Improving Trend Following Systems

trend following systems

trend following systemsThe second half of Clenow’s chapter that defines two trend following strategies focuses on improving those strategies. Once again, I was surprised by the simplicity of the improvements he suggests. Ironically, they are basically the same improvements I often wonder about when looking at other people’s backtesting results. Reading this chapter led me to believe that many of my theories will actually produce positive results.

Here are my thoughts on the four topics Clenow covers with respect to improving trend following systems:

Trend Filter

If we set a criterion for our strategy that we are only allowed to buy when the overall market is in an uptrend and vice versa, we are likely to get two effects: the number of trades is reduced and there is a higher percentage of winners. – Clenow

I have been looking at a number of different trading systems over the past few months. Many times, I have found that the biggest problem is getting whipsawed out of positions that are taken against the long term trend of the market. Obviously, some systems are built to trade counter to the long term trend, but most of the systems I have considered have performed best when they are aligned with a major trend.

In order to determine the long term trend direction, I have been using the 200 day simple moving average (SMA). Clenow suggests using the 100 day SMA. I suspect that arguments can be made for either time frame depending on your trading outlook, but I would always defer to his experience over my assumptions. Regardless, the idea is that using an SMA line is a super simple, and reliable, method for determining trend direction.

This will make sure that your system is only able to trade on the long side while the market is trending up. Likewise, it will only trade the downside when the market is trending down. Many of the Market Wizards have discussed the importance of trading with the major trend of the market. Trading is hard enough, there is no need to do it swimming upstream.

Volatility Stops

Let’s keep it simple, and set our stops trailing, always 3 ATR units away from the best point the position has seen since we opened it. – Clenow

One of the biggest risk factors in some of the systems we look at is the fact that there is not definite exit point. A traditional breakout system could theoretically establish a long position and then trend downward steadily for years without ever triggering an exit signal. Clenow recognizes this open-ended downside and proposes handling it by using a trailing stop-loss order.

Clenow suggests trading a diverse cross section of markets, each of which has a different average volatility. He negates this issue by accounting for volatility with his position sizing algorithm, and then again with his trailing stops, which are based on average true range (ATR). This gives a position enough room to breath, based on its historical volatility, but still sets a maximum loss point.

Many of the Market Wizards have pointed out that properly assessing and accounting for risk is the most important thing a trader can do, regardless of strategy or approach. Using ATR stops to control systems with open-ended downsides is an excellent way to better manage risk.

Adjusting Risk/Reward

trend following systemsIf you want to gear the risk level of the strategy up or down, all you need to do is to modify this input value. The risk factor governs how much each position, in theory, should affect the bottom line of the portfolio on a day to day basis and how much damage or profit the position can potentially do to the portfolio as a whole. – Clenow

One of the most interesting aspects about trend following systems is that they all produce similar results, but the magnitude of those results often differs. This is because of the amount of leverage a trader allows the system to use.

Traders who fully believe in their systems and are trading for maximum return often kick up the leverage in order to get the highest possible returns. Traders who are managing money for conservative clients are often forced to limit drawdowns, which in turn lowers the returns they are able to produce.

Clenow adjusts the risk/reward of his system by a variable he calls “risk factor,” which is built into his position sizing algorithm. He proposes a fairly conservative system, but empowers his readers to adjust their risk factors up or down based on their own personal risk tolerance.

I have seen it written that you should take whatever number you believe to be the maximum drawdown you are willing to withstand and cut that number in half. You can then use that number to calculate your risk exposure. Never underestimate the importance of being able to sleep at night.

Curve Fitting

Optimisations are plain and simple evil and out to kill you, and if you ever catch one stomp on it hard and make sure it does not get away. – Clenow

Clenow closes his best chapter yet by warning against over-optimizing systems. He cautions that many software packages will calculate the ideal variables for a moving average crossover or breakout system based on historical results. Many times, systems that are that optimized will fail because they lose their robustness. Clenow suggests instead that you test three variations of a system and see which on best suits your personality.

As I have started using NinjaTrader, I can see how easy it would be to implement one indicator after another in order to “improve” a system. In order to guard against that, I have been constantly reminding myself to keep it simple and stick to using only the most basic indicators.

The goal is to build a system that will perform well regardless of what the market does in the future. By building a system that would have performed too well in the past, we are handicapping its flexibility to adjust to whatever the future markets might do.