It's 7 a.m. and a slew of reports await you.
Your analyst, working all night, drilled down on the latest impact of the Eurozone crisis-yes, the banks are being rescued again-on a dozen industry sectors, and then wrote notes on 20-interesting stocks. Meanwhile, your assistant manager brainstormed new risk and allocation strategies for six portfolios.
Great job! You say. Bonus-well, at least a latte-you say.
Well, keep your money because neither are human.
The automation of analysis is arriving, thanks to technology that takes data and turns it into written reports-and, hopefully, insights.
Developed by two-year old artificial intelligence firm Narrative Science, the system, called Quill, reads any and all incoming data you assign to it. Then, it generates as many analytical reports as you like, in formats you create. It can be trained to use the same formalized analytical processes your staffers use and tailor reporting for each individual reader.
"Quill looks at the data and figures out what one can take from it, and then it figures out how to make genuinely insightful stories," says Kris Hammond, chief technology officer and co-founder of Narrative Science.
There have been systems screening financial data and handling basic searches perhaps since the days when vacuum-tubed computing devices filled entire offices-Quill goes beyond this.
First, it scours all data that flows into a financial firm: market data, economic stats, credit and risk ratings, news stories and drill-downs on a portfolio's companies and assets.
To create a sentence such as "Hood Chemical failed to meet its fourth-quarter earnings estimate of 10 to 20 cents a share," the system might pull earnings estimates provided, for example, by Zacks Investment Research and compare them to earnings results disclosed in a 10-K filing with the Securities and Exchange Commission.
To find volatility in a story, it might pull price changes from Nasdaq's TotalFeed of stock market data, to produce a statement like "Dynga shares bounced back 5% from yesterday's carnage, when the price of a share fell 10% to $12.50," or an Associated Press news feed to declare "Eurozone leaders agreed to radically restructure Spain's $124 billion bank recapitalization plan.''
Then, just like an analyst or journalist, it starts to mull all the potential interesting "angles" of this new information. Angles could include cases for industries coming in or out of favor or particular companies that might be worth closer inspection.
After the system pulls in all available facts, and evaluates all possible angles, it is ready to write. The "analyst" bears in mind the interests (previously inputted) of its target readers. And it uses familiar writing styles.
For example, the system is taught to spin out effective story "arcs" that work, basically, like a top-down newspaper article. Facts of greatest impact go first with background explanations following after. Stories on similar subjects are referenced for context. Full names and stock symbols aren't spelled out after first company mention, etc.
How does the system know these things?
That comes out of Narrative Science's development process. Roughly a third of the company's 40 employees are writers, journalists and other analysts. They are trained to be aware, like cognitive psychologists or anthropologists, of all their thinking processes as they write. They can break down these methods and translate them into code with help from their programmer colleagues.
If, then, an analyst's thinking processes can be articulated clearly and comprehensively (consider all the in-house guidebooks and training processes you've already developed regarding your "investment strategies"), then the team at Narrative Science will build an addition to Quill around it. The company is continuously developing new strategies for studying and adapting client thinking processes.
"The limits on the application of this product in finance are only the limits that exist in analytics itself," Hammond says.
The company continues to look for new applications: like systems that can analyze all the information on a fund's distribution process and suggest cost cuts, and so on. If a thing produces data, Quill can write about it.
Once Quill is taught to write a particular kind of report, it can then crunch them out at lightning speed and superhuman volume. And customize each one.
"The idea of focusing on an audience, on who I am talking to, is essential. Will they be interested in this characterization, or in this angle? We can distinguish between what one audience would want in comparison to others and tailor accordingly," Hammond says.
Already, Quill crunches out thousands of sports, real estate and financial stories for clients weekly. It also helps other companies, such as restaurants or retail chains, scour through terabytes of monthly sales data and describe what's going on.
Asset manager Charles Sizemore, chief investment officer of Sizemore Capital Management, which manages global macro, strategic growth and tactical exchange-traded-fund portfolios, said he would "definitely consider" Quill. He views the system as a potential time saver.
Sizemore is no stranger to technology; he also manages portfolios on the social media platform Covestor. He said that while a "computer can never replace a talented investor like a Warren Buffett," good technology can certainly allow one to make better use of his or her time, focusing on qualitative aspects and larger "big picture" macro themes while the computer does the grunt work digging through financial statements.
"To the extent that securities analysis can be reduced to 'ones and zeros,' i.e. simple yes/no answers, software such as Narrative Science can save man hours and add a degree of precision that you're not going to get from human eyes," he said. "When it comes to repetitive tasks, humans make mistakes. Machines don't."
However, he says humans still have a place in investing. Computers are only as good as their programmers, and systems have never been particularly good at seeing "curve balls" coming - like a sudden dotcom bust or a crash on mortgage-backed securities.
While such systems would be good for generating insights on past trends, intuiting future possibilities is still a human talent.
"Computers are not good at looking around the corner to see what may come next," he said. "That's where human intuition is still very valuable."