We are in 2011 and yet many financial companies still have no clue when it
comes to risk management.
A good example is swissquote and it's so called "Risk Tools"
Swissquote claims to offer a Value-at-Risk "Risk
Tool" to
their customers.
As we will see, the swissquote calculations have
nothing to do with
industry standard
Value-at-Risk [VaR].
To calculate it's misnamed "historical Value-at-Risk" Swissquote takes the history of your portfolio's past returns to compute the 5th/95th percentile.
Obviously computing VaR is much more involved than that!
Swissquote now claims it also combines a ten day Monte-Carlo VaR when the asset is an option!
If only things were that simple. You first come up with a "weak" implementaiton and then you pepper some Monte-Carlo on top of it, jic !
This is NOT how industry standard Value-at-Risk is computed !
Any Value-at-Risk computation must:
1) Be clearly explained and follow
industry standard.
There are good reasons why these
standards were established.
Putting together your own methodology as you
go along and then calling it Value-at-Risk is unethical . . . .and
sloppy !
2)
All values, including initial risk factor data and intermediate
computations, must be available to the
user so he can audit and reproduce every calculation step in the
sequence.
3) Values
should be consistent according to methodology and simulation horizon.
4) It is good practice to provide all three Methodologies separately. The three methodologies can then be compared ceteris paribus. These three methodologies must then be complemented with proper predictive stress tests and marginal measures in order to identify hot spots.
5) Proper Risk Management solution should also compute excessions
as this is the point of the exercise.
Historical Value-at-Risk was designed to
capture
idiosyncratic
risks which are lost when you use a correlation matrix as past asset price
swings are "averaged"
over the period.
Obviously this backward looking black box provided by
Swissquote defeats twice the purpose of using historical simulation as this
information is lost into aggregated risk factors that are priced into the asset
and then on specific assets with the correlation matrix of the Monte-Carlo
simulation.
This quick and dirty approach has important consequences when measuring risk:
Risk will always be different
for non-linear instruments such as:
Bonds,
Equity,
Commodity Futures and Forwards.
All Other Derivatives
As well as pseudo-linear instruments with pre-defined maturity:
Time-Deposits
Cd's
Fx-Fwds
etc.
On the other hand this approach yields similar results when dealing with linear instruments, such as Cash or Equities (with no dividends) thanks to a numerical coincidence between the pricing function (linear) the sampling frequency (i.e. daily price returns) which are then applied to an identical risk horizon of one-day.
Unfortunately, this does not hold true for non-linear instruments, which are also the most risky.
Conclusion:
Swissquote is clearly not a leader when it comes to Financial Risk Management !
This is not surprising as brokers never deal with Financial Risk themselves, instead they just take a commission and pass on the risk to buyers and seelers!
So, as a customer be Very careful when using the Swissquote "Risk tools" with anything other than cash assets as this is not how it is done in the Financial Industry !