Serving the Quantitative Finance Community

 
User avatar
Siberian
Topic Author
Posts: 0
Joined: June 25th, 2003, 8:56 pm

fx liquidity forecasting

May 21st, 2008, 8:41 pm

have seen a lot of research and data on equity t-cost/liquidity forecasting, not a whole lot on fx though, would greatly appreciate any help.Regards,Ev
 
User avatar
LiKaShing
Posts: 0
Joined: July 12th, 2007, 10:34 am

fx liquidity forecasting

May 23rd, 2008, 12:53 pm

Liquidity is not a problem in the FX-market where you can get a better price for a large order-size. In equities the opposite holds, large order size means large market impact and a high transaction cost.Therefore transaction cost/risk research is scarce for FX (and futures).
 
User avatar
Siberian
Topic Author
Posts: 0
Joined: June 25th, 2003, 8:56 pm

fx liquidity forecasting

May 23rd, 2008, 3:02 pm

is it the same for the em ccies as well?are you saying there is no need for t-cost model in ccies at all?Thanks for the reply.Ev
 
User avatar
Siberian
Topic Author
Posts: 0
Joined: June 25th, 2003, 8:56 pm

fx liquidity forecasting

May 23rd, 2008, 6:17 pm

Also, is it really true if you compare a 1MM trade with a 100MM or 500MM trade? For some reason i doubt that the trader at the bank will be willing to take a large sum like that at the mid.
 
User avatar
opsho
Posts: 0
Joined: April 7th, 2006, 5:46 pm

fx liquidity forecasting

May 26th, 2008, 1:36 pm

It depends on the currency in question. If you are talking about a deeply liquid currency like Eur then maybe you could get out 1 Eur and 50 Eur at the same price. For most currencies though once you get north of about $25 then the bid/ask will widen out. If it's an emerging market or minor currency, the amount could be a lot lower for price impact as there is less liquidity
 
User avatar
Siberian
Topic Author
Posts: 0
Joined: June 25th, 2003, 8:56 pm

fx liquidity forecasting

May 26th, 2008, 7:12 pm

i looked at some USDNOK trades, the price impact is drastically different from small size to mid to large size trades, varies much more than in USDEUR. I would like to come up with a good forecasting model for liquidity/t-cost of an fx trade as a function of the size. When i polled the brokers, all the quotes were upward sloping in size, none of them were exhibiting lower costs at higher sizes, so i am a bit perplexed. I would not assume that a broker will be interested in trading 1bn kiwi at the lower spread than 250mm, and my preliminary research supports that.Wonder if any bank/dealer would sell the fx-trade data, so at least i can get some kind of number to run analysis on.Thanks for the inputs.Ev
 
User avatar
LiKaShing
Posts: 0
Joined: July 12th, 2007, 10:34 am

fx liquidity forecasting

May 27th, 2008, 9:41 am

The best is probably to make your own statistical study. Transaction data for model fitting is not easy to come over.Perhaps you can implement a model similar to the one suggested by Almgren in Risk Magazine 2005 - http://www.courant.nyu.edu/~almgren/pap ... testim.pdf - or, at least something simliar to this.
 
User avatar
Siberian
Topic Author
Posts: 0
Joined: June 25th, 2003, 8:56 pm

fx liquidity forecasting

May 27th, 2008, 12:55 pm

Thanks LiKaShing, i have seen these papers, the problem is that being a relatively new shop we don't have that much of our own data, definitely not enough for exploratory analysis.Will keep digging.Thanks everyone.Ev
 
User avatar
opsho
Posts: 0
Joined: April 7th, 2006, 5:46 pm

fx liquidity forecasting

May 29th, 2008, 7:28 pm

I think you'll be hard pressed to find a dealer willing to share their own data. Most of the data is guarded pretty tightly (speaking from experience).In terms of the spreads widening with transaction size...there are a few points that may help:(1) for the most part, the fx market trades in a certain transaction size for a give currency;(2) fx prices move fairly quickly;(3) dealers have private information regarding client flow and orders;(4) there has to be an economic reason for a dealer to want to take on a large transaction (e.g. suits their risk profile, suits their market view, offsets another position)(5) bigger transactions take longer to move into the marketIf you assume the above, a wider spread for a large transaction is really nothing more than a risk premium for the additional temporal risk.