The math optional, made finite. Daily Practice

The maths optional is no longer a mystery

We mapped thirteen years of maths-optional papers, then held one back and classified it blind. What came back surprised even us — and it's why we're opening a small pilot.

11 min read

A dark field scattered with small charted nodes joined by faint lines — the map built from thirteen years of maths-optional papers. A single fresh sheet of paper drops onto it, its points lighting up on nodes that already exist; one lone node sits off on its own, ringed and uncharted. The unseen paper, almost entirely on ground already mapped.

We did something with the maths optional that we’d been afraid to do, because we weren’t sure we’d like the answer.

For two years the story everyone tells about this paper has been the same: it’s a black box. Unpredictable. A gamble. The optional that can hand you a rank and can just as easily sink you, and there’s no telling which until results day. You take it on faith and a prayer.

We stopped taking it on faith. We ran an experiment instead. And the result is the reason this page exists.

The experiment we almost didn’t run

Start with what we’d already done. We solved every maths-optional paper from 2013 to 2025 — all twenty-six papers, all 804 sub-parts, Paper 1 and Paper 2, worked in full and posted free at /papers/, no wall. While we were in there, line by line, we noticed something we couldn’t unsee: the paper repeats. Not the exact questions — the underlying types. The same families of problem, dressed in new clothes, year after year.

So we built a map of them. We took the thirteen years and sorted every sub-part into its underlying pattern — its archetype. 804 sub-parts collapsed into 434 recurring archetypes. A finite, knowable list of what this exam actually asks.

Then came the part we were nervous about. It’s easy to draw a map of the past and declare it complete — you already know the answers, so of course everything fits. The honest test is the one you can fail. So we held a paper back.

Here’s exactly how we ran it. We built the entire map from 2013 to 2024 only — and we locked it. Frozen. Then we took the 2025 paper, which the map had never seen, and classified it blind — solved it, sorted every sub-part into its pattern, all before we were allowed to check it against what the map already covered. No quietly fitting the new paper to make ourselves look good. A real out-of-sample test. The kind you can fail in public.

We fully expected gaps. We braced for the new paper to wander off our map in three or four places and humble us.

Here is what actually came back.

A student carrying only the 2013–2024 map could have attempted 100% of the attemptable marks on both unseen 2025 papers. Every compulsory question — already on the map. All but one sub-part of the entire unseen paper mapped to patterns we’d already charted.

Read that twice, because we had to.

Across both 2025 papers, 61 of 63 sub-parts landed on ground we’d already charted. The lone holdout was a single 8-mark sub-part on a genuinely new topic — one truly new idea in fourteen years of papers — and it sat inside an optional question you could simply skip and never feel the loss. The paper’s own structure (choose three of six, at least one per section) routes you cleanly around it.

We want to be precise, because precision is the whole point of an experiment like this. We are not saying we can predict next year’s paper — nobody can, and anyone who tells you otherwise is selling something. We map recurrence; we don’t forecast specific questions. What we are saying is narrower and far more useful: the maths optional is not a black box. It is finite, it repeats, and the repetition is now charted — charted, and then validated against a paper it had never met.

That is what we mean by “no longer a mystery.” Not a crystal ball. A map — tested on a road it had never driven, and found to already know almost every turn.

The person who built the map (and why he’d have killed for it)

The map didn’t come from a coaching factory or a research grant. It came from one software engineer who took the maths optional the hard way, and lost a fair bit of blood doing it.

He had a full-time job. The optional got the scraps: evenings after work, weekends that were supposed to be rest. And like most people preparing alone, he made every mistake in the book — not because he was careless, but because nobody hands you the right method, you have to bleed your way into it.

He ground topics one at a time, in long blocks — linear algebra for two weeks until he felt fluent, then moving on and watching it quietly evaporate while he drowned in the next thing. He re-read solved problems and mistook the warm glow of recognising a solution for the hard skill of producing one cold. He had no system telling him what to practise today — so every night started with the small tax of deciding, and too many nights he reached for whatever felt easiest. He confused feeling busy with getting better, which is the most expensive mistake in this entire game.

