I think what you've got here has a lot of potential; let me echo some things and maybe make some other suggestions. First, I think it is a great idea decomposi

I think what you've got here has a lot of potential; let me echo some things and maybe make some other suggestions. First, I think it is a great idea decomposing stock variance. Really, I do want to know what moved my stock. However, it's not really clear what your chart is telling me. I see market and I see the stock price and I see that they are different but I don't know why they are different. What I'd imagined was suppose you start with the return on the value weighted Wilshire 5000 possibly adjusted by empirical beta. Presumably there will be some residual price movement, some of that is coming from sector changes. It would make sense to see the sector effect as a different color tacked on to the market effect. Then you add news effects and then the leftover would be unknown or some such. Perhaps on news days you assume all the residual movement is due to news and then carry forward that price difference through the non-news periods. As you've got it I assumed your chart was just comparing the stock return with the return on the S&P until I read the comments here. You're news history and the associated price change is awesome, in academic finance there is a long tradition of doing event studies (e.g. what is the effect on a stock price to an announcement of a seasoned equity offering?). If I understand what you're doing you've essentially done an event study of every possible event for all stocks throughout time. That's pretty cool. In fact, if you used academic standard methods calculating your expected returns you might even have a dataset that research universities would be interested in buying from you. One possible thing traders might like that you might be able to pull off. Suppose I'm worried a stock is going to get hit with an "identify theft" leak. If you put a little text processing on your news titles you could determine the average effect that kind of announcement will have on the average and maybe for a stock in particular. So lets imagine "identity theft" has an average negative return of six percent. Then I check identity theft with American Standard and I get 0.1 percent because the negative identity theft returns are linked with banking and retail sector returns and not really with the market. I'm a little worried that your design is too cutesy. I would expect websites that do lots of statistics to have more hard corners. I understand you want the site to look accessible but you want it to appear intelligent and accessible. There's got to be a compromise in between somewhere.