0
$\begingroup$

I hedge a portfolio of Global Equities (200 stocks within MSCI World universe) by shorting futures on MSCI World Net Total Return. The hedge is calculated using Beta. Beta is calculated using a risk model looking at historical volatility for the past year.

When there is no market crash, volatility is relatively low and my portfolio Beta to benchmark (to MSCI world) will be say 1. Turns out that during a market crash the observed Beta was like 1.2, so the hedge underperformed (not all of the losses were mitigated).

I am concerned that the model will now take this volatility into account and this will make Beta increase. This means that I will short more futures, and if the market goes up in a less volatile way, I could lose more on the way up.

What's the market practice to avoid this issue?

$\endgroup$
  • $\begingroup$ In practice, empirical betas are often shaded towards 1, whether that is higher or lower. Why? It tends to be a better go-forward estimate for future realized beta. Another thing that may be happening here is you could have on sector bets where those sectors are more sensitive than other sectors to the current crisis so they beta again may be higher going forward. $\endgroup$ – RWP - Down by the Bay Apr 10 at 2:31
  • 1
    $\begingroup$ why not just sell the position? $\endgroup$ – Taylor May 10 at 16:02
0
$\begingroup$

Use a better model to estimate beta.

Try using Kalman Filters:

https://www.quantopian.com/posts/quantopian-lecture-series-kalman-filters

| improve this answer | |
$\endgroup$
0
$\begingroup$

Most of the literature in Finance assumes continuous hedging which is just practically impossible. Minimum variance hedge ratio assumes the same. Unless you’re a bank or HF that can cost effectively re-hedge portfolios regularly, you’ll never get anything near a perfect hedge.

For a retail investor I would say to avoid time evolving betas like with Kalman Filters because even if you can get perfect hedges, the transactions costs of implementing them would eat way your returns.

I think you should look at long-run relationships between them (either co-integration betas or normal long run betas) and do not dynamically hedge if you cannot afford to. Market practice is significantly different than what retail guys can do.

| improve this answer | |
$\endgroup$
0
$\begingroup$

You've highlighted one of the weaknesses of using beta, despite it being a major part of mean variance analysis.

You could focus on long-term beta to reduce the impact of recent volatility.

Or use Kalman filters, but understand the inherent complexity/costs you're getting into there.

| improve this answer | |
$\endgroup$

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.