Doob’s optional sampling theorem
Given a filtered probability space , a process is a martingale![]()
if it satisfies the equality
for all in the index set![]()
. Doob’s optional sampling theorem
![]()
says that this equality still holds if the times are replaced by bounded stopping times . In this case, the -algebra is replaced by the collection
![]()
of events observable at the random time (http://planetmath.org/SigmaAlgebraAtAStoppingTime),
In discrete-time, when the index set is countable![]()
, the result is as follows.
Doob’s Optional Sampling Theorem.
Suppose that the index set is countable and that are stopping times bounded above by some constant .
If is a martingale then is an integrable random variable![]()
and
| (1) |
Similarly, if is a submartingale then is integrable and
| (2) |
If is a supermartingale then is integrable and
| (3) |
This theorem shows, amongst other things, that in the case of a fair casino, where your return is a martingale, betting strategies involving ‘knowing when to quit’ do not enhance your expected return.
In continuous-time, when the index set an interval of the real numbers, then the stopping times can have a continuous distribution and need not be measurable quantities. Then, it is necessary to place conditions on the sample paths of the process . In particular, Doob’s optional sampling theorem holds in continuous-time if is assumed to be right-continuous.
| Title | Doob’s optional sampling theorem |
|---|---|
| Canonical name | DoobsOptionalSamplingTheorem |
| Date of creation | 2013-03-22 16:43:41 |
| Last modified on | 2013-03-22 16:43:41 |
| Owner | skubeedooo (5401) |
| Last modified by | skubeedooo (5401) |
| Numerical id | 8 |
| Author | skubeedooo (5401) |
| Entry type | Theorem |
| Classification | msc 60G44 |
| Classification | msc 60G46 |
| Classification | msc 60G42 |
| Related topic | Martingale |
| Related topic | StoppingTime |
| Related topic | SigmaAlgebraAtAStoppingTime |