THEORY A / RELATIONSHIP LAB

Valuation in contextCorrelations

Explore what moves your stock. Test what brings it back.

LOCAL HISTORY0/3 READY
REVIEW THROUGH2026-09-16
02 / SHARED MOVEMENT

Price paths & relationship windows

All series rebased on their first shared date
Latest in viewAAPL MSFT SPY
No aligned observations in this window.

Drag to inspect a window · Arrow keys explore · Enter sets selection edges · Esc clears

Positive co-movementNegative co-movementInspected windowHighlights: AAPL / MSFT

Provider closing prices, not verified total returns. Corporate actions, currencies, and different market closing times can affect comparisons. Missing dates are never forward-filled.

03 / FIND & QUESTION

Relationship scanner

EXPLORATORY
Correlations with AAPL. Click a comparison to inspect its windows.
ComparisonIn view20 obs.60 obs.120 obs.Windows

No sustained windows at these settings. Try another comparison or adjust the threshold.

Windows mark consecutive rolling estimates above the threshold, not proven significance. Their start is retrospective; “detectable” is when persistence was first satisfied. Rolling estimates may use history before the visible start. Scanning and tuning many windows increases false discoveries.

04 / RESTORING BEHAVIOR

Does the relationship pull back?

AAPL / MSFT

Fit a reference relationship, then examine subsequent departures. Changing the chart window never refits this model.

Add a comparison and fit a formation period to examine its restoring behavior. An initial fit is prepared when sufficient history is available.
05 / WEIGH THE EVIDENCE

What supports acting, waiting, or revisiting?

VALUATION

Define your valuation anchor

Enter expected EPS and a target multiple, then record what would make you reconsider them.

BROADER EXPOSURE

Choose a benchmark

Add a market or sector benchmark to examine how much co-movement remains after accounting for it. Shared exposure does not establish causation.

RESTORING BEHAVIOR

Build a reference relationship

More observations after fitting are needed to judge whether departures tend to recover. This describes price behavior, not fundamental value.

Keep these observations separate: persistent coupling can coexist with a widening valuation gap. Fundamental change requires reviewing earnings and business assumptions in Analyze. These diagnostics do not identify crowded positioning or issue a trading recommendation.

Prices persist on this device; chart manipulation stays local.
Data & methodology

Daily log returns, joined on both interval endpoints. Pearson correlation; benchmark adjustment uses partial correlation on at least 20 shared intervals. The spread is an ordinary least-squares fit of log prices, standardized by its frozen formation-period residual deviation. It is not a cointegration test. All research settings, notes, and thesis snapshots are local to this browser.

Price history refreshes after six hours or on request. A server cache may still be served. History extends up to five years plus a warm-up period. Your provider’s adjustment convention is not guaranteed by the current API. Do not interpret split artifacts, currency moves, or asynchronous market closes as an economic relationship.

  • AAPL: 0 closes · latest unavailable · fetched unknown / stale
  • MSFT: 0 closes · latest unavailable · fetched unknown / stale
  • SPY: 0 closes · latest unavailable · fetched unknown / stale