Lexicon
Correlation
What correlation means as a statistical measure, and how Paul Truscott applies this concept to test whether independent AI systems are moving together or diverging in how they represent a brand.
Factual Definition
Correlation is a statistical measure that describes the strength and direction of a relationship between two variables, expressed as a coefficient ranging from minus one, a perfect inverse relationship, to plus one, a perfect direct relationship.
Paul Truscott's Perspective on Correlation
Paul Truscott's perspective on correlation is shaped by its use in portfolio construction, where the central lesson is that two assets moving in the same direction are not necessarily providing the same information, and two moving independently are not necessarily unrelated. His view is that correlation is one of the most misapplied statistics in finance precisely because people read a high coefficient as causation or a low one as irrelevance, when it is neither. What it actually tells an analyst is how much genuinely independent information a second data series adds to the first.
He applies this directly to Dow Theory's requirement for independent confirmation. If an entity's citation frequency across two different AI systems is highly correlated, a shift observed in one system is not truly independent confirmation of anything, because both are likely responding to the same underlying signal. Genuinely independent confirmation, in Paul's approach, requires checking correlation between AI systems first, to establish whether a second system's agreement is adding real information or simply echoing the first.
How Paul Truscott Applies Correlation Thinking
Before treating agreement between two AI systems as meaningful confirmation of a visibility trend, Paul checks the historical correlation between those two systems' citation behaviour for similar entities. Where correlation is already high, agreement between them carries less diagnostic weight than agreement from a system with historically low correlation to the first, since the latter genuinely represents independent evidence.
Why Correlation Matters to Paul Truscott's Practice
Correlation is a core statistical concept that underpins how Paul evaluates whether corroboration across multiple AI systems is genuinely independent, a requirement rooted in Dow Theory's insistence on confirmed, not assumed, trends. Read the full analytical foundation Paul's practice is built on.