I’ve decided to have a more detailed look at my use of stop losses. By using a stop loss I mean selling a share if its price falls to a predefined level. Stop losses fit naturally within a momentum based strategy, as they discipline you to sell shares that lose momentum.
My current use of stop losses is a bit ad hoc and I’ve been wondering whether and how to be more systematic. In particular, I’m concerned that I might be applying them a bit too vigorously. There have been a few occasions recently when I’ve sold a share on price weakness only to buy back again a couple of weeks later when the price has bounced.
To clarify, when I use stop losses I monitor and execute them myself, rather than leaving them as an automatic order with my broker. This is a matter of personal preference, but my experience is that delegating stop losses to your broker can be risky, unreliable and can sometimes lead to you being exploited by short-lived spikes in volatility. Since I monitor my investments regularly anyway I have no need for automatic stop losses.
To help define a clearer stop loss strategy, I’ve asked myself a series of questions.
Should I use stop losses at all?
The main alternatives to stop losses I can see are: only selling in response to news (or when valuation gets too high) & reviewing holdings after a certain time period (say monthly) rather than using stop losses on an ongoing basis.
Buying and holding and only selling ‘if the story changes’ has some attractions. It’s very easy to implement and takes a lot of the anxiety out of deciding when to sell. However, I’m pretty sure that using stop losses would improve performance beyond this approach. From my experience it is very often the case that a price fall will come in advance of bad news (as at least some investors predict an increased likelihood of bad news occurring). It is also frequently the case that prices will fall and then stagnate for an extended period with no change in news at all. Using stop losses to avoid these situations and more fully exploit momentum is in my view almost certainly likely to be worth it.
I’m more on the fence as to whether using stop losses is better than simply reviewing holdings every month, and deciding at that point whether to sell because of a loss momentum (also taking other factors into account). My mechanical benchmark portfolios follow this approach and the more concentrated version is currently outperforming my actual portfolio. This suggests this alternative can’t be too bad.
I think the main advantage is that it would make it easier to make decisions more carefully and dispassionately, ensuring that changes in momentum had time to play out before a decision to sell was made. It could definitely help discipline me from selling too quickly if indeed that’s the mistake I’m making.
The flip side would be that the delay in response to falling momentum would sometimes cost me. In this way this strategy seems suboptimal. I prefer the idea of a stop loss strategy that I can refine over time, even if my current execution of it is imperfect. I think the key thing though is to see how my current use of stop losses is working out.
How much have stop losses been costing me?
A few months ago, I did a simple analysis to get an idea of whether I might be overtrading as a result of being too quick to sell. This compared the performance of shares I had sold after I sold them against the return of the rest of my portfolio since the sale. At that point I found that only four out of the 29 shares I had sold over the latter half of 2016 went on to perform within 5% of the rest of my portfolio or better. The 5% is to account for the trading costs of selling and then reinvesting. This suggested that my selling decisions had been working out pretty well at that point.
I’ve redone this analysis for the first half of 2017 until the end of July. Out of 36 sales in first half of year, only seven have gone on to perform within 5% of the rest of my portfolio or better (and nine within 10%). The rest have done quite a bit worse. This is a surprisingly positive result and suggests my sell decisions haven’t been too bad.
Looking in a bit more detail, a couple of points jump out. One is that the performance of the rest of my portfolio has been pretty strong over the period. This has meant that, while several of my share sales went on to rise in price after the sale, generally this was by considerably less than the rest of the portfolio. As many of the shares in my portfolio have done well, the ‘opportunity cost’ of holding an underperforming share rather than selling and reinvesting into the rest of portfolio has been high. As a result, what has looked like a mistake, as the price has bounced shortly after a sale, often hasn’t actually been one. While there has not always been a further slide of the share price following my sales, there has often been a period of volatility or consolidation.
The other point is that it is pretty difficult to tell from the data what my stop loss approach has been. There is a large variation in how much I have let a share price fall from its high before selling. In general I have been pretty tight and have only allowed shares to fall on average 8% below my buy price before selling, but in individual circumstances this has varied widely. This accords with my recollection of the thinking behind these trades: how far they have fallen from the high has typically only been one of several factors I’ve accounted for when deciding when to sell.
