Tool lesson

Correlation Matrix: Choose Windows Without Cherry-Picking

A practical Correlation Matrix lesson for comparing 60D, 90D, 180D, and 365D relationships as evidence windows before writing a desk note.

13 minBeginner6 chapters

Educational only

The examples teach workflow and risk framing. They do not provide trade recommendations, personalized advice, leverage guidance, or guaranteed outcomes.

Chapter 01

Treat window length as a model assumption

Trader question

Which completed-history window built this relationship read?

A period setting is not decoration. It decides which aligned returns enter the sample, so the first safe habit is to name the window before interpreting the coefficient.

Desk checklist

Name the active period before reading the cell.
Treat the period as an assumption.
Avoid copying a conclusion across windows.

Interactive proof

Matrix period control and period rail

Open the window lab and choose the period label that belongs in the desk note.

1WindowModel assumptionA 60D relationship and a 365D relationship answer different questions, so the period label belongs in the note.

2BaselineLonger contextA longer window helps the learner see whether the short read is normal, stretched, or unusual.

3ComparisonNo single-window proofMulti-period comparison protects against choosing only the window that supports the preferred story.

4Direction changeReview promptStrengthening and weakening describe what changed in the sample; they do not forecast the next move.

5Pause ruleDislocation or regime?If one short window disagrees with the baseline, the safest output is a pause and a regime check.

Window choice is part of the evidence. A short-term relationship that disagrees with the longer baseline becomes a review prompt, not a forecast.

Interactive desk lab

Correlation Window Lab

A practical Correlation Matrix window-comparison lab for toggling 60D, 90D, 180D, and 365D reads, spotting cherry-pick risk, and writing a cautious pause reason.

Native scroll

44s Remotion sceneCorrelationWindowAssumptionStackVideo

Window as an assumption

Four period cards appear as assumptions, not proof buttons.

Storyboard beats4 cues

A relationship question opens above an empty desk note.

60D, 90D, 180D, and 365D cards stack in sequence.

Each card stamps a different sample label on the same pair.

The final frame says: period belongs in the note.

Remotion code

CorrelationWindowAssumptionStackVideo

The snippet is stored with the lesson so a future Remotion project can render the chapter video.

Show component snippet

import {AbsoluteFill, Easing, Sequence, interpolate, useCurrentFrame} from "remotion";

const windows = ["60D", "90D", "180D", "365D"];

export const CorrelationWindowAssumptionStackVideo = () => {
  const frame = useCurrentFrame();

  return (
    <AbsoluteFill style={{background: "#fff8e8", color: "#071126", padding: 72}}>
      <h1 style={{fontSize: 56, lineHeight: 1}}>The period is an assumption.</h1>
      <p style={{fontSize: 26, color: "#5c6576", maxWidth: 820}}>The same pair can answer different questions when the window changes.</p>
      <div style={{marginTop: 54, display: "grid", gridTemplateColumns: "repeat(4, 1fr)", gap: 14}}>
        {windows.map((item, index) => {
          const progress = interpolate(frame, [index * 18, index * 18 + 18], [0, 1], {extrapolateLeft: "clamp", extrapolateRight: "clamp", easing: Easing.bezier(0.16, 1, 0.3, 1)});
          return <div key={item} style={{opacity: progress, transform: "translateY(" + (28 - progress * 28) + "px)", border: "2px solid #ad862d", background: "#fffdf7", padding: 24, minHeight: 150}}>
            <span style={{fontSize: 18, color: "#805407", fontWeight: 900}}>Window</span>
            <strong style={{display: "block", fontSize: 52, marginTop: 12}}>{item}</strong>
            <span style={{display: "block", color: "#5c6576", fontSize: 18}}>sample label</span>
          </div>;
        })}
      </div>
      <Sequence from={88} layout="none">
        <div style={{marginTop: 34, background: "#071126", color: "#fff8e8", padding: 22, fontSize: 28, fontWeight: 900}}>Desk note must say which window built the read.</div>
      </Sequence>
    </AbsoluteFill>
  );
};

Sources used for this tutorial

CORRELATIONNIST/SEMATECH Engineering Statistics HandbookUsed for Pearson correlation direction and coefficient framing across different historical samples.Use and Misuse of Correlation CoefficientsPenn State STAT 509Used for sample-selection, outlier, and causation caveats when a learner changes windows.Evaluating Correlation Breakdowns during Periods of Market VolatilityFederal Reserve Board IFDPUsed to explain why relationships can change during volatility regimes.International Asset Allocation with Time-Varying CorrelationsNBERUsed as a research anchor for time-varying correlations and regime-sensitive diversification.Spread Trading Sector Index FuturesCME GroupUsed for practical rolling-sample and unstable-estimate caveats.Commodity Trading Systems Sold on the InternetCFTCUsed as a guardrail against turning period comparisons into mechanical system claims.

Next step

Open the tool with the checklist beside you.

Move from the lesson into the matching Bullion Brains tool, keep the checklist visible, and treat the output as evidence until the caveats are clear.

Open Correlation Matrix