# Definition of risk factors for market risk scenario testing

I am doing a research for stress testing in market risk. The usual process I found out for scenario testing is:

1. Define risk factors upon the portfolio
2. Define the desired scenarios
3. Vary the risk factors and execute the different scenarios
4. Interprete the scenarios

My problem is that I do not know how to define the risk factors for a portfolio? Which R functionality could I use? For example, what are the risk factors for a simple FX portfolio when I consider the portfolio: EURUSD, USDMXN, AUDUSD, USDJPY and USDKRW?

-
I voted to close this as it asks for interpretation of scenarios - something open-ended and subjective, all of which, with agreement from @Quantlbex, validates removal. – madilyn Jul 19 '13 at 5:19
@kristine What did Quantlbex write on here? – chrisaycock Jul 19 '13 at 16:21
@chrisaycock I haven't written here yet. I guess kristine is refering to (part of) the justification I gave to close one of her questions which was about conferences and networking events. – QuantIbex Jul 19 '13 at 16:42
@kristine In my opinion, you misunderstood the question as it does NOT ask for interpretation of scenarios. It asks for the methodology to define risk factors, an implementation of it in R, and the risk factors for a specific portfolio all of which are not subjective nor open-ended. As such, I don't think the question should be closed. – QuantIbex Jul 19 '13 at 16:43
@QuantIbex Thanks for the clarification. That conference question was totally off-topic, whereas this question about risk factors looks totally fine. I think kristine is just upset about something since she voted to close a bunch of questions all at once. – chrisaycock Jul 19 '13 at 16:46

There exist a lot of way to choose risk factors and the choice differs according to the kind of underlying assets.

In your case, particularly, since the portfolio is composed by currencies, I would choose the risk factors mainly among all the macroeconomic variables available in your dataset or data provider.

After that, to choose on which of them basing the stress test, evaluate which ones are more predictive of the portfolio returns; again, there exists in academic literature different ways to evaluate the predictive ability of a variable and it is up to you choose which one (correlation, linear regression model $adj-R^2$, granger-causality (particular cases only),...). I suggest to read:

Wickens, Michael R., and Peter N. Smith. "Macroeconomic sources of FOREX risk." Univ. of York Econ. Discussion Paper 2001/13 (2001).

that shows synthetic measures of risk in the forex market.

Alternatively, choose them on judgemental/expert way, trying to give an economic/rational explanation.

Hope this helps.

-
Link to paper, www-users.york.ac.uk/~pns2/forex.pdf – pyCthon Jul 6 '15 at 0:10