Mining info from Web for financial sector

Mining info from Web for financial sector

There are one billion people online every day expressing their views on one issue or another.

There has to be information out there which is useful in the financial or commercial world. Based on this premise, a Singapore-based company has set out to mine all that's being put out on the Web every day, and distil it into a few sentiment metrics that investors can use to trade the market.

Two of the shortcomings of financial market data - price, volatility and correlations - are that one, they reflect new unstructured information only with significant delay; and two, they don't "price" properly the shifting market sentiment and insights, says Marcello Fontana, founder and managing director of Wavecatch.

"We try to capture that sentiment on the ground as it happens, combines it with market data, to come up with short-term price forecasts for equities," he says.

According to Mr Fontana, the company's system scans the 800 million documents published in the various channels of the Web every day - news, blogs and social media platforms such as Twitter - in English and Chinese, and filters them for certain relevant keywords.

Documents with these keywords will go through further downstream analysis. Typically, 2.5 million "relevant" documents will be identified each week. The system will identify, for example, brands, sub-brands or names of companies with positive or negative mentions. The information will then be scaled and standardised.

This measure of sentiment will then be combined with market data such as company-specific time series, macromarket indicators such as index futures or bond indices.

The output is short-term, that is 24-hour, forecasts of volatility and price movements for that particular stock. Three or four signals daily will be generated across the markets.

The system now covers 240 big-cap stocks in the United States, UK, Greater China and Australia.

Wavecatch is focusing on the equities market first and intends to start off by licensing its sentiment metrics and forecasts to institutional investors, hedge funds and proprietary trading desks of banks. Over a longer term, it is looking to partner online brokers to develop products for the retail investors.

For retail investors, the system will have to be tweaked to predict the one-week performance of stocks, says Mr Fontana. Twenty-four-hour trading does not generally make sense for retail investors because they incur higher transaction costs.

According to Mr Fontana, Wavecatch has back-tested its systematic strategies. "Potentially, after transaction costs, the strategies could return around 15 basis points per day."

The brain power behind the company is impressive.

Mr Fontana, "half German, half Italian", has over 15 years of global experience in financial services and technology/media companies such as Wells Fargo Bank, Intuit, Standard Chartered Bank, Aegon and Fox/News Corp.

Prior to founding Wavecatch, he was managing director for Asia-Pacific at Fox Mobile (Fox/News Corp), and global head of R&D and innovation at StanChart, where he established and managed the $1 million a year Innovation Lab @ SMU.

He holds an MSc in econometrics/statistics, and earned his MBA (technology management) at UC Berkeley (Haas). He's been in Singapore for eight years.

Your daily good stuff - AsiaOne stories delivered straight to your inbox
By signing up, you agree to our Privacy policy and Terms and Conditions.