Perry Kaufman is a master of trading systems, and he shows it in his outstanding book, Smarter Trading: Improving Performance in Changing Markets. Kaufman details a number of interesting issues for developers of mechanical systems, and I've summarized some points I found of interest below.  Of particular interest to me as is Kaufman's concept of an Efficiency Ratio, which can measure the amount of directional trend you get relative to the corresponding volatility of the stock over time. 

In addition, Kaufman has published another excellent book on trading systems and their development, entitled Trading Systems & Methods.  These books are required reading if you are looking to test your own ideas as potential trading methods, as Kaufman illustrates a number of dos and don'ts that all investors can benefit from as they seek to test their theories on past data and put them into action in real-time trading.
 
1.  Efficiency Ratio = (10-day price change)/(Absolute Sum of 10 daily closing price differences) - the idea here is that you can get a read of the net directional movement in the trend compared to the total amount of distance the stock moved over the period based on the sum of its daily ranges. We liked this indicator so much, we added it to our BigTrends Option Charts package. We'll be doing an upcoming daily writeup on a current example of an "efficient" trend.

2.  The more mature the market, the "noisier" it is - too much participation leads to poor trend-following results (can you say "Cisco" 12-18 months ago?).

3.  "Use of spontaneous judgment only gives you the opportunity to delay the exit, hold onto that losing trade, and wait for a worse price to come along."  Prepare before market open.

4.  "Unables" are likely to stay winning trades.

5.  The shorter-term the trade, the more bid/ask & slippage cut into return over time (not to mention the higher commissions of fast, more frequent trading).

6.  "If market relationships change, you cannot use old data to forecast the future."

7.  "It is not the forecasting attributes of a trending system that allow profits, it's the risk controls."

8.  A reliable strategy is one that uses relevant information or specific indicators only when they are important.  The market does not always have something to say.

9.  The extra profit that may be gained at the end of a move tend to have much higher volatility associated with marginal returns - the definition of a change in trend.

10.  Stops are reactive; Kaufman not a fan of tight stops due to short-term market noise.

11.  Long-term systems are more predictable than short-term systems - use weekly & monthly systems over daily systems.

12. The level of volatility that occurs during a sustained sideways directionless period is a convenient measure on intrinsic noise.  A price trend will be unreliable if it is signaled by a move that is no greater than the intrinsic level of market noise.

13. Adaptive Moving Average - speed up in trends & slows down in ranges.

Price Headley is the founder and chief analyst of BigTrends.com.