October 22, 2013

Everyone seems to agree with the notion and existence of a "trend", but no one seems to agree on its definition. Some want a universal definition with no compromise, like in, this is "the" trend, period. Geez, it's evident, see, it starts here and stops there; it can plainly be seen by anyone of age on any chart of past stock price data, whatever the selected time frame.

They want a one-size-fits-all definition that a trend spans a microsecond, a week, a month, 100 years, anything in between, or longer. Not only that, but this "trend" should apply to all classes of tradable assets at the same time in the same direction. Please...

Most, with a little hindsight, will see a trend as it started there. But with its other endpoint undefined since this other endpoint is in the future somewhere. Then the question becomes: will the trend continue or not? Was it a "trend" in the first place? If one looks at population growth, for instance, the answer will have to be yes on both counts and to say the least; the trend has been up for quite some time now... It's another matter for stock prices. Their history is much shorter.

Technically, you should not care what the trend definition of somebody else is. All that is important to you is that the trend definition you use is consistent with your trading methodology. If you swing-trade for weeks to months at a time, why should you care for the minute wiggles here and there since it might just be random noise in your chosen timeframe? The intrinsic value of a 20-year trend on your few weeks' swing trade might amount to pennies, so again, why worry about the value of this long-term drift?

What seems important is your style of trading and your methodology. This methodology "must" be consistent with what the market has to offer or can offer on "your" selected stocks and trading time frame. If your specialty is the 100m dash, you might get the early lead in a marathon, but I don't expect you to reach the finish line!

The simplest definition of a trend, as related to stock prices, has to be: the line between two endpoints. Whether this trend line is flat, going up, or down, it defines the trend: the general behavior of price movements. This trend, in its simplest form, can be defined as ΔP, the difference between the two endpoints P(t) – P(0), your entry and exit prices. You now have a "trend" definition for whatever duration Δt and direction: ΔP = P(t) – P(0).

It's simple: ΔP > 0 → trend is up; ΔP < 0 → trend is down; ΔP = 0 → trend is flat; not only flat, but no profit: P(t) – P(0) = 0, and for sure, no loss either. You need a change in price to make a profit, you need, ΔP <> 0, and your bet should be in the same direction as this price variation ΔP whatever the time span |Q|ΔP > 0. However, note that over the long term (20 years or more) and over a large number of stocks, the average ΔP of US price indexes has tended to be greater than zero with a probability measure approaching asymptotically to one. This means that with time (long-term), stock prices, on average, tended to be up, ΔP > 0. With this notion, it would appear that just holding your positions for the long term would be sufficient to finish ahead: Σ(H.*ΔP) > 0.

You can't make any money on past trends; they are gone. You can only bet on their future continuation or reversal. This translates to P(0), whether you take the trade or not. You can rely on your experience and expertise, on all the information you can gather up to your entry decision point: P(0), and there make your bet. But you are faced with uncertainty, P(0) is the now price, it is always available, you just press execute on your keyboard, or let your program do it, and the purchase or sale is done in a split second.

The Point of Interest

Your real endpoint of interest is P(t). What will it be at some time Δt in the future? You could simply gamble and see what happens, and there is nothing wrong with that. I have back-tested some trading strategies that do just that and win. That, like many other trading strategies, might not be "your" cup of tea or the best way to play the game. There are so many different and profitable trading strategies out there. A lot of discretionary traders do make money in the markets, and some do not even use any trading software.

Nonetheless, you want to play the game and hopefully win. Therefore, whatever your methodology, you will need to backtest to show that at least over past data, your own trading strategy, your own style of trading, not only survived but thrived sufficiently to, as a bare minimum, beat the long-term Buy & Hold scenario. And there, when I say backtesting, I do mean making a good job of it.

Not backtesting your trading methods is like a shot in the dark. What makes you believe that your trading idea has value? What could make you believe that your idea is new and has not been tried or tested before? Maybe an alternative might be to use already developed systems. But even there...

Say you adopt the idea, the concept developed by someone else who "says" that his trading strategy outperforms the Buy & Hold long term. Say something like the Turtles; however, statistical proof is not provided, not even a portfolio simulation test, only a few chosen profitable examples. Why would you accept this without testing the trading strategy for yourself? You say you don't trust anyone in the market, and yet you would seem ready to fork out some $3,000 on a Turtles course without system verification! And then wonder why you are losing money.

I do think that the only way to see if an outside trading strategy has merit is to test it yourself. This way, you can get a better understanding of the trading procedures used, their objectives, their strengths and weaknesses, and even maybe gain the ability to modify them to your own needs. As a minimum, I would want to see backtests where the trading strategy generated thousands of trades over a long-term trading interval over many stocks.

I have often taken ordinary and even non-performing trading scripts, modified them to my needs, and produced what I would call outstanding trading scripts by adding my own trading procedures and software routines that backtested real fine. But you need to backtest properly; otherwise, you might be just curve fitting. See a related article on trends: Trends or No Trends.

If you don't do your testing properly, you start an over-fitting iteration process that will extract more and more profits from your past data. But the future is not over-fitted, it is there to unfold without regards to your particular trading strategy. You have to work hard to design long-term trading strategies. At each stage of development and as portfolio size grows, you will find new challenges and problems to solve, which, if not incorporated in your trading strategy design from the start, will deliberately and inexorably restructure your portfolio to the downside. The market is not designed to be kind to anyone. It is just a marketplace where you can exchange cash for shares or vice versa at any time during market hours.

Designing a profitable trading strategy must comply with what is out there. It must comply with all the academic research and trading history. You must build on the shoulders of the best. Your trading strategies need to comply with Markowitz, Sharpe, Jensen, Fama, and the list goes on and on. When I see academics describing their unique view of the market using a Sharpe ratio for comparison purposes, I view their research and mathematical models holding up as long as they describe their trading universe within their Gaussian environments.

But the big question is: is the market really Gaussian? To that, I answer no, with all its implications. I prefer to look at the market as a chaotic universe where a Gaussian distribution is at best a low approximation process and where randomness is stronger than admitted in most of those research papers.


Created... October 22, 2013,    © Guy R. Fleury. All rights reserved.