Your trading model might have a random risk element and you might not even be aware of it. In particular longer term models need special care to avoid ending up with random risk.
The world doesn’t stop spinning when you open a position. The assumptions you used when opening your position may not be valid a week later. Even less so a month later. The world is dynamic and therefore your position sizes must also be.
When you read about ‘money management’, as changing position sizes is often referred to as, it’s mostly about how to double and half your sizes depending on your conviction or current gains. Classic methods involve doubling your position size once you have a certain amount of profits for instance, so called pyramiding up. I’ve always seen this type of thinking as more suitable for gambling than for professional trading.
Here’s a different way of looking at the classic methods of scaling position sizes based on past results. Let’s say you have a method of doubling your position size once you have a certain amount of gains on it. That way of thinking implies that your gains somehow have some predictive power about the future. Why would you want to double your risk just because you had a good run? Your gains or losses doesn’t affect the current market situation.
This is a common logical fallacy though. It’s same when you consider whether or not to exit a position. Say you’re long the Nasdaq and you’re considering whether to keep it or to close it. That decision is exactly the same as if you would have no position and consider whether or not to buy. Exactly the same. If you decide to keep your long position, but you wouldn’t buy on the current levels if you were not already in it, then you’re not acting rationally.
But, back to the topic of this article. This article is not about changing risk. It’s about attempting to keep it relatively constant. This is a much more common way to approach position sizes for professional strategies.
Time randomizes risk
I’ve written many times in the past on how to set position sizes based on volatility. There are many ways of doing that, but they aim at accomplishing more or less the same thing. The idea is to take on an approximate amount of portfolio level risk per position. You look at how much a market tends to move up or down in a normal day and size your position after that. Easy.
Setting an initial risk level of the position is easy. The problem comes soon after. What if the volatility of the position changes over time? You’d have a different risk than you intended. What if you have a large gain on the position? And, as often is forgotten, what if the position in question stays exactly the same, but your other positions in the portfolio changes?
Perhaps your position does pretty much nothing for a month, but at the same time you make large gains on other positions. Suddenly your portfolio as a whole is larger than when you opened your position. That means that the portfolio level risk of that position is now lower. Organic growth of the portfolio as a whole made the weight of this particular position smaller, even though it itself didn’t change.
As you can see, there can be many reasons why your risk will end up quite different from you had intended. Yes, dear institutional reader, the word I’m getting at is rebalancing.
Long term models require rebalancing
For institutional money managers, the word rebalancing is both common and clear. For retail traders, perhaps not. The idea is that you set your risk levels at one point, and then you recheck at regular intervals and reset the risks as you had initially intended.
This sounds boring, and it often is. It means making a whole bunch of trades that might seem odd and counter intuitive at the time. If you would see my trade blotter, you would wonder why I keep doing all these trades. You might ask why I’m selling in the middle of a massive bull market for instance. Perhaps if you looked at that blotter, you would think that I suddenly went bearish. No, not at all. I was just rebalancing. It had nothing to do with market views or trade signals.
Surely this is just a boring task that only institutional money managers have use for, right? Well.. it can actually greatly enhance your trading results, all while giving you a better control over your risk.
Let’s do a simple demo.
Remember the 12 months momentum model that I wrote about in the past? Let’s use that for the demo. It’s simple, it works, and it’s long term. That’s what we need.
The trading rules are:
- Check trade signals only on Fridays (to reduce whipsaws).
- If yesterday’s price is higher than 250 trading days ago, go long.
- If lower, go short.
- Use a simple ATR based position sizer.
- That’s it.
Now this model has an average holding period of around half a year. A lot can happen in that period. So let’s make two versions of this simple trend model. The first one is just as written above. Just that simple. The second model rebalances positions once a month. That means that once a month, we recalculate the position sizes and adjust current open positions to the size they would have if they would be opened just now. That is, we don’t just let the position sizes run wild and end up all over the place.
In the simulation results below, the two versions are compared. The standard line, in black, uses the same size throughout the position life time, while the purple line uses a monthly rebalance.
Well, it looks like the standard version won. Right? Hm.. perhaps not.
See how the standard version goes nuts at times? It has massive short term profits and ends up giving much of it away. The drawdowns are considerably steeper.
In this simulation, here are the key stats to look at:
- Both versions produced an annualized gain of about 23%.
- The standard version saw a max drawdown of 40%.
- The rebalanced version had a max drawdown of 25%.
I mentioned above that you might find an actual trade blotter a bit confusing. If you didn’t realize that most trades are about risk rebalancing, you could spend days trying to figure out why all those trades are done. Take the S&P for instance. Trend followers have been long this for quite some time now. The model above has been long for several years actually. But it’s been more work than just buying a few years ago and sitting on it. Look at the trade chart below as an example…
None of these trades that you see in the chart above reflects any change in market view. The model is long and has remained so for a long time. But if we don’t do all of these small trades, the actual risk will drift far away from what the model intended.
Yes, reality can be a little bit more messy this way. But wait, it gets worse. The logic shown here still assumes zero fund flows. As an asset manager, and in particular a fund manager, you would see constant flows both in and out. Hopefully more in than out of course. If your fund has monthly dealing, you could expect to see money going both in and out every month. Each time, you’d have to adapt your position sizes to reflect the intended portfolio level risk. If you get a sudden inflow of 3% of the previous fund value, you’d in theory have to increase all positions by that amount. This may not always be possible of course, unless you’ve got a quite high AuM. For some reason, the exchanges don’t like it when you try to buy 0.6 gasoline futures contracts…
How exactly does one go about rebalancing a portfolio?
As I’m sure I’ve mentioned once or twice,
I started publishing a weekly futures analysis, covering about 75 different global markets. In this report, I detail how we approach each market at my asset management company, along with the quantitative analytics that we use. Every week I illustrate each market with a real life trading model to teach the ideas and methods of the industry.
Oh, and once a month I send out a deeper bonus document. This is different every month and contains something of interest to teach the subscribers something new. Last month’s bonus doc detailed two trading models based on a counter trend entry approach. Complete with full source code of course.
Next month’s bonus issue of the Clenow Futures Intelligence Report will focus on position size rebalancing. What can be done, why and how. Sign up for a free month trial and get next month’s issue.
Actually, sign up now and I’ll send you the last report with the counter trading models as well.
Sign up for a free trial here