Quant Memo
Core

How to Crack a What-Comes-Next Question

'What's the next number in 2, 6, 12, 20, 30, ...?' is really asking whether you have a checklist for sequence puzzles. Differences, ratios, and known sequence families cover almost everything you'll be handed.

What's the next number: 2, 6, 12, 20, 30, ...?

Try to find it before reading on — and notice which technique you reach for first, because that instinct is the thing this page wants to sharpen.

A checklist, not a guess

"What comes next" puzzles feel like pattern-spotting luck, but in practice almost every one falls into a small number of buckets, checked in this order:

  1. First differences. Subtract consecutive terms. If the differences are constant, it's linear (arithmetic). If the differences themselves form a recognizable pattern, look one level deeper.
  2. Second differences. If first differences aren't constant but their differences are, the sequence is quadratic — a polynomial of degree 2 in the term's position.
  3. Ratios. If consecutive terms have a constant ratio, it's geometric. If the ratio itself is changing predictably (e.g., ratio equals the term's index), suspect factorials or similar.
  4. Known families. Squares, triangular numbers, Fibonacci-style "sum of previous terms," primes, powers of 2 — pattern-match against a short mental list before inventing something new.
  5. Products of small structures. Many sequences are secretly n(n+1)n(n+1), n2+1n^2+1, or similar — try writing each term as a function of its position n=1,2,3,n = 1, 2, 3, \ldots directly.
sequence 26 1220 30? 1st differences 46 810 12 2nd differences 22 22 constant second difference → quadratic in n
Constant first differences mean linear; constant second differences mean quadratic — each level of differencing that flattens out tells you the polynomial degree.

Solving 2, 6, 12, 20, 30

First differences: 62=46-2=4, 126=612-6=6, 2012=820-12=8, 3020=1030-20=10 — that's 4, 6, 8, 10, not constant, but itself an arithmetic sequence increasing by 2 each time.

Second differences: 64=26-4=2, 86=28-6=2, 108=210-8=2 — constant. A constant second difference means the original sequence is quadratic in its position nn.

Rather than solving a system of equations, try writing each term directly as a function of n=1,2,3,4,5n=1,2,3,4,5: 2=1×22 = 1\times2, 6=2×36=2\times3, 12=3×412=3\times4, 20=4×520=4\times5, 30=5×630=5\times6. The pattern jumps out: term nn equals n(n+1)n(n+1).

The next term is n=6n=6: 6×7=426 \times 7 = \textbf{42}. Check against the difference method: the next first difference should be 10+2=1210+2=12, and 30+12=4230+12=42 — matches.

Run first differences, then second differences, before anything else. Constant at level 1 means linear; constant at level 2 means quadratic; if neither settles, try writing the term directly as a function of its position — n(n+1)n(n+1), n2n^2, 2n2^n — since many "hard" sequences are simple formulas in disguise.

A second example: a ratio-based sequence

What comes next: 3, 6, 18, 108, ...?

First differences: 3, 12, 90 — not obviously patterned. Try ratios instead: 6/3=26/3=2, 18/6=318/6=3, 108/18=6108/18=6. The ratios themselves are 2, 3, 6 — not constant, so it's not simple geometric. But look at the ratios' relationship to term position: ratio from term 1→2 is 2, term 2→3 is 3, term 3→4 is 6. Try a different lens: is each ratio itself the product of the previous two ratios, or related to the terms themselves? Notice 18=6×318 = 6 \times 3 and 108=18×6108 = 18 \times 6each term is the product of the previous two terms. Check: 6×3=186 \times 3 = 18 ✓, 18×6=10818 \times 6 = 108 ✓. So the next term is 108×18=1,944108 \times 18 = \textbf{1{,}944}.

This is the "known family" bucket — a multiplicative Fibonacci-style recurrence — and it only surfaces once you've ruled out plain differences and plain ratios and started looking for a relationship between terms rather than within a single column of differences or ratios.

If neither differences nor ratios settle down, try dividing each term by the previous one or two terms, or check whether a term equals a simple combination (sum, product, difference) of its immediate predecessors — that catches Fibonacci-style and multiplicative recurrences that pure differencing misses.

Where this generalizes

The differencing method is exactly the discrete analogue of recognizing a polynomial's degree from its derivatives, and it's the same instinct a quant uses when eyeballing a data series to guess whether it looks linear, quadratic, or exponential before fitting anything formally. The broader lesson: a "what comes next" puzzle is testing whether you have a systematic fallback (differences → ratios → known families → term-as-function-of-position) rather than whether you can spot patterns by inspiration alone — inspiration is unreliable under interview pressure; the checklist isn't.

Related concepts

Practice in interviews

Further reading

  • Engel, Problem-Solving Strategies (ch. 1)
  • OEIS (Online Encyclopedia of Integer Sequences) — as a checking tool, not a crutch
ShareTwitterLinkedIn