ResearchNotes · Methodology note

Thematic universes without hindsight

Selecting a theme's members from today's market overstates the theme's history. This note sets out the three sources of that bias and the four rules we use to remove it.

Most thematic baskets are built from the present. A theme is named, the companies that fit it today are selected, and the group is tested over the preceding decade. The history that results describes the selection more than it describes the theme.

Three sources of hindsight

Selecting from today’s market carries three advantages into the historical record. None requires bad intent; each follows from starting in the present.

Survivorship. Companies that were acquired, delisted or liquidated are absent from today’s market, so they never enter the basket. They are often the theme’s least successful participants, and leaving them out removes losses that belong in its history.

Transformation. A company that is central to a theme today may have had little to do with it ten years ago. Treating it as a member throughout credits the theme with a change in the business that could not have been identified at the time.

Definition drift. Themes change shape. A definition written today reflects what the theme has become and, implicitly, which companies succeeded in it. Applied to earlier years, it selects with knowledge that was not yet available.

The same theme, two histories Panel A shows four companies assumed to be theme members from 2014 to 2024. Panel B shows membership decided from disclosures on each date: two of the four enter years later, and two further companies qualify for a period and then exit, one acquired and one delisted. A · SELECTED TODAY, PROJECTED BACKWARDS Company A Company B Company C Company D B · DECIDED FROM DISCLOSURES ON EACH DATE Company A Company B Company C Company D Company E acquired Company F delisted 2014 2016 2018 2020 2022 2024 Member, as projected from today Member, as decided at the time Exit
Exhibit 1. The same theme, two histories. In panel A, the four companies that qualify today are assumed to have qualified throughout. In panel B, membership is decided from the disclosures available on each date: two of the four enter years later, and two companies that qualified for a period and then exited appear in the record. Illustrative; the companies are hypothetical.

Four rules

We construct theme membership under four rules. The first two set what membership means and how it is evidenced; the third and fourth remove hindsight from its history and make every result reproducible.

1. Membership is defined by who pays

A company belongs to a theme when the theme describes what it sells and what its customers buy, not what the company mentions, uses or is classified as. A manufacturer of the equipment a theme depends on qualifies, including a developer that has yet to reach revenue. A company that adopts that equipment, or discusses it in its annual report, is a customer of the theme rather than a member.

The rule describes the economics of the business. It carries no view on whether the theme will perform, which keeps the investment thesis out of the membership decision. Because it describes what customers buy rather than which companies have succeeded, it is also less exposed to definition drift.

2. Every membership is evidenced from disclosures

Each member is linked to the passage in its own disclosures that establishes its role, typically the business description or a segment disclosure. Where a company reports the revenue it earns from the theme, that share is recorded with it and sets the member’s tier: how much of the company the theme accounts for. Revenue share measures exposure; it is not a condition of membership. Any membership can be traced to the company’s own account of its business.

3. History is decided with what was known at the time

To decide whether a company belonged to a theme on a given date, we use only the disclosures available on that date. A company enters when the evidence first appears and leaves when it lapses. Companies that were later acquired, delisted or liquidated remain in the record for the period in which they qualified.

4. Definitions are versioned

Theme definitions are revised as themes develop. A revision becomes a new, dated version; it does not silently rewrite results produced under the old one. Any result can be reproduced under the version that produced it, and the difference between two versions can be traced to the decisions that changed.

What changes

Research built under these rules is less flattering than a basket assembled from the present, in three predictable ways.

  • Themes appear younger. Many themes become identifiable in company disclosures only in recent years, so the usable history is shorter than a projected one.
  • Histories include failures. Companies that qualified and then exited are part of the record, with their full contribution to returns.
  • Membership changes over time. The constituents on any past date differ from today’s, sometimes substantially.

Each is a cost to presentation, not to analysis. A history built this way shows what an investor could have known at the time, and that is the history on which an allocation decision can rest.

In subsequent notes

We will apply these rules to individual themes, publishing their definitions, their membership through time and the evidence behind it. The full methodology is available to institutional investors on request.