# Hedge Fund risk management on a daily basis

Since Hedge Funds/Fund of Funds report on a monthly basis usually within 10 days after the month end, monitoring and managing (hedging) potential risks is quite a difficult task. Having done some research, there seems to be some solutions to this problem.

Let's just say that there is a FoF, under which there are around 50 HFs with different strategies. Managing risk can be done on a FoF level as well as on each HF level. The latter is more tedious but offers better understanding (as in if a particular HF is a Long/Short Equity fund, its obvious that they are exposed to equity risk).

Thus far I have looked at HFRI Equity Hedge Index and regressed it against Fama-French factors. The result is not perfect (24 months rolling regression), but is alright. However, this exact approach is not yielding anything meaningful for Credit, Multi-strat, Fixed Income funds (maybe I am not using correct factors). I will probably keep trying for a while.

There are two more options - PCA and Factor Analysis. So I thought maybe there is somebody who have taken a look at this problem and is willing to share his/her experience. Any input is appreciated.

• Does anybody have any input? – wannabe Dec 9 '15 at 11:29

Have a look at the following discussion between PCA versus FA : https://stats.stackexchange.com/questions/1576/what-are-the-differences-between-factor-analysis-and-principal-component-analysi, and choose what you think is theoretically adequate to your case.

[EDIT]

In regards to Risk Monitoring, EM Algorithm FA could be used to build Latent Factor Models on Alternative Indexes. My advice is to access some databases such as : indexiq, alternativeucits, etc. to obtain sufficient time series (returns) on the different strategies adopted or tracked by your funds. This might be relatively less costly (in terms of Time) than replicating those third-party index methodologies.

Ps: Few data (index strategies) could also be found on Bloomberg such as Newedge CTA Trend..

Once your Multi-Factor Model is built using EM algorithm (e.g loading, Score..), you could derive your Ex-Ante Sigma, and also easily apply Monte Carlo simulations on your Factors to derive VaR...

• great discussion, will definetily read. thnx – wannabe Dec 10 '15 at 10:28
• @wannabe: glad :) the feedback matches your expectation. Best, – owner Dec 10 '15 at 11:07

Stumbled upon IndexIQ's methodology here: http://www.indexiq.com/docs/iqhgms/iiqhedgeiqindexes.pdf

Might be another alternative to replicate HF returns/risk and risk manage accordingly on a daily basis.

Feel free to comment/discuss.

• what kind of risk monitoring are you doing? which risk metrics would you like to use in conjunction with requested info on PCA / FA? explain a bit why do you think that you need to replicate returns... – owner Dec 10 '15 at 14:27
• @owner, we have several portfolios - Bond Portfolio (mostly IG bonds), stocks, Hedge Funds portfolio, Private Equity portfolio, etc. Monitoring the Bond/Stock Portfolio is pretty easy and everything is available on a daily basis; HF performance comes in on a monthly basis with 10 days lag. Let's say EM is selling off, so we predict that some EM oriented funds will also report bad results. Ideally we'd like to hedge this out, but for now we're trying to just see replication results and risk monitor (sigma, VaR, DD, worst return). – wannabe Dec 11 '15 at 2:29
• cheers. see edited answer. hope it helps, Best – owner Dec 11 '15 at 8:27

I think you are approaching your portfolio from the complete wrong starting point. Your "securities" are the individual hedge funds. You should be using individual strategies to "hedge" the other strategies. It sounds like you're attempting to create a hedge overlay.

If you wanted a diversified equity portfolio, would you buy a 50 names in with $\beta = 1.5$ to one specific subsector, and "hedge" by shorting an ETF for that same subsector and going long an ETF for another sector?

Also, you should not be overlaying a hedge for your hedge fund portfolio. They should already be hedged. The risks that hedge funds take exposure to should be calculated and precise. You should not have unwanted sector, geopolitical, etc. risk in your portfolio because a good hedge fund manager will have a very narrow strategy that only exposes itself to certain risks. The more risk you hedge away, the lower your returns will be. That defeats the purpose of even hiring the fund.

• "Also, you should not be overlaying a hedge for your hedge fund portfolio. They should already be hedged." - Haha :) – Chris Taylor Jul 5 '17 at 7:34

Here are a few things you can do:

1. There are software vendors where quite a few hedge funds submit their positions on a monthly basis so you can run daily risk assessments assuming that the portfolio stays the same (might be a wrong assumption though pending on the turnover of the underlying HF)

2. Regarding the return-based analysis, I have been recently having a look at the kalman filter instead of the regression. There is also the dynamic style analysis based on FLS.

3. Try to find statistical relationships between the returns and measure the 'true exposure' of your portfolio.