Always appreciate your articles. Been recently reading Billy Williams’ book, “Trading Chaos”, 1995. Point of book, can’t use linear models to interrupt non-linear systems. The use of fractals are a better approach.

Quote from book, pg. 134:

“The early 1980s produced the $3,000 black box systems, RSI, stochastics, sentiment indexes, and so on. Then Trade Station and other program developers made back testing and curve fitting a fun project for new traders. Mechanical systems were the rage — popular but not profitable.”

“Parametric statistics are not appropriate to examine nonlinear behavior.”

Whether or not one accepts Mr. Williams ideas is another matter but, it does provide food for thought.

Red

I was curious though if your back-testing software allows you to optimize the buy/sell threshold levels of 40 and 60?
Thank you again!

That’s a good idea. I’ll put that down as an idea for a future idea for another article. Until then remember, the code is available to download and test as you please. 🙂

Thanks for the great comments, Andy. A couple of points. While the strategy may not work well from 1966 to 1982, that is not much concern for me since I can only trade today. So I’m much more concerned about “lately”. While it’s true the longer the history, the more confidence I have, today is what matters. Next, I would not call the strategy presented a trading model. It’s more of an indicator to help gauge the overall market direction. Anyway, thanks again for the comments about your trading style and levels.

However, if you use $INDU with this strategy going back 119 years, you’ll see there are very long periods of time that this is not useful AT ALL. The 1966 to 1982 period is particularly awful! The fundamental flaw in so many strategies is the age old, and incredibly outmoded, “lookback” period.

All my work is based on some new fundamentals:

* No Tradestation function calls. They are cost is prohibitive.
* It’s about THE LEVELS. There is no idea more important than this. Buy near support in uptrends and cycles. Sell near resistance in downtrends and cycles. Unless you use the “traditional” tools to discern support and resistance accurately, your strategies are destined to fail under different circumstances. A strategy that eventually fails or goes flat as the circumstances change is a flawed strategy (i.e. 1966 to 1982 in the above example). You may not have lost much money during that period, but would you be able to SIT through it the whole time? Of course not.
* Minimal, if any, “lookback” periods. I use cumulative and “break” processing. Moving average “lengths” are accumulated, and reset at appropriate times. As price “breaks”, levels are trailed up or down with a possible reset of the MAs (and I use VERY few MAs, as they are LAGGARDS! – IT’S ABOUT THE LEVELS!!!! ).

System Trading is like magic. The old hands will only show you enough to get you started, and it’s all OLD. You’ve GOT to break new ground, because they aren’t going to show you stuff the works CURRENTLY! Why would they share that? It’s proprietary! And you better believe they are willing to SELL it to you! 🙂

Andy Shaffer

Thanks for the idea. I update the article’s last paragraph (“Where Are We Now?”) whenever the signal changes. As it stands today, the last update was July of 2016.

Jeff, thanks and excellent link which addresses what I had said.

Salvo, you’ll be interested in reading http://systemtradersuccess.com/the-top-three-pitfalls-of-stock-and-etf-system-development/