Choosing your risk

By Marcus Williamson

Risk is not a simple entity, it comes in many flavours, and requires respect and consideration even when you least expect. If one was to try ‘totally avoid’ risk in their market endeavours they would most likely just be holding cash – where are the returns in that?

The truth is that in almost all circumstances one must take at least some form of risk to have the chance of potentially generating returns. Let’s face it, you’ve got to crack some eggs to make an omelette!

Those, including myself, who have experienced first hand what it is like to feel the full effects of taking on too much risk, will most likely find it a life changing experience. You will either withdraw your remaining investments and consider other ventures, or you will revisit your mistakes with single goal; of never ending up in a similar situation again.

The trick to risk is not all out avoiding it, but instead choosing it. Ask yourself ‘What risks am I willing to take to give myself the chance of generating some prospective returns?’. Nothing is for certain – except uncertainty and its associated risk!

By choosing your risk wisely, not only are you enforcing damage limitation but you may even improve you risk returns profile as a result.

Turning to the most recent example I have dealt with regarding risk; the student run equity fund at Bath University (SOBIC). I recently took on the role of managing risk within our portfolio with a view to help us better choose our risk going forwards. Whilst I haven’t had much in the way of input on the risk side as of yet, I have had the opportunity to revisit some of the investment decisions which were made by the funds previous managing partners.

At SOBIC we are principally a long only equity fund focussing on value, using DCF models to identify areas of opportunity for investment. Historic management has left the portfolio in a horrific state with overweight exposure to mining and energy and no stop losses on any positions. Whilst I could not change the past I did look to see what could have been done differently or additionally at the time.

Breaking down the objectives of the fund; SOBIC aims to build a portfolio of diversified value based investments, which are selected through sector based research. Each company is evaluated using a DCF model to highlight the upside potential. SOBIC is run for profit, however it is primarily a learning experience, we are happy to sacrifice professionalism & performance for developing the knowledge and ability of budding analysts.

The portfolio holdings are no longer in the state which I discuss below, we have trimmed and cut undesired positions to give us a relatively clean slate to work with going forward.

Portfolio positions shown below are as of 01/04/2013, (they were not all acquired at the same time), percentages are that of invested portion of the portfolio. There was some cash left over but I disclude it from calculations.

Position Ticker Allocation Book cost
Meggit MGGT 11.82% £399.12
Pennon PNN 17.68% £597.09
Sirius Minerals SXX 23.68% £799.94
Unilever ULVR 17.14% £579
Anglo Asian AAZ 29.68% £1002.49

When I initially went to evaluate VAR in the portfolio I built a very simple Google Sheets model with the common assumptions of normally distributed data and a zero mean. Market returns are not normally distributed in actuality and the mean is not zero, this model was simply used as a proxy tool to get a feeling of where we stood. The VAR model can be made considerably more accurate and realistic by using distributions which actually represent the data rather than just sticking with a normal distribution, a paper such as this one would help out with achieving such an aim. The next time I get round to improving my existing models I will definitely be utilising this paper!

After inputting my data and selecting and naming a few ranges I end up with this:

Var Proxy

Looking at the VAR (One day horizon, zero mean & normality assumption):

Value At Risk (Unhedged): σ
Invested: £3,377.91 £ £160.93 £565.97 £945.47
Positions: 5 % 4.76% 16.76% 27.99%

Looking at the portfolio VaR we can see our 1,2 and 3 σ moves explicitly defined in cash and as a percentage of the investment, the 3σ result is quite worrying, a potential loss of over £900! This reads as, the portfolio in its current state has the potential to make or lose £160.93 (4.76%) in one day under the assumptions of the model (normally distributed daily returns & a zero mean) with a 1σ move (occuring 68% of the time over a 1 day period).

Now for the fun part: choosing our risk!

When I evaluated the activities of SOBIC, it was clear that a solid portion of the analyst efforts went into reading company reports and understanding where the business model fell short or was stronger than its competitors. They attempt to explicitly define the upside potential of the company with a DCF model, which gives a rough gauge of opportunity for the Fund. SOBICs emphasis was on the company, not on the wider market. Consideration was given to the sector in some cases, but not to the financial health of the UK or any major macro news. SOBIC had no holistic view on the wider picture, simply the bottom up micro picture (I have since set up a macro team to address this issue!). My thinking is that if SOBIC were to ‘choose’ its risk exposure they would have greater consistency of performance, regardless of prevailing market conditions. I suggested exposure should be limited to factors they considered and had a view on.

What are the risks here?

I would argue our risks were / are:

  • Market risk
  • Sector risk
  • Company risk

I know there will be a few of you reading this quietly tutting at my lack of inclusion of factors such as systemic risk, liquidity risk, currency risk and so forth. However in this case I was focussing on the most appropriate and tangible risks to SOBIC.

I am broadly defining market risk as all the macro based noise which leads to large proportions of the major indices, currency pairs and key commodities moves, these are understandably very influential on stock positions in a portfolio. Simply by having an open position, long or short, in the market you are exposed to these factors. Sector risk, as I am sure you can guess, are the risks arising from exposure to a sector, such as mining. Finally, company risks are the idiosyncratic factors arising from holding shares of a company which is free floating in the market.

To quickly visualise this I have plotted an example of Coca Cola (Red) against Pepsi (Blue) against the S&P 500 (the shaded region). On the left we see company risk as the performance difference of Pepsi vs Coca Cola; whilst the explicit figure is irrelevant what I am highlighting is a differing in share price over time, despite them both being a drinks manufacturer in the same sector. In the middle we have sector risk; I am showing this as the difference between both Coca Cola and Pepsi (roughly representing the sector!) and the market (S&P 500). Finally on the right I show market risk as the absolute move of the market over the most recent month. It is clear to see that both the blue and red lines typically trend in the same direction, this makes sense as both are constituents of the market!


