Stochastic

# These walks are called left continuous since they

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: e values of M0 , X0 , X1 , . . . , Xm 1 . Having introduced the gambling strategy “Bet \$1 on each play up to time T ” our next step is to compute the payo↵ we receive when W0 = M0 . Letting T ^ n denote the minimum of T and n, i.e., it is T if T < n and n if T n, we can give the answer as: W n = M0 + n X Hm (Mm Mm m=1 1) = MT ^n (5.9) To check the last equality, consider two cases: (i) if T n then Hm = 1 for all m n, so Wn = M0 + (Mn M0 ) = Mn (ii) if T n then Hm = 0 for m > T , so the sum in (5.9) stops at T . In this case, Wn = M0 + (MT M0 ) = MT Combining (5.9) with Theorem 5.12 and using Theorem 5.9 we have Theorem 5.13. If Mn is a supermartingale with respect to Xn and T is a stopping time then the stopped process MT ^n is a supermartingale with respect to Xn . In particular, EMT ^n M0 As in the discussion after Theorem 5.9, the analogous results are true for submartingales (EMT ^n M0 ) and for martingales (EMT ^n = M0 ). 168 CHAPTER 5. MARTINGALES 5.4 Applications In this section we wi...
View Full Document

## This document was uploaded on 03/06/2014 for the course MATH 4740 at Cornell.

Ask a homework question - tutors are online