A Few Rambling Thoughts on Trading

I am going to share on my blog -- my trading journal, if you will -- some thoughts that I wrote to a fellow trading friend earlier today. There are several themes that meander through this email message. The primary one is the subject of achieving an automated system of software-based trading. This is a subject I don't believe I've mentioned on my blog previously. Other subjects include the use of the Weighted Moving Average along with the Exponential Moving Average, trading UP from a shorter to a longer time frame, the dynamic and changing nature of the financial markets, etc.

Here is the text of some of that message:

I may have mentioned in some of my blog posts that I was using some other methods of exiting trades (not Cahen’s). I still look for Cahen’s exits, but I have found that I can time them better than Cahen teaches in his book. The EMA usage is the best one. Right now, I’m also playing with an 11 period Weighted MA. It is slightly smoother than the EMA, and it tends to get me both IN and OUT of a trade slightly earlier. I want to code the one in TS so that it changes color like the one I coded for EMA (which I have attached to this email). I have put the WMA on my charts as a dotted white line (which you might notice in the charts on my blog – they’re barely noticeable). I’ve noticed that at some intermediate point in a trade, the WMA crosses over the EMA. At that point, I begin to ignore the Klinger Volume indicator. That point is somewhat of a turning point, because the trade is no longer vulnerable to possibly losing money. I am thinking about this WMA/EMA in terms of possibly programming it into some code at some point.

By the way, in his Encyclopedia of Technical Analysis Indicators, Robert Colby found that the WMA was almost as successful as the EMA – but not quite. It tends to be slightly smoother, much like the Hull and Guassian MAs. However, he told me in an email that he found the EMA he tested was only effective on daily charts --- not intra-day charts. He indicated that he ONLY tested on the daily charts. He didn’t know if the EMA was successful on intra-day charts. But he went out of his way to mention this fact. However, he also wasn’t using tick charts, as I am. My own experience with the EMA is that they are much more successful with tick charts than short-term time interval charts like your 10 minute one. I think this is because tick charts measure orders through the system rather than by some arbitrary time interval. The more orders that come through the exchange, the faster the tick chart candles print across the screen. While within 5 minutes, much can happen on a 5 minute chart if a shock hits the system, on a tick chart, as soon as the shock hits, the candles start to print across the screen very rapidly. And that affects the Bollinger Band patterns very quickly that Cahen teaches. In an automated system, entries and exits could happen very quickly and dynamically.

I have also found that on a strong move in the markets, prices may exhaust themselves on a shorter (tick charts, 3 min) time frame temporarily, but as time proceeds and prices get close to the EMA on the higher time frame (10-15 min), they tend to rebound off the EMA in the higher time frame (15 min) and continue in the original direction. I suspect it is because traders who are trading the higher time frame tend to add to their positions as prices begin to exhaust themselves. It is a rather remarkable phenomenon. Hence, if I am adhering to Phantom’s Rules, I will be constantly adding and removing contracts with the ebb and flow of the market. This would be extremely difficult – not impossible – to code into an automated system. Perhaps an automated system is one that will evolve over time as we find new twists on how to trade more accurately.

I can read Cahen’s book a dozen times, and I learn something new each time. While the translation is a little rough, most of it makes sense. You should have seen the translation of his earlier book published in 2001. It was so bad that some parts couldn’t be understood – by anyone! Cahen told me that a non-trader did the translation. It showed!

I agree with your thoughts on coding a completely automated system for the reasons you stated. Mark Douglas, in his book, says that a completely automated system ultimately fails. I’m not quite so fatalistic about it as he is. I know there are people who have accomplished it. However, they tend to work on one trading instrument, but not another (for example, it might work on the S&P mini, but not on gold or crude oil). Douglas says in the book that automated systems tend to work for awhile, but as the market changes, they lose effectiveness over time. This is another reason why I believe that any trading system must be DYNAMIC in nature. It must use indicators, support and resistance, etc. that are as dynamic and changing (able to adjust to change) as the market itself transforms and transmutes itself. What a challenge!

Complete automation might be very difficult to accomplish. For example, I keep very tight stops when I jump into a trade, but as the trade makes more money, I become somewhat more easy-going about it. Phantom describes this in Chapter 10 of his book (the one about John Denver). In fact, as a trade becomes very profitable, I will move to a higher time frame to manage it. I have moved from the tick chart to the 3 min chart for soybeans this morning. I will probably manage the soybean trade on the 3 min chart the rest of the day. How does one code THAT into software?

Another problem in designing an automated trading system is what to do when the prices lock limit. A system would need some mechanism to detect that and either exit a trade or not execute new trades. Fortunately with the expanded limits, this is becoming less and less of a problem.

There are numerous little iterations and subtleties that I use in my trading. Another one is that the longer a trade continues, the poorer the Klinger Volume indicator works. I also tends to work only for the direction of the prevailing trend. Klinger is great for getting me into a good trade early, but then it loses its effectiveness. Klinger is also very good when the market moves back and forth, up and down like a sine wave between flat Bollinger Bands. It is used for entry and quick exit of non-profitable trades, but is not effective for exiting highly profitable trades. Once my trade reaches about $150-200 per contract, I pay little attention to the Klinger Volume indicator.