He came through it. But he came through it carrying a very specific ache: the certainty that he’d done it the slow, lonely, wasteful way, and that the thing that would have saved him — a map of what actually mattered, and a system that simply told him what to work on each day — had never existed. So he built it. Afterwards. For the people walking in behind him.

"I guide others to a treasure I cannot possess."

He’d laugh at that line, and then admit it lands. The map arrived too late to win him an easier run. That’s exactly why he was so stubborn about building it for everyone after — so that no one preparing alone, with a day job and a stubborn dream, has to learn the method the way he did: the slow way, by losing.

This whole platform is that one engineer’s argument, made in code: that the maths optional is winnable for ordinary people with ordinary hours, if someone just hands them the map he had to draw himself.

What the daily loop actually buys you

Now the optimistic part, and we’re going to be openly optimistic here — but we’ll only promise what the design can actually deliver, not what would sound good on a poster.

Here’s what the daily loop actually does. It hands you a practice sheet built for you — a small, mixed set of problems each day, chosen by our proprietary algorithm to sit exactly where your strengths and weaknesses are that morning. Not a generic worksheet everyone gets. Yours. The algorithm draws from a large and continuously growing practice bank — seeded from the thirteen-year map and patterned on it, so every problem you solve is the kind of problem the exam actually asks, at the depth you need it. The map is the foundation the bank is built on; the personalised sheet is what you practise.

Sit with what that means. The exam has a finite, knowable list of question types — and the engine works that list with you specifically in mind, day after day. The thing that made this paper feel infinite was never its size. It was the fog, and the loneliness of facing it without a plan for today. Lift both and what’s left is a bounded, repeatable body of work, served to you in the order that serves you best. A student who sticks with this program walks into the hall having already met, many times over, the kinds of questions that make up the paper — at the depth they personally needed — which is precisely what the blind 2025 test showed the map contains.

We’ll be honest about the floor, because you’ve earned honesty more than you’ve earned hype: nobody has taken this exact loop into an exam hall yet, so we have no scores to wave at you, and we won’t invent any. What we can promise is the shape of the work — stick with the program and the engine keeps you practising exactly where you’re weakest, while quietly keeping warm the ground you’ve already won, so nothing you’ve earned slips away while you fix what hasn’t. The execution on the day is still yours. But you’ll arrive having practised every pattern the paper draws from, at the depth you needed — not the territory in general, your territory.

There’s a quieter relief folded into the design, too. The loop hands you one bounded set a day — a thing you can finish and then close. No twelve-hour panic marathons, no guilt spiral when the marathon collapses. The science behind it is plain and well-established: spreading practice across days beats cramming it into one, and mixing problem types beats drilling one type in a block. Cognitive scientists call the first distributed practice; Doug Rohrer’s controlled studies on maths homework specifically (Rohrer, Dedrick & Stershic, 2015) show the second — students did markedly better on a later test from nothing more than having their problems shuffled into mixed order. Pulling an answer out of your own head beats reading one back, every time (Roediger & Karpicke, 2006). The loop is just those settled findings, made automatic, so you don’t have to be your own scheduler and your own discipline at once, alone, every night.

We’re opening it to a small first group

Here’s where it stands. The loop is built and running. We’re not throwing the doors open to thousands on day one — we’re starting with a small, hand-picked first group of aspirants, so we can sit close to them, watch how the loop actually serves a real day of real preparation, and get it right before it grows.

So yes — the first round is limited, genuinely. Not as a trick, not a countdown clock ticking in the corner to rush you. Just the plain truth of how we’re starting: a handful of people, by invitation, soon. If you want to be among them, the door is the sign-in page — leave your details there, or in the form above, and we’ll reach out as places open.

And the most important line on this whole page: none of this is what you start with. What you start with is free, and it stays free, forever. We solved all thirteen years — every Paper 1 and Paper 2 from 2013 to 2025, all 804 sub-parts, worked in full and ungated — alongside a full guide to the syllabus, the same way. No email, no wall. That corpus is the gift; the daily loop is just the next step for someone already getting value from it. Open any paper tonight and read a whole solution. The map is right there, waiting for you.