DCT Lexicon · Optimization
Stochastic dominance
Plan A dominates plan B at first order when, at every level of value, A is at least as likely as B to reach it; at second order when A's average shortfall below every level is no larger than B's.
No book source: the volumes do not define stochastic dominance (measured: no occurrence in either volume or the online appendices). The definition is the standard one; in AXIOM it is a proposed reading (AXIOM.PROPOSED.STOCHASTIC_DOMINANCE)
In depth
Formal statement
For value distributions and : first-order dominance, for every ; second-order dominance, for every .
Drafted for the author's approval: the standard definitions, not the book's
How to read it
First-order dominance means every board, whatever its attitude to risk, prefers A as long as it prefers more value to less. Second-order dominance means every risk-averse board does. When neither plan dominates, the choice depends on the board's appetite.
Drafted for the author's approval
What the book does instead
The book compares plans path by path on common random numbers and by their breach (Vol. II, Proposition 15.2). AXIOM shows dominance beside that comparison, as a proposed reading, and treats a violation smaller than the standard error of the mean difference as within noise.
Vol. II, Proposition 15.2; AXIOM's engine documentation
How AXIOM uses it
Scenario comparison plays two plans on 100,000 common paths and reports how often one beats the other, the distribution of the difference, and whether either dominates the other at first or second order, tested exactly on the sample.
Drafted for the author's approval, from AXIOM's method passport "Scenario comparison" (in flight on 29 October 2026, not yet on main)
In the book: Vol. II, Ch. 15: Enterprise Digital Twins and Autonomous Enterprise Transformation
Related: Common random numbers · Risk appetite


