There are thousands of companies with very low capitalization. Their stocks are worth order of a dollar and they can change tens per cent a day. In other words, they are extremely risky. Large players ignore this market because it can't digest their capitals -- the liquidity is limited. Each penny stock lives its own life which is fairly separate from the life of the stock market or even the economy in general. It's understandable because for a tiny company, its local news are much more important than global news. A good management decision, acquisition of a strategic partner or a client can easily double the earnings. The penny stock prices are moved by news real or imaginary and can be manipulated by various means, including sales campaigns targeting naive investors.
For the same reasons why poor people outnumber the rich, tiny companies outnumber the big ones. Because they are so numerous and move quite independently from each other, there is a chance that at any point in time, you can find one or two of them which are at a certain point of some familiar pattern -- within a certain degree of approximation, of course. The pattern or a group of them can be inferred from historical data, aggregated for this class of stocks. An automated system might scan the real-time market data for these thousands of stocks (thousands of time series) and search for those "archetype" patterns which do repeat themselves. The system might then generate a buy or sell signal. Since there are thousands of opportunities, the system may have a very high selectivity threshold and still generate a few signals a week.
Tim Sykes (TimothySykes.com) trades these stocks in a way informed by long experience and sells alerts and educational materials. In the past couple of days, I went over Tim's trades for the past month, posted on his site, to see the patterns and in hope of learning from his trades. The patterns do exist and are very clear in this market, unlike the market of the larger companies which attract the media attention and big traders. There are two problems however: one is lack of liquidity. Tim says he limits his trades to 1% of liquidity. Which means that as his personal fortune must be at about $2M, it unfortunately can not grow exponentially. It can only grow linearly since he must be trading only a small fraction of his account, more or less fixed in the nominal dollar amount. If a stock has daily volume of 100,000-1,000,000 shares and trades at $1, a trader with such a rule can trade 1,000-10,000 shares, and if his capital is order of $1M, he is limited to 1% of his account. No wonder he needs no leverage. He also is not diversified -- there is little need to since a much better way of managing the risk is to apply extreme selectivity when adopting a trade idea. It is doable in this market because of a huge selection of companies and therefore, a huge selection of potential ideas to consider.
That leads me to the second challenge: who is going to consider all those ideas. The best way is to use an automated, programmed pattern recognition system. If I decide to do that, there should be very little competition as institutional players who are able to hire someone capable of researching and implementing that (they would need a quantitative research department) are not interested in the market which is too small even for a guy with a couple millions. This may be a real chance to make the amount of money big enough to lose interest in this market before moving into something else.
For now I am going to keep monitoring Tim's trades and posting charts here.