When it comes to typical stock market data, people assume it to be price data, company fundamentals and some technical indicators – most of which is nicely structured and straightforward to analyze. But off late it’s been seen that “alpha” lies in information that is not easily available, rests in disparate sources and is generally quite difficult to navigate through. Of course, it’s only natural that something that takes extra effort to understand, might give you an edge if you are able to parse it well enough.
When we, at Stockal, first decided to pursue the idea of democratizing investing intelligence and stock market information, we didn’t quite know how challenging it will be. Why? Because we were not quite aware of exactly how unclean, unorganized and ambiguous such information really is.
So – challenging it may be; but it’s still a lot of fun and helps keep us sharp. It also promises to be rewarding for an early-stage company such as Stockal because we have seen how much our users value the information we bring to them. So let’s look at what we do.
How we look at information
As mentioned, structured price and fundamental data is easily available so there are limited opportunities in finding additional insights there. So which “information” are we talking about here? News, social media and analyst opinions, to be precise.
There are thousands of sources of unstructured information on the web – many of which offer pretty useful insights to the reader, but most of which are tough to automatically analyze at scale by a machine. Another minor challenge, also, is that new sources of news and analysis keep cropping up every now and then so it’s not optimal to create a pre-fixed list of sources and analyze only those.
Thankfully, search engines like Google and, to a certain extent, Bing and DuckDuckGo offer great source discovery mechanisms that we have made use of.
Starting with a list of NASDAQ and NYSE listed stocks and ETFs, our engine looks for news about each of the listed companies and their stocks on hundreds of websites everyday. Whenever it finds something remotely relevant, it brings the content of the said news item into our fold. Alongside discovering news on popular websites, as mentioned our engines also constantly discover new sources of news everyday – so that we may not miss that obscure website or that little read reportage.
Then our curation engine goes to work. Starting with the headlines, this goes through each line of content and flags irrelevant material based on date of posting, frequency of mention of the company it’s tagged to and overall mention of certain keywords we deem avoidable because they are potentially manipulative, salesy or misleading. The content that remains is analyzed for sentiment.
Unlike news, social media sources are much easier to manage. Because they are fewer in number, are discoverable through Twitter and StockTwits, and tagged to users whose reputation can be automatically ascertained.
Yeah, so we look at Twitter, StockTwits, discussion forums and comment threads of analyst blogs.
This is easier to deal with – especially on Twitter and StockTwits – because of the extensive use of “cash tags” (“$” as in $AAPL, $GM, $TWTR etc.). While it may be limiting in some sense, it helps keep our data clean by looking at only those social conversations and discussions which are about the stock and not about the company or its products in general. There are definitely some exceptions – for instance, discussions on iPhone sales, reviews and expectations must be analyzed for their potential to impact Apple stock price.
Once again, data once curated based on timeliness and direct correlation to the stock, is analyzed for sentiment. But there are two other things we do with social media that we don’t do with news or analysis.
We look for trends such as sudden increase social media chatter volumes – and we call it Social Media Pulse. This is important because an unusual increase in chatter typically indicates that “something must be brewing”. Something that needs your attention if you’re interested in that stock. This helps flag stocks from our users’ portfolios and watchlists so that they can pay attention to just the stocks that need to be looked at.
Stockal also has an algorithm that catches potentially manipulative chatter on Twitter. There are many bots on Twitter that automatically try to pump stocks in the hope that some trading algo will catch the trend and trade on it. So it’s important to make sure such attempts are flagged and avoided.
This is a very interesting offering from Stockal. If you use websites like Yahoo Finance or MarketWatch you must have seen that almost every one has an analyst recommendation section that shows how many analysts are bullish, bearish and neutral about any stock.
What most of these websites/platforms miss is that such information is not of great value to most readers because it’s not quite complete unless you know opinions of which analysts were analyzed and how good each of those analysts really is. In other words, its nice to have an opinion of an “expert” but not one must take into consideration what the value of said opinion is.
So we invented a new parameter that takes analyst reputation into account. And called it Confidence Meter.
We’ve written a detailed piece on the Confidence Meter here.
Fed up of having to read through tons of news content and unable to keep with so much social media chatter for your stocks? Feel like you are missing information? Use Stockal to stay on top of things that may be impacting your money.