This lack of consistency means that the further piece of analysis I had planned to do, looking at how a different stop loss level might have affected my results, seems a bit meaningless now. I can say that extending the stop losses across the board to a consistent higher level (e.g. 15%) wouldn’t have helped overall. It would have helped avoid a small number of the unsuccessful sales, but would have also made me continue to hold a far greater number of losers for longer. On the other hand, the rare occasions where I have allowed a share to fall by more than 15% (Boohoo in particular) appear in retrospect to have clearly been mistakes.
Selling a share soon after purchase requires admitting the original purchase was a mistake. Consequently, my frequent sales of recently purchased shares, and on occasion subsequent repurchases of these shares at higher prices, have often left me feeling a bit stupid, especially since I’ve committed myself to record the trades in my blog. I imagine I would feel this even more if I was a professional fund manager. However, overall I’d say my fairly tight use of stop losses seems to actually be working quite well. I expect it’s the case that the best approaches in investing are often going to be the ones that make you feel the most stupid – after all, it should pay to be contrarian.
To refine my stop loss strategy further and ideally make it more consistent (or at least clearer to myself), I’m going to need to look a bit more into the qualitative detail of how I apply it.
What form of stop loss works best?
There are quite a few different ways of applying stop losses I can think of:
- A trailing stop loss of a fixed percentage (so the share is sold at a fixed percentage off the high)
- A stop loss based on whether a share falls below its moving average
- A stop loss based on support or resistance levels
- A stop loss based on a break in trend.
I use a combination of all of the above depending on what seems most appropriate at the time from looking at the previous share price behaviour. In principle I’m looking out for signs of a break to the upwards trend in the share price, but allowing a bit of leeway for the volatility you typically get in illiquid shares. In general the ways above can be used more or less interchangeably to achieve this.
The one situation where I’ve noticed it make a significant difference is when a share price shoots way above its trend line and becomes overextended. In these cases letting it fall back to the trend line can mean quite a large fall and, depending on how big this fall is, I think it can make sense to sell a bit sooner using a fixed percentage trailing stop loss e.g. 10% from the high.
Other than that I think I face a trade off between choosing the apparently more appropriate method for a particular situation or always using the same method for consistency. Greater consistency is something that I am after so I’m going to focus on fixed percentage trailing stop losses unless I have good reason not to.
How tight a stop loss?
I’ve had a hunt around for research into stop losses and came across this article reviewing some of the existing research. The research highlights the benefits of a stop loss strategy over buy and hold but also the trade-off between limiting losses to a greater extent with tighter stops but also allowing for less volatility. It points to using stop losses of around 15-20%. It also suggests applying stop losses on a monthly rather than ongoing basis to allow for volatility.
There are some differences between the analysis in the research and how I apply stop losses, the key one being that my stop loss policy involves individual shares rather than a whole portfolio. This results in the additional opportunity cost highlighted above that money from sold shares can be reinvested into the rest of the portfolio. My analysis of my previous sales that takes this opportunity cost into account suggests a tighter stop of loss of about 10% may be a bit better.
The additional benefit of a tighter stop loss strategy is that it will get you out of your whole portfolio quickly and automatically in the event of a market correction. This potential benefit is not captured in the above analysis as we have not had any corrections recently. Getting out quickly when the market starts to crash seems to me to be critically important. Selling up can be a very difficult decision to take if your whole portfolio starts to fall and it can be very easy to be caught like a rabbit in the headlights. It would be easy to underestimate the benefit of a strategy that does this for you automatically.
How much discretion?
Are there benefits to using discretion to give wider stop losses for shares which are more illiquid and volatile, or those in which I have greater conviction? I think so, but there are definitely costs to allowing too much and the rules have to kick in somewhere. Being more rigid with the rules can need to make so many decisions. I think I’m prepared to allow a little bit of discretion but only in more exceptional circumstances and within tightly defined limits.
My current approach to stop losses doesn’t seem too bad. To make it more consistent and explicit I’m going to try to use a 10-12% trailing stop in most cases unless I have a very good reason to deviate.