Dynamic Stop Losses in a Multi-Signal Trading Strategy - Research Article #71
How to set up a moving stop-loss in a multi-signal trend-following strategy?
👋 Hey there, Pedma here! Welcome to the 🔒 exclusive subscriber edition 🔒 of Trading Research Hub’s Newsletter. Each week, I release a new research article with a trading strategy, its code, and much more.
If you’re not a subscriber, here’s what you missed this past month so far:
If you’re not yet a part of our community, subscribe to stay updated with these more of these posts, and to access all our content.
Picture a fisherman planning a new fishing trip. In order to be able to pay his crew, the tools, and to generate some profit, he needs to decide on the strategy that will maximize the returns of his fishing trip. What would be the optimal strategy?
Using a small net in a narrow corner of the sea?
Using a very large net and sail through the largest possible area of the ocean?
I take the same approach to trading strategies. We don’t want to be constrained to a single signal, indicator or length. I don’t think that’s a viable long-term solution. We want to capture as many regimes as possible, and also maximize robustness of our models. That means not being reliant on a single point of failure, which is a problem when using a single signal.
However, this approach comes with a problem for traders that use the old trend-follower methods. As we’ve discussed last week, there’s multiple schools of thought when it comes to trend-following approaches.
Modern trend-followers believe that exposures should be dictated on a continuous spectrum, and regularly rebalanced. This is meant to keep positioning in line with our current view of the market. While old trend-followers, with a binary signal approach, tend to believe that we should enter a full position at a specific price, and only close at the exit or at a stop-loss.
I have been quite interested in the old trend following way of binary signals. My goal is to eventually run a blend of both models on my portfolio, but I think that as an account builder, the absolute return approach has it’s own advantages, if you can stand the volatility.
I see trading strategies as our earlier fishing example. We know that our fish is there (effect), so we want to use the widest net possible (lengths, indicators, etc), so that we can capture it from different angles.
As I was designing the process to test it, I came across the problem of stop-loss setting. How do you set a stop-loss on a position that is shifting across time? That is what we will address today.
By the end of today’s article, we’ve achieved some really interesting results.
Let’s get into it.
Overview of ATR-Based Risk Positioning
Old trend follower philosophy typically uses a volatility based position sizing, by using ATR multiples to set stop-losses and manage risk. Each trade’s size is determined by the distance from the entry point to the stop-loss. This is useful because then we can set a maximum risk based on our equity. For example, if we only want to risk 1% of equity on any given trade, we can measure the distance of the entry to the stop-loss, and calculate the position size we need, so that if the stop-loss is hit, we only lose 1%.
When the ATR is low (or volatility) , this means that the position size will be larger and when the ATR is high , the position size is smaller. This makes sense right? When the volatility is high, we should be taking less risk as prices move faster and sharper.
For example, let’s say we have a $100,000 account, and we want to only risk 1% of equity ($1,000) on a single trade. The ATR at the time of our trade is 2%, and our stop-loss is 5x the ATR. The stop-loss is placed 10% (5 ATR * 2%) away from entry, making our position size $1,000 / 10% = $10,000, ensuring that we only lose those $1,000 if the position is stopped out.
Using stop-losses can be an inefficient way to protect downside, as we’ve discussed in the article below, but I am interested in these type of old models.
Pyramiding: Multiple Entries in the Same Trend
Like I’ve introduced above, rather than a single-entry signal, where you are completely exposed to the underlying assumptions of that signal, classic trend-followers are known to pyramid into a position, as the trend consolidates.
Let’s look at a practical example of a breakout with lengths of 20-days, 50-days and 200-days.
As the trend develops, your position grows stronger and stronger. We’re also making the assumption that as a signal gets stronger, so does the expected return, but I won’t get into it at this time, perhaps for another article.
The opposite is also true. When we are looking to exit a position, instead of exiting a full position at one signal, we scale out of a position as the trend gets weaker and weaker.
In the example above, we can see that the position never fully exited, because price never closed below the higher timeframe length signal (200-day).
Why Pyramid? This method offers a few key benefits:
It limits losses on false breakouts. If the first breakout fails and reverses, the trader only had a partial position on (small risk) and thus loses much less than if they’d entered the full intended position.
Pyramiding maximizes gains on strong trends. By adding more units as the trend continues, the trader fully leverages big moves after they have confirmation the trend is real.
You’re not dependent on a single point failure risk. For example, if market regime changes benefiting short-term timeframes, or long-term timeframes, you can capture them within the same system, rather than being off the market because your current signal didn’t capture it.
Even though the performance decreases when we add a multi-signal approach, as we’ll see below, I do believe that it is the optimal approach if our goal is to maximize long-term robustness of the model. Markets change, regimes change, and believing that your selected length will survive forever, is not a good bet in my opinion.
But we’ll discuss that in a different article.