Saturday, December 31, 2022

Simplicity and robustness - learning from Linda Raschke, Ken Tropin and Curtis Faith


"The minute you think you have found the key to trading, I promise you the markets will change the lock." - Linda Bradford Raschke

It is very easy to think that there is a magic set of parameters which will work best in a chosen market.

The problem is that a certain set of parameters, on a particular timeframe, may work well for a specific period of time, in specific market conditions. But as soon as you dare think "Eureka - this is it!", the markets will change character and those parameters will not work as well.
My own views on this, which I have discussed before, are different to a lot of people, but to summarise:
  • The market 'state' (trending or non-trending, stable or volatile) can change at any time, for any reason. This can be for the markets in general, or in the individual stock, commodity or instrument you are looking to trade;
  • Different traders will interpret what price is doing differently depending on their chosen beliefs, timeframe and parameters;
  • The market you are looking to trade may go through phases where longer-term trends work better than shorter-term trends, or vice versa.
What is the best way around this?

Some will say that backtesting will help you - but if you do that, you are always basing your future trading approach on what has worked best in the past. And it will be sod's law that you will effectively be chasing your tail the whole time. This is effectively what Linda Raschke was referring to.

Long-term readers of this site will know my own views on backtesting.

It is almost a given that, as soon as you ditch a particular market to trade, or shelve a particular timeframe or set of parameters, things will start to 'click' and profitable trends will appear.

If you have the available equity, then you may want to consider adding some diversity, either by adding new markets to trade, or by using more than one system covering different timeframes or parameters.

If that is not an option, then above all you need to avoid very specific, highly tuned methods with lots of rules (degrees of complexity) or tightly defined criteria. 

Trend following approaches are purposefully designed to be simple, but robust.

This is Ken Tropin's take on it:

"In order for a system to be successful, it has to be what I call robust. Robust means that I can test that system in a market I designed it around. Say I’m using it in the treasury bonds, and then if I switch that market and I try that system in the Euro, it still works. And if I change its parameters, it still works. And if I switch it over to corn — something totally different than treasury bonds — it still works. And if I look at some data that was out of sample from what I designed it around, it still works. Then I have something that might be interesting and have a chance of living in the future. 

Because the nature of data is it changes a little all the time. And so the key to success in systems trading is to have what I call a loose fitting suit. I can’t have a suit that’s so tight and perfectly proportioned to me that if I gain two pounds, it won’t fit the data anymore.” 

As Tropin says, trend following systems are, by their very nature, a loose fitting suit in terms of the complexity and number of rules used. The resultant returns are volatile and 'lumpy'. No-one knows when any profitable trends will start or finish, or when the market(s) may change their state.

And finally, here is what former Turtle Curtis Faith had to say in his book Way of the Turtle:

"Simple rules make systems more robust because those rules work in a greater variety of circumstances. complex systems generally are complex because they have been designed to take advantage of some conditions or market behaviour.

Simple rules that are built on more durable concepts will hold up in actual trading better than will complex rules that are tailored to more specific market behaviour. Keep your systems simple and you will find that they hold up better over time."

No comments:

Post a Comment