DCT Lexicon · Simulation
Standard error
s/√N
The size of a simulated estimate's sampling error: s/\sqrtN for an average over N paths, and \sqrtp(1−p)/N for a probability p.
Drafted for the author's approval from Vol. I, Section 8.9, and Vol. II, Theorem 9.3(i)
In depth
In the book
Errors. The averages are unbiased, with standard errors and .
Vol. II, Theorem 9.3(i) (Sample Average Approximation), verbatim
Executive example
Liquidity falls below $350 million in at least one quarter in 24.6 percent of the paths, with a standard error of percentage points (Figure 8.4).
Vol. I, Worked Example 8.1, verbatim
How to read it
Read an estimate as its value plus or minus about two standard errors. Two plans whose difference is smaller than that have not been told apart by the simulation.
Drafted for the author's approval
Common misreading
A common mistake is to report a simulated probability without its error bar, or to trust a few hundred paths for a rare event.
Vol. I, LOS 8.5, verbatim
How AXIOM uses it
Every statistical result in AXIOM carries its standard error: breach probabilities, the chance of reaching the value target, the value of reading cash before a decision. A result whose passport says "statistical" is never shown without it.
Drafted for the author's approval, from AXIOM's method passports
In the book: Vol. I, Ch. 8: Stochastic Enterprise Dynamics · Vol. II, Ch. 9: Stochastic Enterprise Optimization
Related: Monte Carlo simulation · Breach probability
V1.ME.8.MonteCarlo · V2.Thm.9.3


