Monday, January 7, 2013

Thinking about trading methods

The dirty little secret of the markets is this: Most trading is based on faith, not in a higher power but in the received wisdom of the great analysts who came before.

Do we trade on the fundamentals? We follow the wise men Benjamin Graham and Warren Buffett.  Or maybe even the wise guy Jim Cramer.

The relative strength index, the most widely used indicator in the world? J. Welles Wilder is the source of wisdom.

Elliott Waves? We take Robert Prechter as our guide.

Perhaps we prefer the Turtle Trading method, which is the basis of the way I trade? We know it works because Richard Dennis and his disciples told us so.

Whatever we do as traders, we do it because we've been told that it works. Which seems like a sorry state of affairs for women and men seeking to make their fortunes rationally in the maelstrom that is the markets.

The emphasis is on "rational". It is, in my opinion, irrational to make trading decisions based on faith. Sola fide is a very poor principle for the purpose of making money. Faith in trading, without the works to to back it up, is a fast track to failure.

The methods of those wise men (and they are all men for some reason) are enshrined in brokerage and charting software used around the world.

We are even reluctant to change the standard time periods used in the analysis. Wilder's RSI and other indicators tend to use 14 days as the period of analysis? What's so special about 1-4/5 weeks? We don't care. It's the default, so we use it.

When Wilder developed the RSI in the 1970s, 14 days was a period that worked for him. But 1978 was long ago and far away. The markets of 2013 are a different sort of beast, with shorter holding periods than anyone dreamed of 35 years ago.

The same can be said for all of the popular trading schemes. They've been around a long time. They are hoary with age.

Moreover, when I read about the popular methods, the testing tells me how they did against the market as a whole, but not against any individual stock that I might be interested in. Does anyone for a moment believe that the price of AAPL behaves the same way as that of SO? Or of FB? Or of APLL?

Private traders (what some call "retail traders") have up to now had little choice but to trade on faith. It takes data to do custom analysis. Data cost a lot of money.

Happily, the price of data has come down in recent years and big data can easily be incorporated in a trader's information base without adding much to the overhead of trading.

I'm working with a database of daily stock prices from the NASDAQ, NYSE and AMEX exchanges -- Open, high, low, close, volume -- dating back to the start of 2009, about three months before the market's recession low. As of today, I have 7,460,837 records to work with. Since each record has five pieces of information, that means I'm working with 37,304,185 data points.

I think that's enough data to draw some interesting conclusions from, but I'll need to work with it to see for sure. Time will tell.

(Intraday data is also available, but it is more costly, and I'm always mindful of the overhead in my trading operations. I considered it, but in the end decided it was unnecessary to support the way I rade.)

During the past four weeks I've written software (using the coding language perl as my platform) to apply a version of my Turtle Trading rules to the data, identifying 213,958 breakouts beyond the 20-day price channel.

The most obvious thing to do with the data is to follow the sports and elections quant Nate Silver by calculating the odds of success. From my data, I know the number of  breakouts, and I know how many of those were profitable. I also know the average percent profit on the profitable trades. And I can analyze this further by breakouts to the upside and to the downside.

And there is much, much more.

It is often said that in the markets, past performance doesn't determine future returns. But it is also said that prices tend to revert to the mean.

These are contradictory statements. For what is the mean if not the record of past prices? What is reversion to the mean if not past prices influencing future returns?

That is the underlying assumption of my way of analysis (and, in fact, of all technical analysis).

My version of the Turtle Trading rules can be found here. I've made some changes from those in order to analyze my daily data.

One change is in the time frame. Under the rules I have been using, when a breakout occurs intra-day, I wait 30 minutes, and if the price is still beyond the breakout level, I enter the trade. With the daily data, I'll be analyzing breakouts from the prior trading day, and entering the next trading day if the breakout signal remains valid.

Second, I won't be using the RSI or any other auxiliary indicator as confirmation. The analysis is pure Turtle, except for the time delay.

Third, for this analysis I've chosen to use the classic Turtle measure of success or failure: The price moving beyond the 10-day channel in a direction opposite the trade. For my operations up to now, I've modified that by using a trailing stop/loss equal to twice the average daily trading range. And in practice, I may still do that. But for analytical purposes -- for now at least -- I'm calculating success and failure based on the 10-day channel rules.

There were 15 breakouts of highly liquid stocks, all to the upside, on Friday, Jan. 4. I'll be looking at those today, and picking the best to analyze, bringing my new data capabilities to the table in making a trading decision. Look for a posting to come.

Disclaimer
Tim Bovee, Private Trader tracks the analysis and trades of a private trader for his own accounts. Nothing in this blog constitutes a recommendation to buy or sell stocks, options or any other financial instrument. The only purpose of this blog is to provide education and entertainment.
No trader is ever 100 percent successful in his or her trades. Trading in the stock and option markets is risky and uncertain. Each trader must make trading decision decisions for his or her own account, and take responsibility for the consequences.

No comments:

Post a Comment