Let's say I wanted to test two commodities to see how much they correlated over time. What's the starting point? I mean I could take today's prices, or any day's prices as long as they're the same day and say there's no divergance. I'll take that day's difference in prices as 0% divergence in price. Is there a more correct way to do this? It seems as if I start the analysis on an arbitrary day I may skew the whole study. For example if I started the analysis today and the two commodities have moved away from each other significantly in the past few time periods, wouldn't the results be totally screwed up. I may be thinking about this too much though. A historical test of price divergence may hold true no matter what day the test starts, right? This brings me to another question: What to do about scaling the y-axis for price difference? ex. comm A trades @ 2500 and comm B trades at 600. Is it best to mult 600 and all future prices of comm. B to match comm. A prices?