What you are trading is one data set against another. Regardless of what the data comprises of.
Wether it's a positive correlation or a negative correlation is irrelevant. What you have is a calculable relationship. A strategy for example would be to look for a break down in correlation on the short term and then trade towards the long term correlation.
EurUsd vs GbpUsd for example. They're normally about 80% correlated on weekly. So lets say EurUsd goes down and GbpUsd stays flat (which is a break down in correlation) then you'd buy EurUsd and sell GbpUsd. It might go to a temporary negative correlation in such a case on a lower time frame.
Of course that gives you a EurGbp position. But to me it's a better method than technical analysis. It's calculable.
If it was EurUsd vs UsdChf, strong negative correlation, you'd just buy both or sell both on short term breakdown in correlation.
Problem is every now and then correlations break down completely for long periods of time, especially in this type of market where prices are heavily and openly manipulated. But as with any type of system you can come up with account protection methods, like stop outs on NAV losses, trade size reductions if you detect a series of failures etc...
Another way of doing it is to simply overlay highly correlated pairs on chart, then it will become obvious where to trade and in which direction, but the position will always consist of two trades. You trade both pairs and regard it as one position.
Hope that helps.