It is hard to imagine stock prices as being anything but random, and in a certain sense they definitely agree with my definition of randomness. This article focuses on some of the pockets of nonrandomness that appear from time to time. I will dismiss with long-term trading up front; relative strength is the single best thing to be watching for making multi-month investments, and coupling these with an eye for strong earnings growth and a contrarian bent can produce stunning returns in a healthy equity market.
Unfortunately, equity shorting strategies based on decelerating earnings and poor relative strength are an extreme challenge, and your only hope may be a long-short equity hedge fund-of-funds for that sort of asset. What follows is an attempt to make predictions over 30 days or so that can be exploited to hedge the other risks in the market on one's own.
For those of us not on the floor, there is evidence that minute-by-minute movement of stocks is noise. That is, the autocorrelation of price movements is noticeable at very small time horizons (under a minute) and dissipates shortly after that. One reason for this is that market-makers (or anyone acting as one) act to remove as much momentum from short-term prices as they can, enabling them to supervise a significant directional move without carrying a large position that has been shorted (or bought from) those making directional trades.
Daytraders can make money using sheer cleverness and experience, suggesting that there are tricks to be learned, but this is beyond what I'm writing here. The net effect of all this business is to give simple noisy oscillation around some temporary market equilibrium. For a $100 stock with 30% annual volatility (around twice an index) this results in an average move of about 0.3% every trading hour. What is interesting are the distortions that cause the market to find new equilibria.
One of the most compelling technical indicators is the simple moving average. At first, viewing price charts with these lines, it seems there is lots of mean-reversion to trade on; when the stock moves significantly away from the moving average it magically snaps back. The problem is that the average can simply catch up with the price rather than vice versa!
Having tried this, the next compelling thing to do is to combine a short-term SMA with a longer one. The lines will cross when downward movement is opposite a longer trend for a little while. My experience is that a trading system based on the crossover of moving averages is probably not going to be profitable after transaction costs. It seems about 50-50 that a given SMA or EMA crossover is a precursor to significant additional movement in that direction, so we are prompted to look harder.
Support and resistance levels are interesting and very visual. There do seem to be key price levels for a lot of stocks, but devising a system that works from this assumption is hard. This is where we get the simplest thing that I think presents real, tradable effects: breakouts above or below key levels often proceed in the same direction for a little while. I think this happens when the level of buying or selling pressure becomes too much for technicians and traders to offset, because little pockets of stock demanded or offered are waiting for them at those levels. I think this system presents opportunities for up to 3% gains when an initial risk level is established (via stops) to 1%. I would describe it as labor-intensive, and the potential for slippage in these distorted markets is high. For large accounts (but not too large), breakouts in stocks with very liquid options markets could be profit opportunities.
Another promising indicator which comes in many forms is the momentum oscillator. Stochastic, for example (a poor name) often identifies buying opportunities after a selloff and is very sexy when displayed on a chart. There are opportunities to devise a system that risks around 2.5% for a 7.5% gain over several days. Options speculators in liquid markets can do exceptionally well here because options are very profitable upon trend reversals. This is because those who are chasing the latest move in the opposite direction raise demand for the option (or market-makers simply mark them up in anticipation of this pressure).
One challenge with this system is being equally long and short so as to lose the systematic S&P 500 risk. Diversifying across many options is hard because your market exposure changes dramatically if the delta of an option on a high-beta stock changes. Also, many stocks tend to price overbought or oversold at the same time, and when betting on a bounceback rally for the market as a whole, there may not be ample shorting opportunities out there since everything has been hit already. At the moment I think careful hedging via S&P puts, synthetics, or futures can do the job. In the aforementioned case, a simple short on the market averages can be chosen so that if all the bounceback calls are hurt at once, they will be paid for by a rise in the puts' price. If we do win and there is a bounceback, nonlinearity will assure that the option portfolio wins in the end.
Another promising effect is revealed through volatility and standard deviation bands. Plotting the 21-day volatility of a stock reveals certain short-term troughs where the market noise has been quiet. This is also represented by the narrowing of two-sigma bands to a small region. It seems that this indicates a pool of supply or demand is building up in price levels somewhere nearby, because a large price move often follows this situation. This is the stock market's way of making sure that volatility reverts to the established mean. To trade on this effect one needs to buy volatility using option strangles, calendar spreads or straddles. However I think 15% gains are possible with risk being around 5% from time decay and other factors. Directionally trading this effect could be possible as well but it is unclear to me which way to speculate.
To summarize, I've indicated three real technical trading possibilities with system reward:risk ratios of around 3:1. This means a win rate above 40% is all that is needed to ensure profits in the long run. I would appreciate feedback on what luck traders have with these strategies. Since each is rather different, deploying all three at once in an active trading account (along with, perhaps, simple merger arbitrage, convertible hedging, and the like) could present a good diversification to more traditional portfolio management.