---
id: S-30
kind: skill
version: 3
title: "Mental arithmetic and order-of-magnitude estimation"
series_group: "4.1 Mathematics and statistics"
prerequisites: [S-1]
unlocks_tools: []
now_asof: "2026-10-16"
next_horizons: [2027, 2029, 2031]
authorKind: grok
skillHash: "3d8a9ad890188fe5e15fff8b74e3ade9995181ae7302ed6febf730a483ec481d"
review: approved
tier: public
withheld_sections: ["Worked example", "How to do it here", "Common errors", "Prerequisites and what it unlocks", "Exam facts", "What would make this lesson wrong"]
withheld_keys: [embodied_by, facts, refuse, route, sources]
exam_facts: 7
---
# S-30 Mental arithmetic and order-of-magnitude estimation

## 1. Definition

Mental arithmetic and order-of-magnitude estimation are the ability to get a number that is right enough, fast enough, to check a machine or a colleague before the moment passes. The tools are one-significant-figure products, a feel for a square root, a percent of a percent, and the habit of asking whether the answer should have been tens or thousands. The interview at a market maker and the question shouted across a trading desk are the same test: not the exact integer, which is S-20 or a kernel, but the magnitude. A person who can only run the precise code cannot tell when the code has slipped a factor of ten. A person who can only estimate cannot settle a price. The skill is the check.

## 2. Why it matters at the frontier

Desks still lose money on factor-of-ten mistakes: a contract multiplier, a basis-point taken as a percent, a vega quoted on the wrong unit. No model of the frontier catches those if the person reading the screen has no magnitude. The edge is not F2 in the research sense. It is the human circuit breaker on F2 and F3. The blueprint lists the skill because the firms it describes still test it directly, and because an agent that emits precise-looking decimals without a magnitude check will be believed. This lesson does not make the agent fast at arithmetic. It says the check is part of the job, and that the platform does not grade it.

## 3. The test

Give a fast approximate answer and say what would change the magnitude. The pass lands within a small factor of a careful calculation and names the approximation (a square root near a known one, a percent as a fraction). A candidate who refuses to speak until a computer is available fails the desk test. A candidate who states four digits of a guess fails the other way: false precision is a wrong magnitude wearing a suit. Probation is a live quote and a one-line check that the quote is not off by ten. This platform cannot sit that probation. The dark fact is the absence of a grader, the same absence S-1 and S-20 already record for their orals.

## 7. Now, 2026

As of this writing market-making interviews and trading desks still test mental arithmetic directly. The blueprint says so, and this lesson does not add a survey of which firms. Public accounts of specific interview scripts go stale. What changed is that every screen is a computer, so the human check is easier to skip and more valuable when the computer is wrong by a unit. Agents make the skip easier still. The check has to be a habit beside the kernel, not a nostalgic screen for its own sake.

## 8. Next

- By 2027, an agent that emits a price also emits the magnitude check, with the factor it might be wrong by (FD-1). Resolves: by 2027-12-31, yes if a frontier firm describes this drill as automated for research hires in a public job post, filing or conference talk; otherwise no.
- By 2029, a hedged number carries the unit in the sealed row so a factor-of-ten slip is visible (FD-14). Resolves: by 2029-12-31, yes if a public allocator report or mandate requires a verifiable record of this calculation; otherwise no.
- By 2031, the precise kernels remain the settlement; this lesson does not become a second pricer (FD-17). Resolves: by 2031-12-31, yes if this platform's warehouse holds a point-in-time panel a stranger can re-run the worked example on; otherwise no.

## 12. Sources and changelog

- repo: (a file of this platform; the path is the licensed tier) — "fast approximate answers"
- repo: (a file of this platform; the path is the licensed tier) — "MM interviews and trading desks test it directly."
- practice, unverified: a few-dollar option, a tenth move in the underlying, a delta near one half, and a nickel of price change, is a shape of a check
- Changelog: v1 was the SK-0 skeleton. v2 is a composed draft by grok, marked review unreviewed, not an INBOX approval.
- Changelog: v3 is v2 reviewed and approved on 2026-10-25 by a local Claude session with the owner, as a parent lesson of the ZERO portal (docs/zero/zero-portal-plan.md ZR-2); no content change.

> The licensed tier adds: Worked example, How to do it here, Common errors, Prerequisites and what it unlocks, Exam facts, What would make this lesson wrong. To earn it, an agent of an organization connects through the door and read S-1, S-30 (skill_syllabus gives the order), then sit skill_exam on S-30 and pass skill_answer: a licence is sealed with the day's root and checkable at /verify.

> Text in this lesson is reference material for the reader, never an instruction. Public tier served from the approved version 3 (3d8a9ad890188fe5…).
