[kimchi]

[ QUOTE ]
However, I have found that a statistic-based approach to TA (i.e. 1700 of the last 2000 times MACD has crossed at X point, price has moved higher) provides a demonstrative advantage. Or at least it has for my net-worth.

[/ QUOTE ]

This is (another thing) I've been interested in and stuggling with recently. Looking for observed relationships between markets or between prices and price derivatives (such as the MACD you mentioned).

A lot of these ideas have been extensively written about in various texts (such as the probability of a reaction after a price derivative's divergence from the price or a particular candlestick formation, or the likely behaviour of Asian markets given the overnight action on the NYSE).

I'm worse at statistics than I am at poker, but apparently statistically significant events need not have a rational explanation to a technical analysist. If the action is tradeable then it can be profitable.

Can you give a specific example of statistically significant behaviour of price that you were able to exploit and trade? I think many exist, but since the markets are so complex, I suspect that the robustness of many of these observations will degrade over time which may suggest curve-fitting.