Skip to content
WORK SMARTER, NOT HARDER

How our
algorithm works

While most study apps don't even keep track of your progress, FourtyFive takes it a massive step further.

Recency
Last 6 attempts
Difficulty
7-mark Paper 2
Speed
0.84× exam time
Consistency
σ = 4.2
Memory age
11 days
Mastery score
82 / 100
Stoichiometry — solid and stable. Next review in 9 days.
Proven capability 91
Memory retention 74
mastery = ƒ(performance, memory, confidence)
Topic · Derivatives Math AA HL
Capability86
Fluency72
Retention58
Confidence80
Awareness66
Momentum78
Mastery — last 7 sessions
Mastery score
76 / 100 +5 this week

Strong recent performance, but retention is fading on chain-rule problems answered >14 days ago.

Capability · 86 Retention · 58 Confidence · 80
OUR PHILOSOPHY

Absolute
Accuracy

Most revision apps track your performance by the number of questions you answer. FourtyFive builds a living picture.

See our subjects
Three questions, three layers

Not one number.
Three honest ones.

FourtyFive's mastery algorithm does not reduce your ability to a single number. It asks three distinct questions about every topic you study — and combines the answers into a score that reflects what is actually happening in your revision.

1 Question 1

How well have you performed?

Recent attempts speak loudest. Hard questions count more.

The algorithm tracks every answer you give, but it does not treat them all equally. Recent attempts matter more than old ones. Hard questions count more than easy ones. And if you answered quickly and accurately, that signals real fluency — not just a lucky guess. As you answer more questions on a topic, older attempts gradually fade in importance.

Attempt log · weighted
Today 7/8 Hard
Yesterday 6/8 Hard
4 days ago 4/8 Med
2 wk ago 7/7 Easy
6 wk ago 3/6 Med
Weights fade exponentially — old attempts whisper, recent ones speak.
2 Question 2

How much do you actually remember?

Memory is modelled — and your forgetting rate is personal.

Performing well once is not the same as knowing something. The algorithm models how memory works — knowledge fades over time if you do not revisit it. But it does not assume everyone forgets at the same rate. Answer a topic correctly many times with high accuracy, and the memory window grows. Practise once, and the algorithm treats that knowledge as fragile.

Memory retention model
1 attempt 30 high-accuracy attempts
3 Question 3

How confident should we be?

The algorithm widens its uncertainty when the data is thin.

Answer 30 questions on organic chemistry and the algorithm has a lot of evidence — it's confident in your score. Answer one question on wave optics and it doesn't pretend to know how good you are. Instead, it blends the limited data with your own self-assessment and waits for more evidence before making strong claims.

Confidence interval · narrows with data
The algorithm doesn't overclaim — it admits when it doesn't know yet.
What happens behind the scenes

What runs
after every
answer.

Every time you submit an answer, the algorithm runs through a five-step sequence. The result is a mastery score from 0 to 100 for every topic you have studied. Not an average. Not a guess. A calibrated, evidence-based assessment of where you actually stand.

$ submit answer Q14
# scoring · memory · confidence …
mastery 0.61 → 0.66
1

Every answer counts — but not equally

Difficulty + pace, both count

Getting a hard question right after thinking it through shows more than acing an easy one in ten seconds. We weigh how hard the question was, how fast you answered and how recently — so your score reflects how you actually performed, not just a tally of ticks and crosses.

2

It knows when you're about to forget

Pulled back before it fades

Knowledge fades — and different topics fade at different speeds for different students. The algorithm builds a sense of your personal forgetting curve on each topic and pulls it back into your queue just before it starts slipping, not weeks too late.

3

It earns your trust slowly

Earned, not assumed

With only a few attempts under its belt, the algorithm doesn't pretend to know you yet — it leans on your own confidence rating and stays cautious. The more you practise, the more it trusts the numbers, and the more your score reflects evidence rather than guesswork.

4

Steady beats a lucky day

Consistency over flukes

A single 90% on a topic doesn't mean you've mastered it. Five 80%s in a row often does. We watch for consistency — because the IB doesn't care about your best day, it cares about your average one. Reliable performance is what moves your score, not flukes.

5

When you slip, it catches it fast

Caught instantly

If your scores suddenly tank on a topic you used to ace, the algorithm doesn't quietly fold that into a long-term average. It treats it as a signal — something's changed — and surfaces the topic immediately for review, before mock week, before exam week, before it costs you marks.

The result?

You study less
(while learning more)

Instead of repeating questions you already know or wasting time on random topics, our algorithm focuses on the questions you're most likely to forget, or that need the most work. It’s like a playlist that adjusts to your performance, helping you lock in what matters most.

GET STARTED

Want to have early access
to FourtyFive?

Be among the first to try FourtyFive, the AI-powered platform built to make IB revision more efficient. Get algorithm recommendations, real IB exam simulations, and powerful study insights.