Market Environment Applications
Before initiating a trade, seek to define, even if roughly, directional movement and volatility. What do you see? Do they fit in with the conclusion you reached from the analysis of your other tools? If not, why not? Is it important? In your plan, this analysis could be either near the beginning of your heuristic -- to spot pairs with good general conditions -- or near the end, as a confirming tool.
Look at the time rhythm and price rhythm. Is the timing of both rhythms good for a trade? If either the time rhythm average or price rhythm average is off substantially, it may be good to take a bit longer look before pulling the trigger.
A market environment profile is the complete set of MEs for a given chart. A brief notation might look like this:
Specific currency pairs will sometimes exhibit stable market environment profiles over relatively long periods of time. For each trade you make, keep a short notational record of the directional movement and volatility for that market. Once a month, compare your winning trades with your losing trades. Almost all traders find they do better in some primary MEs than in others.
To dig deeper, keep ME profiles for all ME elements on your trades, and look for winning ME clusters and losing ME clusters.
Mutual and hedge funds, which use multiple managers, may use this last idea to allocate funds to specific managers for specific anticipated long-term MEs; managers receiving more money to trade in markets in which they excel, less in markets in which they do poorly. An ME cluster is a contiguous grouping of ME pairs in an ME matrix. Trading systems that work well over historical data only to almost immediately flop in real time almost always had a high majority of their big winners in one or two small ME clusters.
Market environments may also be used to back-test systems and methods using historical data. Rather than looking for the usual suspects of Sharpe Ratio and so forth, look for methods that did well in a wide range of market profiles.
Tip: A short, well-constructed ME data set will be a better test than years of data concentrated in a few clusters or even a real-time test. A long real-time back-test may be very deceptive with respect to possible future performance Consider: 1) markets may go for long periods of time without exhibiting every ME, 2) most of the profitable performance may be hidden in just a few MEs, and 3) the distribution of MEs may be very poorly distributed even in a long data series.