The Long Bet · Article 4 of 5

The Unfashionable Four

Why the A-level combination top universities have always wanted hasn't changed – and why most students don't pick it

Schools present A-level choice as a menu. Computer Science for the future engineer. Biology for the future doctor. Business for the future entrepreneur. Psychology, Economics, Design Technology – all legitimate, all offered. The message is: pick what's relevant to you.

The admissions data from 31 STEM programmes across 100 ARWU top universities tells a different story. There is a hard core golden standard – four subjects that top research universities have required for decades. Not approximately. Not with occasional exceptions. For decades, across 19 countries and every major STEM field. The menu around them grew. The four didn't move.

The Hard Core Golden Standard
Mathematics  ·  Further Mathematics
Physics  ·  Chemistry
Stable across 31 programmes, 100 universities, 19 countries

Here is what the data shows – and why the subjects that feel most relevant are not the ones that are load-bearing.

A note on sourcing before the numbers: 9 of the 31 programme requirements were verified directly against current admissions pages – Cambridge, Oxford, UCL, Edinburgh. The remaining 22 are extrapolated from those verified patterns and published entry requirements. Checked and estimated rows point the same way; the exact counts below inherit that mix.


Mathematics – required by 30 of 31 programmes

The baseline. Not a finding – a precondition. A student without Mathematics is not applying to STEM at a top research university.

What the data shows beyond this: the A* needs to land here specifically. A*A*A means three A-levels at grades A*, A, A. The natural assumption is to put the A* in whichever subject is strongest. Cambridge CS admissions data says otherwise – the A* is required in Mathematics, not just anywhere across the three subjects.

A student can achieve A*, A*, A and still not meet Cambridge's requirement.
If Mathematics is not one of the A*s, the Cambridge CS entry condition is not satisfied – regardless of total grades. Not every university states the rule this explicitly – but at the most selective programmes the pattern is consistent: the top grade is expected in Mathematics. The golden standard runs all the way down to where the top mark falls.

Further Mathematics – recommended by 25, required by 2

Formally required by 2 of 31 programmes. Recommended by 25 of 31.

Cambridge says "if your school offers it." Oxford says "preferred." UCL says "strongly recommended." None of them say required – they don't want to exclude students from schools that don't offer FM. But a student from a school that does offer it and chooses not to take it is at a structural disadvantage across 25 of 31 programme categories.

At A*A*A entry level, "strongly recommended" and "required" are the same thing.
The subject top universities most want is not the one they say they require. This is the first layer of the standard that official admissions guidance obscures.

Physics – the key to the engineering cluster

Formally required by 12 of 31 programmes. All twelve are in the engineering and physical sciences cluster – Electrical, Mechanical, Aerospace, Biomedical, Civil, Nuclear, Systems Engineering, plus Physics and Astrophysics.

Universities write "recommended" in the edge cases. The core is not ambiguous. Electrical Engineering at Cambridge: Mathematics and Physics required. Biomedical Engineering at Cambridge: Mathematics and Physics required. Aerospace, Mechanical: the same.

A student without Physics is not a weak candidate for Electrical Engineering at Cambridge. They are an ineligible one.
Physics is the second layer of the standard. It unlocks the largest and most research-active cluster in the dataset. Top universities assume serious applicants already have it.

Chemistry – the subject nobody told you about

Formally required by 4 of 31 programmes – the same number as Biology. Most students and parents assume Biology is the relevant choice for anything biology-adjacent. The data does not support this.

The difference is in where each subject's requirements apply.

Chemistry's four programmes – Chemistry, Biochemistry, Chemical Engineering, Neuroscience – apply globally. At US, French, German, and UK universities alike.

Biology's four – Biochemistry, Bioinformatics, Computational Biology, Neuroscience – are UK-specific gates. At the 37 US universities in the ARWU top 100, none of these programmes require Biology A-level. Biology, chemistry, and physics are all taught from scratch within the degree.

And for Computational Biology specifically – the field that sounds most "biology" – top US universities consider Chemistry the stronger preparation. General Chemistry and Biochemistry appear in year 1 of every US Bioinformatics and Computational Biology degree. A student with Chemistry arrives prepared for those courses. Biology is taught from scratch regardless.

Verified June 2026:

The field name says Biology. The degree structure says Chemistry.
Biology A-level is not a global requirement. It is a feature of one country's admissions system. Chemistry is the third layer of the golden standard – and the one nobody states.

The standard confirmed

Subject weight across 31 STEM programmes
ARWU top 100 universities · 19 countries · June 2026
Formally required
Recommended / expected
No formal requirement
Further Mathematics: "recommended" at A*A*A entry level is functionally required. Economics, Psychology, Business: 0 requirements across all 31 programmes.
0
Economics
programmes requiring it
0
Psychology
programmes requiring it
0
Business
programmes requiring it

The subjects that fill the school menu carry nothing in the admissions data. Four subjects. Each one load-bearing. Each one present at every level – required, recommended, or structurally expected. None of them are surprising. All of them have been there all along.

The proof is in what the alternatives don't do.

1
Computing A-level
formally required
programmes
vs
30
Mathematics A-level
formally required
programmes

The subject that signals CS interest most loudly to a student is almost never required by the university itself. Cambridge CS doesn't require Computing. It requires Mathematics – and Further Mathematics if the school offers it.

Programme access by choice of 4th subject
Base: Maths + Further Maths + Physics · out of 31 programmes
Chemistry
28 / 31
90%
Biology
27 / 31
87%
Computing
25 / 31
81%
Computing is the weakest of the three options – and the one that feels most obviously right.

The golden standard was never hidden. It was always visible in the admissions data. The expanded school menu created the appearance of choice. The data shows the choice was always narrower than it looked.


For Andrey

The research was built for a specific decision. Andrey is preparing to choose his A-levels. The subjects he picks now determine which universities remain open – and which close before he has written a single application.

The answer from 31 programmes, 100 universities, and every layer of this analysis is unambiguous.

The answer
Mathematics  ·  Further Mathematics
Physics  ·  Chemistry

That combination opens 28 of 31 programmes in the UK system. The three it misses are UK-specific biology gates that don't apply at US, French, or Chinese universities. At MIT, Stanford, UCSD, and CMU, that combination is not a limitation. It is exactly the profile those programmes are looking for.

The A* goes on Mathematics.

For other students reading this: the data above shows the standard. Whether it applies depends on which universities you are targeting and what you want to study. That decision is yours.
Also in this series
The Problem Statement The four-constraint optimisation problem and why every obvious answer fell apart.
How I Built It Why ARWU over QS and THE, why 100 universities, and what 1,200+ normalised pairs made possible.
What the World Actually Offers CS at 95 universities. AI at 12. France above Imperial. What the coverage matrix actually showed.
The Floor Keeps Rising Seven scarcity cycles since the 1880s. The compression curve that makes a four-year degree longer than a full cycle.