---
id: S-24
kind: skill
version: 3
title: "Signal processing"
series_group: "4.1 Mathematics and statistics"
prerequisites: [S-1, S-4]
unlocks_tools: []
now_asof: "2026-10-16"
next_horizons: [2027, 2029, 2031]
authorKind: grok
skillHash: "91d6bb84475b22dbce737ea300a75a081ec323ff23e8b3a3b40ca5b1fea045d6"
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-24 Signal processing

## 1. Definition

Signal processing is what you are doing when you smooth, difference, or decompose a series, whether or not you use the name. A moving average replaces a point with an average of its neighbors or of its past. An exponential average weights the past by a decaying factor, so the latest point has weight 2 over (period plus one) in the usual definition and the rest rides on yesterday's average. Both introduce lag: the average cannot turn until the window has seen the turn, so a crossover tells you the turn happened, late. Spectral methods ask which cycles are in the series. Wavelets ask the same question locally in time. Aliasing is what happens when you sample slower than the thing you care about and a fast wiggle shows up as a slow one. The interview asks why a moving average introduces lag and what to do about it. The answer is that the average is a past, and the repair is either to accept the lag and not call it a forecast, or to use a filter that states its delay.

## 2. Why it matters at the frontier

Charts full of averages are how discretionary and systematic screens still talk. The frontier failure is to backtest a crossover as if the average had known the turn. It knew the turn one window late, and the backtest that does not pay that lag is a different strategy from the one you can trade. F2 on a series. Aliasing is the quieter failure: a daily bar of a process that lives inside the day is not a milder version of the same process. It is a different object. This lesson is the warning. It is not a spectral trading system.

## 3. The test

Why does a moving average lag, and what will you do about it? The pass shows that the average sits behind a step, names the delay, and refuses to call the cross a prediction. A candidate who says "it filters noise" has described a hope. Noise and signal are not labeled in the window. Probation is a crossover backtest with the entries shifted by the window, to see what remains. A wavelet or a spectrum is the same interview if the candidate claims one: where is the frequency, and what sampling rate would have aliased it.

## 7. Now, 2026

As of this writing moving averages remain the most common overlay on a trading screen and one of the least respected as a research object, which is a dangerous combination. Spectral and wavelet methods are standard in the academic time-series literature and rare as a production signal at a trading firm. That split is practice, unverified. What has not changed is the lag. Faster computers draw the average sooner. They do not move the window into the future. A chart that does is using data the trader did not have.

## 8. Next

- By 2027, a crossover shown as a signal states the lag, or it is labeled a description (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 smoothed series used in a decision carries the window and the one-sidedness in the sealed row (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 same bars can be re-averaged from the warehouse (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) — "export function ema(values: readonly number[], period: number): (number | null)[] {"
- repo: (a file of this platform; the path is the licensed tier) — "const k = 2 / (period + 1);"
- repo: (a file of this platform; the path is the licensed tier) — "Null until the window has enough bars"
- paper: the standard moving-average and aliasing material; no DOI is claimed
- practice, unverified: a step and a late average is a shape, not a backtest
- 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-4, S-24 (skill_syllabus gives the order), then sit skill_exam on S-24 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 (91d6bb84475b22db…).
