Snell envelope

The Snell envelope, used in stochastics and mathematical finance, is the smallest supermartingale dominating a stochastic process. The Snell envelope is named after James Laurie Snell.

Definition

Given a filtered probability space $(\Omega ,{\mathcal {F}},({\mathcal {F}}_{t})_{t\in [0,T]},\mathbb {P} )$ and an absolutely continuous probability measure $\mathbb {Q} \ll \mathbb {P}$ then an adapted process $U=(U_{t})_{t\in [0,T]}$ is the Snell envelope with respect to $\mathbb {Q}$ of the process $X=(X_{t})_{t\in [0,T]}$ if

$U$ is a $\mathbb {Q}$ -supermartingale
$U$ dominates $X$ , i.e. $U_{t}\geq X_{t}$ $\mathbb {Q}$ -almost surely for all times $t\in [0,T]$
If $V=(V_{t})_{t\in [0,T]}$ is a $\mathbb {Q}$ -supermartingale which dominates $X$ , then $V$ dominates $U$ .[1]

Construction

Given a (discrete) filtered probability space $(\Omega ,{\mathcal {F}},({\mathcal {F}}_{n})_{n=0}^{N},\mathbb {P} )$ and an absolutely continuous probability measure $\mathbb {Q} \ll \mathbb {P}$ then the Snell envelope $(U_{n})_{n=0}^{N}$ with respect to $\mathbb {Q}$ of the process $(X_{n})_{n=0}^{N}$ is given by the recursive scheme

U_{N}:=X_{N},

U_{n}:=X_{n}\lor \mathbb {E} ^{\mathbb {Q} }[U_{n+1}\mid {\mathcal {F}}_{n}]

for

n=N-1,...,0

where $\lor$ is the join (in this case equal to the maximum of the two random variables).[1]

Application

If $X$ is a discounted American option payoff with Snell envelope $U$ then $U_{t}$ is the minimal capital requirement to hedge $X$ from time $t$ to the expiration date.[1]

gollark: I dislike how browsers made CSRF a thing, it is total bees.

gollark: One of these days I really ought to add login and CSRF prevention.

gollark: ```javascriptimport m = require("mithril")import * as RPCTypes from "../common/rpc"export const sendMessage = (msg: RPCTypes.Message): Promise<RPCTypes.MessageResponse> => { return m.request( { method: "POST", url: "./rpc/", body: msg, }).then(res => { const [ type, p1, p2 ] = res if (type === "error") { throw new RPCTypes.RPCError(p2, p1) } else if (type === "ok") { return p1 } else { throw new Error("Invalid RPC response") } })}const handler = { get: (target, prop) => (...args) => sendMessage([prop, ...args])}export const serverProxy = new Proxy({}, handler)```

gollark: The RPC thing and some JS hax on the client mean I can basically just call any function the server provides as if it's a local one (except asynchronously).

gollark: minoteaur is just plain RPC - you do `POST /rpc` with a function and its arguments as JSON.

References

Föllmer, Hans; Schied, Alexander (2004). Stochastic finance: an introduction in discrete time (2 ed.). Walter de Gruyter. pp. 280–282. ISBN 9783110183467.

This article is issued from Wikipedia. The text is licensed under Creative Commons - Attribution - Sharealike. Additional terms may apply for the media files.

[FollmerSchied-1] Föllmer, Hans; Schied, Alexander (2004). Stochastic finance: an introduction in discrete time (2 ed.). Walter de Gruyter. pp. 280–282. ISBN 9783110183467.