[ts-gen] Beware of IB's notion of a position for FX
Ken Feng
kfmfe04 at gmail.com
Wed Oct 7 00:45:38 EDT 2009
I have no questions in this post - this post is for informational
purposes in case others find it useful.
-----------------------------------------------------------
There are at least two notions of a flat portfolio (I will omit others
for the sake of clarity/simplicity):
A. If I buy a security and then sell it later, I go from flat to open
position to flat.
B. If my portfolio only holds my book currency, then I am flat.
IB uses A., while a typical fund or risk manager's perspective of flat
is B. Now, if you are just trading domestic equities or futures, A.
and B. are effectively the same. However, once you start to trade FX
or foreign securities, A. and B. may be different.
For example, suppose I start with a USD base-currency portfolio funded
with USD only. I may buy USDJPY, sell EURJPY, and buy EURUSD, a
triangular arbitrage.
In case A., it looks like I have a long USDJPY, short EURJPY, and long
EURUSD position.
In case B., I am flat, because I have no exposure to any foreign
currencies - I only hold USD, my book currency.
In TWS, you can see an example of A. when you look at any trading tab
down the "Position" column - IB says I have long and short positions
when in fact, my portfolio has no exposure to market forces. If I
follow what TWS says, I would incur additional commissions and bid-ask
crossing expenses by trying to liquidate those positions, when in
fact, I am already all in USD!
This difference has implications on what information I want to pull
out of IB/TWS. If I care about B., then I can't rely on Position from
IB. Luckily, when doing a select acct; the "CashBalance" entries do
give sufficient information to tell me if I am flat or not, with
respect to B. There may be other places where calculated values for
FX are no good - you must keep your own accounting from lower level
raw data like trades or cash balances, and avoid reconciling with
higher level data.
- Ken
More information about the ts-general
mailing list