Thanks all, these are great suggestions. Re: Cointegration. I understand that instruments that are highly correlated may not be cointegrated, but will instruments that are cointegrated also be highly correlated?
As mentioned TOS has a study "Comparison" that allows you to overlay a line chart of any symbol over your primary instrument. Or you can compare any 2 at BigCharts.com by clicking on their "Advanced" tab. Compared here are UBER & LYFT for yesterday http://bigcharts.marketwatch.com/advchart/frames/frames.asp?show=&insttype=Stock&symb=uber&time=1&startdate=1/4/1999&enddate=1/25/2020&freq=9&compidx=aaaaa:0&comptemptext=lyft&comp=lyft&ma=0&maval=9&uf=0&lf=1&lf2=0&lf3=0&type=8&style=320&size=3&x=55&y=16&timeFrameToggle=false&compareToToggle=false&indicatorsToggle=false&chartStyleToggle=false&state=13
You could fit a "vector autoregression" (VAR) to test for "Granger causality". The equations for a VAR(1) with two variables x1 and x2 are x1(t) = c11*x1(t-1) + c12*x2(t-1) + e1(t) x2(t) = c21*x1(t-1) + c22*x2(t-1) + e2(t) where e1(t) and e2(t) are noise terms. If c12 is statistically significant, x2 helps predict x1. If c21 is statistically significant, x1 helps predict x2. Note that it is possible for x1 to be predicted by past values of x1 and x2 *and* for x2 to be predicted by past values of x1 and x2. There are many R packages and at least one Python package for VARs. In general you should read about "time series analysis".