Ask HN: Please review my startup

As I understand, you basically attempt to correlate stock movement with two major components: (1) market/sector movement, and (2) news.

The first one is easy. Market/sector goes up, stocks go up, market/sector goes down, stocks go down (in absence of major news, more on which below). I doubt you would have created the website to prove the obvious, so I'll leave this one alone.

The second one is where I see the major problem. The problem being that NEWS DO NOT MOVE STOCKS. You'd think they would. After all, good earnings report, stock goes up, bad earnings report, stock goes down, right? Except that's not what happens. Granted I have limited trading experience (selling/buying/sometimes shorting several times a month at the most) but I've seen enough to conclude that there is no meaningful correlation between news and stock movement.

I just tracked several stocks that I happen to follow where I have a reasonable idea as to why they've moved in the last year. Your chart only proved the above. News were well presented along the spikes/dips (again, the UI is awesome) but the clear presentation only served to show that no correlation exists. (To clarify, by news I mean anything and everything, e.g. earnings reports, analysts upgrades/downgrades, company announcements, actual news, etc; and by no correlation I mean good/bad news accompanied spikes/dips equally.)

Anyone care to argue with me? Prove me wrong!

From the questions in the comments here on HN, it seems clear that everyone else is in the same boat, not really getting why this is any different from google finance.

You need to experiment with better visual ways of communicating that the blue 'market index' curve in the graph is something you invented, defined by the pie chart to the right.

It also wouldn't hurt if there was some unobtrusive way for you to emphasize why that's important. You're trying to isolate market movements out of stock movements, so that you can correlate news with times when the stock diverges from what would be expected based on the market and the industry/sector it's in. To me, that's great, but neither your intention to do that nor the effect of everything you've done to try to illustrate that is quite coming across.

- It might be a good idea to note stock:market divergences on your chart

- Don't know where your parsing your news, but you might want to incorporate a filter for SEC filings and PR releases along with a way to eliminate the fluff/SEO (Motley Fool comes to mind)

It seems (as blurry pointed out) that you are assuming that news leads/lags a stock. There's also a whiff of efficient market theory in there, and at the expense of pissing off half of HN, the markets aren't efficient.

Other than that, I find a hard time finding the difference between what you have and other tools I use.

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.

I don't understand the use of the "index"...

The logo and interface looks a little '5 years ago'..

>>> For every stock we create an index that measures market and sector influence and identify the news that make the difference. <<<

But I think you're trying to say too much in that sentence anyways. A simpler and better statement that's easier to understand and probably has greater impact might be this:

>>> We measure market influences for every stock in America, and we identify the news that makes a difference. <<<

If you use this new wording you can pay me $1000 ... :)

Aside from this change, I like the grab-and-pull interface that works like google maps, but I'm not into investing so I don't know if real investors will find any value here.

As far as site wise -too much space on the bottom -newstrends are pretty useless for terms, since you get results like you have now "EPS" etc, I'd replace that with newstrends for actual companies.

1. Change your domain name from whomovedmystock.com to whomovedmycheese.com.

2. Change your product from a website to a childishly simple business book.

3. Stop coding and start lecturing and conducting business seminars.

Other than those few small tweaks, I'd say you're on the right track. Keep up the good work!

1. it looks pretty

2. when I move the bar at the bottom of the chart, the applet keeps reloading. This is annoying. Is it intentional or a bug?

3. the bit at the top left where it shows a pie chart of Google's "Market index allocation" -- how is this calculated?

4. On the chart, it's not obvious to me what the blue line "market index" means

On another note, I'd like to see a site called whomovedmysocks.com

Can you put in the extra effort to chart indicators, like VIX and TED and Case-Schiller? CSCO and MSFT stock charts might be tapped out, but this might become useful if it acts as a portal to lesser-known stats.

as others have pointed out, this doesn't provide anything more than what google finance already does.