Every number on this page is computed by the Volume II Chapter 11 notebook's own code, on Chapter 9's 100,000 paths (seed 8) and 100,000 fresh paths (seed 26209); the lab's seed 26211 draws only the simulated records of the coverage check. Your notes stay on this page until you leave it; copy the memo before you go.
The audit committee accepts Chapter 9's law as the treasury's and the engineering office's best estimate, but not as a fact. It asks what the records behind the law allow, what keeping the appetite under every law they allow would cost, and whether a ball around the law would do instead. You support the team preparing the answer.
Use AXIOM to turn records into a family, find its worst law and price it, compare balls around the law, and set a radius for a record's own sampling error.
Worked Example 11.1: the treasury's record holds two closures in twenty years, the declared rate of 0.10 a year; the engineering office's record holds forty quarters of deviations, 39 innovations of spread $50 million. Chapter 9 kept the appetite, at most a 5 percent chance that liquidity falls below the $350 million floor in some quarter of the program, with Volume I's package: a $170 million line and two-thirds of a quarter of delay, $10.44 million.
In the record panel, compute confidence bounds for the rate of closures and the spread of deviations; sweep each across its range and record the chance of a breach of Volume I's package.
Find the family's worst law, design the cheapest package that keeps 5 percent there, and check it on fresh paths and across the family.
Chapter 9's design at each declaration's worst law: the line on a $5 million grid up to $400 million, the least delay at each line. Chances in percent on the book's paths (fresh paths in brackets).
In the divergence and transport panels, compute the worst chance of a breach over balls of growing radius and the cost of keeping 5 percent; find the radii that hold the records' doubts.
A path moves by shifting its deviations, its closures fixed, at a cost equal to the largest shift in any quarter: its distance from a breach is its thinnest margin above the floor. The worst law in the ball pushes the thinnest paths below the floor first.
Design on records of 1,000 paths with radii and with margins; check each design on fresh paths and compare reliability per dollar.
Compute each quarter's Cantelli bound under a package and the cost of keeping all of them at 5 percent.
Under Volume I's package: each quarter's margin above the floor, its mean and spread over the book's paths, and Cantelli's worst chance σ²/(σ² + μ²) that it falls to the floor.
Chances of a breach under Chapter 9's law and under the records' worst law (closures 0.315 a year, spread $61.6 million), on 100,000 paths; fresh paths in brackets ($ million; delays in quarters; percent).
A chance that moves little with the rate of closures and much with the spread (Theorem 11.1(iv); Proposition 11.2); balls whose radii hold the doubts but whose appetites no package can meet (Theorems 11.2(ii) and 11.3(iii)); radii that buy reliability on small records at about the price of a margin; and a moment bound that fears a lump (Proposition 11.1(iii)).
The records' doubts, the family's worst law and its package, the price of ambiguity, and the record worth improving.