Assume we have an FX portfolio with $n$ currency pairs, such that $w_i$ is the weight of currency pair $i \in \{1, \dots, n\}$ in the portfolio, and $\sum_{i=1}^n w_i = 1$. All pairs have USD as one of the legs (USD is the base currency of the portfolio, and I would like to allocate to other currencies using Markowitz). I would like to compute a concentration measure of the portfolio, and believe that the Herfindahl-Hirsch-Index could be an option (see Chammas, Portfolio Concentration link ). However, Chammas assumes that the portfolio only contains long positions. How should I proceed if one of the portfolio weights is negative?

For example, lets assume that the weight of the USDJPY is $-0.5$. Can I simply input that weight in the computation of the HHI?

Is there a more appropriate concentration measure than the HHI for my problem?

  • 2
    $\begingroup$ You might consider the Yager Entropy $E_Y = \sum_{i=1}^N | w_i-\frac{1}{N}| $ where the $w_i$ are the portfolio weights. The minimum (i.e. least concentrated portfolio) is achieved when $w_i=1/N$ like for Shannon Entropy but the advantage is that the $w_i$ can be negative. It is a bit obscure and there are only a few papers about using it in a portfolio context AFAIK. $\endgroup$
    – nbbo2
    Jun 12, 2021 at 23:30
  • 1
    $\begingroup$ At the risk of extreme stupidity if you are insisting on long-only positions, why not treat a 0.5 short in USDJPY as a 0.5 long in JPYUSD? Returns measured in the log-price wouldn't be different. $\endgroup$
    – demully
    Jun 15, 2021 at 1:35


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.