Long/Short market neutral exits
Posted: Thu May 20, 2004 11:46 am
Hello all – I have an issue
My partner and I are in the process of creating a long/short market neutral stock strategy for a hedge fund. He has done some very original research, and what he has created seems to work well as a stock selection filter for medium term (4-15 week) time periods.
We’re running into a few problems with creating the basic stop regime to be used with this strategy.
Market Neutral stock strategies are typically Beta, Sector, and dollar neutral, and ours is heading in this direction. Right now, we are in the initial exploration phase of research, and so the portfolios we’ve created are roughly beta neutral (pretty close) and dollar neutral. We plan on testing sector neutrality soon. I expect it will show good results.
We don’t really have a good exit strategy. Yes, I know – its all about the exits, and that’s why I’ve decided to ask for some help.
We currently are looking at several ideas for initial, risk control exits
1. A typical turtle style, where we use a volatility measure to extract an expected price movement. If the stock moves against us by this amount, we exit the individual stock. This is a simple method, but leaves the portfolio unbalanced. We then balance the port using ETFs or futures.
You can see the massive problem with this in a market neutral portfolio context. If the market has a large move in one direction, all of the stocks will get stopped out for a loss. We are attempting to profit from the relative movement of the individual stocks in the portfolio, so getting stopped out really takes away ½ of the potential return. That suggests not having a stop at all and just letting the portfolio go. However, as a futures trader, I can’t stand the thought of being without some protective stop!! It’s really bothering me, so I came up with…
2. Create an ‘index’ out of the stocks in the portfolio. Base the stops in relation to the movement of the index. For example if the index is going down overall, and one of the long stocks in the portfolio is going down much more relative to the index, we exit this stock, and replace with blah, blah…
This is a little better, as it fits better into the objective of the system – to capture the relative price movement of stocks. Stocks that are not performing relatively well will be eliminated. To actually implement this regime, we could just calculate the volatility of the index. The index will move some ‘x’ volatility units in a day in one direction – adjust all individual stock stops by ‘x’ volatility units in the same direction. So each individual stocks stop will move a different price amount, but will move the same amount in volatility.
This approach has several problems. One, the individual stocks that make up the index are in the portfolio too. As a result, their movements will impact the index. A move in several of the stocks will greatly impact the index, and so there might be an occasion where the stops just keep moving, and we’re losing tons of money. It’s unlikely on any individual run, but we all know that unlikely and impossible are very different. Over a 10 year run, we can expect it to happen. The alternative of creating an index where the individual stock is excluded is massively complex, and would extremely difficult to execute without errors.
Another problem is how to measure risk for this method. As the overall movement is aggregated, the risk would have to be aggregated as well. This leaves us in a situation where were are risking a high percentage of equity on what is essentially one giant trade. If we are trading several sectors, it could turn out where we lose like 50% of equity in a matter of weeks after several good years of performance – a typical short option strategy style payoff.
Using an outside index, for example an ETF for a sector solves some problems, but also introduces others. What is the correlation (and all of the problems with that measure) of the portfolio with the ETF? It could be less than ideal – like in the 70% range. In other words just good enough to semi-trust, just bad enough to get burned.
So – does anyone have experience, ideas or comments on this problem? Is there books, articles or anything on this? Also, note that I didn’t even get into trailing stops. I need help – and know it.
I would think there are several common approaches to this problem, I am just not familiar with those approaches.
Thanks, and I look forward to this discussion.
My partner and I are in the process of creating a long/short market neutral stock strategy for a hedge fund. He has done some very original research, and what he has created seems to work well as a stock selection filter for medium term (4-15 week) time periods.
We’re running into a few problems with creating the basic stop regime to be used with this strategy.
Market Neutral stock strategies are typically Beta, Sector, and dollar neutral, and ours is heading in this direction. Right now, we are in the initial exploration phase of research, and so the portfolios we’ve created are roughly beta neutral (pretty close) and dollar neutral. We plan on testing sector neutrality soon. I expect it will show good results.
We don’t really have a good exit strategy. Yes, I know – its all about the exits, and that’s why I’ve decided to ask for some help.
We currently are looking at several ideas for initial, risk control exits
1. A typical turtle style, where we use a volatility measure to extract an expected price movement. If the stock moves against us by this amount, we exit the individual stock. This is a simple method, but leaves the portfolio unbalanced. We then balance the port using ETFs or futures.
You can see the massive problem with this in a market neutral portfolio context. If the market has a large move in one direction, all of the stocks will get stopped out for a loss. We are attempting to profit from the relative movement of the individual stocks in the portfolio, so getting stopped out really takes away ½ of the potential return. That suggests not having a stop at all and just letting the portfolio go. However, as a futures trader, I can’t stand the thought of being without some protective stop!! It’s really bothering me, so I came up with…
2. Create an ‘index’ out of the stocks in the portfolio. Base the stops in relation to the movement of the index. For example if the index is going down overall, and one of the long stocks in the portfolio is going down much more relative to the index, we exit this stock, and replace with blah, blah…
This is a little better, as it fits better into the objective of the system – to capture the relative price movement of stocks. Stocks that are not performing relatively well will be eliminated. To actually implement this regime, we could just calculate the volatility of the index. The index will move some ‘x’ volatility units in a day in one direction – adjust all individual stock stops by ‘x’ volatility units in the same direction. So each individual stocks stop will move a different price amount, but will move the same amount in volatility.
This approach has several problems. One, the individual stocks that make up the index are in the portfolio too. As a result, their movements will impact the index. A move in several of the stocks will greatly impact the index, and so there might be an occasion where the stops just keep moving, and we’re losing tons of money. It’s unlikely on any individual run, but we all know that unlikely and impossible are very different. Over a 10 year run, we can expect it to happen. The alternative of creating an index where the individual stock is excluded is massively complex, and would extremely difficult to execute without errors.
Another problem is how to measure risk for this method. As the overall movement is aggregated, the risk would have to be aggregated as well. This leaves us in a situation where were are risking a high percentage of equity on what is essentially one giant trade. If we are trading several sectors, it could turn out where we lose like 50% of equity in a matter of weeks after several good years of performance – a typical short option strategy style payoff.
Using an outside index, for example an ETF for a sector solves some problems, but also introduces others. What is the correlation (and all of the problems with that measure) of the portfolio with the ETF? It could be less than ideal – like in the 70% range. In other words just good enough to semi-trust, just bad enough to get burned.
So – does anyone have experience, ideas or comments on this problem? Is there books, articles or anything on this? Also, note that I didn’t even get into trailing stops. I need help – and know it.
I would think there are several common approaches to this problem, I am just not familiar with those approaches.
Thanks, and I look forward to this discussion.