Wilmott Forums - Market Impact Costs

Credit Suisse

Forum Navigation:

login

FORUMS > General Forum

Topic Title: Market Impact Costs
Created On Mon Nov 25, 02 02:30 PM

Topic View:

View thread in raw text format

pjfitz
Junior Member

Posts: 8
Joined: Jul 2002

Mon Nov 25, 02 02:30 PM

User is offline

Hi,
I am looking for information on market impact costs for Equities and FI. Does anayone know where i can get this (free) on the internet.
Thanks,

Rouletabille
Member

Posts: 108
Joined: May 2002

Fri Mar 14, 03 06:58 AM

User is offline

(I re-activate this old but interesting forum)

I'm looking for a simple model describing the market impact (% stock prices) as a function of size order (% dayly volume).
Any idea ? Thanks.

-------------------------
Nonius : say "I abjure USA and I pay allegiance to France". Say it !!!!!

Symplecto
Member

Posts: 40
Joined: Jan 2003

Sat Mar 15, 03 03:14 AM

User is offline

Quote
Originally posted by: Rouletabille
(I re-activate this old but interesting forum)

I'm looking for a simple model describing the market impact (% stock prices) as a function of size order (% dayly volume).
Any idea ? Thanks.

My hunch is that there is no such thing.

MobPsycho
Senior Member

Posts: 1948
Joined: Mar 2002

Sat Mar 15, 03 12:19 PM

User is offline

*****I'm looking for a simple model describing the market impact (% stock prices) as a function of size order*****

To understand the answer to this question you don't want Wilmott, you want eBay.

Find an auction for the only two remaining share cerificates from some long gone company. Buy those first two shares. You might pay, say, several hundred dollars a piece.

Now, put in a market order to buy another 200.

Since nobody expected you to be buying these shares, nobody has built an inventory. Market impact, therefore, is a function of other people's expectations of your behavior. So you need to plug in some combination of how predictable you are, minus how unpredictable some stranger who might jump in is, together with how far in advance you have warned somebody of your intentions.

To understand liquidty, understand that you don't need to buy an option to secure the right to sell to someone who was going to buy anyway. All you need to buy is information, which might be free, if it is thrown off as a byproduct of securing some other quantity.

MP

Trevor
Senior Member

Posts: 331
Joined: Oct 2001

Sat Mar 15, 03 04:32 PM

User is offline

Rouletabille,
The simplest model I've ever seen is: Order size/Average daily volume * Daily Standard Deviation. This is the old "a day's volatility for a day's volume". Some variations of this take into account the bid/ask spread, but you need an estimation of the number of trades.
Another variation I've seen is a model that take the historical relationship between order imbalance (day's volume vs. average for period), and regresses this against the resulting change in VWAP. This assumes basket trading, which normally gets you something close to the VWAP. They usually also incorporate a "market" function (probably something akin to Beta) to get an accurate picture of the potential adverse impact of the market whilst trading.

I think in the end, you'll probably have a very hard time modelling it, and you are better off taking historical results (if you have any). Slippage is very hard to model, because it's a function of liquidity and volatility-on-volatility.

Trevor

-------------------------
Sua Sponte

MobPsycho
Senior Member

Posts: 1948
Joined: Mar 2002

Sat Mar 15, 03 04:46 PM

User is offline

*****The simplest model I've ever seen is: Order size/Average daily volume * Daily Standard Deviation. This is the old "a day's volatility for a day's volume".*****

Works great until 4:15 in the afternoon

Then turns into a pumpkin...

MP

Trevor
Senior Member

Posts: 331
Joined: Oct 2001

Sat Mar 15, 03 04:52 PM

User is offline

Doesn't everything turn into a pumpkin at 4:15PM...or is that just me?...

Trevor

-------------------------
Sua Sponte

Rouletabille
Member

Posts: 108
Joined: May 2002

Mon Mar 17, 03 05:04 PM

User is offline

"A day's volatility for a day's volume", I've never heard of that but it sounds well. Simple and quite sensefull... Thanks Trevor !

-------------------------
Nonius : say "I abjure USA and I pay allegiance to France". Say it !!!!!

View thread in raw text format

FORUMS > General Forum

Forum Navigation:

© All material, including contents and design, copyright Wilmott Electronic Media Limited - FuseTalk 4.01 © 1999-2010 FuseTalk Inc.