A quick thought on the chart above; I am quite aware that one could class the market risk (as i’m visually displaying above) as the ‘roll up’ of company risk which collectively rolls up to form sector risk, which then rolls up to form the market risk. However I argue that moves in global indices from a very top down macro perspective are heavily driven by institutional plays where ‘on / off risk’ moves in global portfolios drive market returns considerably. Said differently, the market moves not always on the merit of a single sector, a group of sectors or an individual company’s performance, but simply because risk is picking up and the manager needs to derisk by selling off a percentage of the portfolio holdings.

As previously mentioned, SOBIC doesn’t formally form a consensus view on the market or any factors which may affect returns over the next year. In some cases the sector prospects were also ignored, very much in line with the top down view not being formulated. My solution to this was setting up a ‘macro team’ and suggesting SOBIC should choose to expose themselves only to company based risks and some sector risks where desired. By minimising market based risks they would improve their performance over a market which was underperforming and limit their exposure to factors they were not considering. By choosing when to take on sector risk they could practically eliminate the sectors performance as a whole when desired, or they could simply leave sector exposure to further benefit from any growth in the wider sector which they have a view on.

Giving some examples to illustrate:

Position(s) Market risk Sector risk Company risk
Anglo Asian (Long) Yes Yes Yes
Anglo Asian (Long)

Apple (Short)

No* Yes Yes
Anglo Asian (Long)

Anglo American (Short)

No* No* Yes
Cash No No No

* No is asterisked as there will always be some inherent residual risk but it has been greatly minimised, effectiveness of risk minimisation is determined by the opposing asset held and its ability to hedge the target position

Taking this into consideration I decide to pick a rival to short from the same sector for each of the companies in the portfolio. I am choosing to limit my risk simply to the companies, trying to mitigate any sector or wider market based exposure. Whilst my choices of companies may unintentionally contain biases and look ahead issues (retrospective construction flaw) what I am showing here still holds in concept whilst the explicit figures will be flawed by design.

In setting up the theoretical portfolio, just to reiterate, whilst the stocks I have picked are ‘relevant’ they are not the focus here.

Long position Short position % Long % Short £ Long £ Short
MGGT RR 7.40% 10.55% £250 £356
PNN SVT 8.59% 12.51% £290 £423
SXX GEM 6.01% 16.44% £203 £555
UNLV AEP 8.33% 13.89% £281 £469
AAZ AAL 8.59% 7.69% £290 £260

To get the % allocations I built a simple beta based hedge tool where using the same investment total to reallocate amongst all the positions with a holistic view of the various betas.


Some understanding of what I have inadvertently done here:

  • I have rebalanced the portfolio to a more beta neutral stance, high beta stocks which were overweight are now relatively underweight
  • There is more short % allocation than long, again this is a product of beta strengths rather than market opinion (we are not short on the market – we are effectively beta neutral)
  • I may or may not have paired a portfolio stock with a winning or losing trade, however this is not what we are evaluating
  • The short positions are put on when the corresponding target long position was historically placed
  • We do not consider transaction costs, margin funding or any costs for that matter, simply just daily stock price

Looking again at the tool after inputting some new data and naming ranges:


Turning to the VaR figures:

Value At Risk (Hedged): σ
Invested: £3,377.91 £ £77.35 £272.23 £454.44
Positions: 10 % 2.29% 8.06% 13.46%

We have seen a huge improvement here with the explicit risk defined earlier reduced by at least half in this calculation!

Reviewing the act of hedging the portfolio, we have:

  • Chosen to expose our portfolio to company risk only, all but eliminating market and sector exposure
  • Increased our costs by doubling the number of positions and corresponding trades needed
  • Decreased our portfolio risk, as explicitly defined by the VaR tool with more than half the percentage risk being reduced across 1, 2 and 3 sigma events

Of course I couldn’t finish without showing the theoretical portfolio performance alongside the actual portfolio in SOBIC (not including any transactions fees or alterations to positions over the duration i.e. totally unmanaged – as it clearly was!)


Despite the poor performance in both portfolios, we have clearly reduced the volatility as shown by the flat yellow line vs the blue line. Here there was no alteration to the companies we actually invested in other than weighting them differently and shorting a rival. I shall caveat once more that the direction of the yellow and blue line is not the focus of this analysis, but rather the volatility of the holdings and consistency of returns.

Reflecting on the concept of VaR and ‘choosing risk’:

VaR is called out on occasions such as in the Swiss Franc unpegging case, as it did not give traders explicit maximum risk parameters. The move in the Franc was about 180 standard deviations – which should occur every billion years or so, whilst very unlikely still theoretically possible!  This brings me to the core of the issue: VaR does not define maximum loss. It was never supposed to do this, and could not be blamed for not alerting traders to the absolute risk surrounding the Franc peg. VaR is simply a tool, and although not perfect it serves purpose. It is not designed to explicitly define risks associated with the ‘once in a billion year’ occurrences – one would have to design a customised model dealing with the specific thematic risks and defining all possible cases to do this.

When it comes to the concept of choosing your risk you should look to focus the desired exposure toward factors you have a view on. The downside to channeling exposure is that if you are wrong in your decisions then there will be no other factors to ‘carry’ the investment (this would be bad management from a discipline standpoint anyhow!). So my suggestion to those out there who may or may not have suffered losses over the recent market moves: don’t avoid all risks, just try to choose them to suit your portfolio better.

0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *