Skip to content

Research

Literature reviews

What the published evidence actually supports for each mechanism, with citations, and where our reading of it stops, before any test of our own.

Before a mechanism is tested here it is read. A literature review in this collection asks a narrow question: does the published evidence describe an economically motivated effect that was replicated outside its original sample and persisted after publication? If so, the claim must be precise enough to pre-register.

That last requirement does most of the work. A great deal of published finance describes an effect in terms that cannot be turned into a specification: the universe drifts, the horizon is chosen after the fact, the cost assumption is absent, or the data used to establish it was restated afterwards. A mechanism that cannot be stated as a testable identity is not a candidate here, however well cited it is, and several reviews end at exactly that point.

Each review states where our reading of the evidence stops. That boundary matters more than a verdict: an effect that is well documented in institutional US equities from 1990 to 2010 is not thereby documented in crypto perpetuals in 2026, and a review that quietly transfers a finding across that gap is doing the reader more harm than one that reaches no conclusion.

These reviews exist to be used against us. If one misreads a paper, overstates a replication, or ignores a contrary result, that is checkable from the citations, and it is checkable before any capital or any trial budget has been spent on the idea. Reading first is cheaper than measuring first, and it is the only stage of this process where being wrong costs nothing.

The literature reviews documents