_season=pl.lit(8),
               _alpha=pl.lit(0.5),
               _sensitivity=pl.lit(1.5))

 # ---- type-aware neutral density ----
 .with_columns(pl.when(pl.col('Season') == 0)
               .then(3.5)
               .otherwise(2.0)
               .alias('_neutral_density'))

 # ---- softened exposure ----
 .with_columns((1 + pl.col('_alpha') * (pl.col('_episodes_per_season') * pl.col('Season') - 1))
               .alias('_exposure'))

 # ---- type-aware person weighting ----
 .with_columns(pl.when(pl.col('Season') == 0)
               .then(2 * pl.col('Person') + pl.col('Quote'))   # movies
               .otherwise(4 * pl.col('Person') + pl.col('Quote'))  # series
               .alias('_weighted_content'))

 # ---- density ----
 .with_columns(_density=pl.col('_weighted_content') / pl.col('_exposure'))

 # ---- calculate score ----
 .with_columns((4 * ((pl.col('_density') - pl.col('_neutral_density')) / pl.col('_sensitivity')).tanh())
               .alias('Mutenge_Score'))

 # ---- cleanup ----
 .select(pl.exclude("^_.*$")) # remove cols that start with underscore.
 .sample(5, seed=31)
)

shape: (5, 8)

Type	Title	Season	Release_Year	Person	Quote	Watch_Year	Mutenge_Score
str	str	i64	i64	u32	u32	i16	f64
"Series"	"The Split"	3	2018	2	2	2025	-2.656147
"Movie"	"Roommate Wanted"	0	2015	1	6	2025	4.0
"Movie"	"Frances Ferguson"	0	2019	1	2	2025	3.980219
"Series"	"Halt and Catch Fire"	4	2014	5	28	2025	2.165392
"Series"	"StartUp"	3	2016	13	10	2025	3.848377

If you have any questions about the Mutenge score, email me and I will be happy to offer a more detailed explanation.

What movies or series did you find gripping in 2025? Share them with me. I will watch them, assign the Mutenge score to them, and include them in a follow-up article.