How to predict the S&P 500 Index by following the stock price of Dun&Bradstreet
There is logic behind our assumption. To better understand why we use the stock price of Dun&Bradstreet to predict stock market cycles we will dive into a short introduction what Dun&Bradstreet actually is.
Dun & Bradstreet, Inc. is an American business services company headquartered in Short Hills, New Jersey, US that provides commercial data to businesses on credit history, business-to-business sales and marketing, counterparty risk exposure, supply chain management, lead scoring and social identity matching. Dun & Bradstreet maintains a database of over 240 million companies globally and over 100 million professional contact names using a variety of sources including public records, trade references, telco providers, telephone interviews, print, digital and trade publications, among others. The company derives revenues through subscriptions as well as pay-per-business report.
You already realize that their is a wealth of important information which can be exploited in order to predict market sectors and the broader stock market. We are going to keep things simple by not extracting data from their database instead we will follow the stock price of D&B. Why do we follow the stock price ? The price of the stock is reflecting the revenue flow of D&B. The revenue increase and decrease depends very much on the number of subscriptions and new companies registering for database inclusion. Isn’t it a logic assumption that there will be more companies registering and more subscriptions in an economic uptrend and vice versa. In other words the state of the economy will be reflected in their revenue stream and subsequently in their stock price. Inevitable the stock performance of D&B will influence the price of the S&P 500.
Of course their are other databases but none of them comes close to D&B in terms of data quantity. Its safe to say they have a sort of monopoly. The Company operates through two segments: Americas, which consists of its operations in the United States, Canada and Latin America, and Non-Americas, which consists of its operations in the United Kingdom, the Netherlands, Belgium, Greater China and India.
Dun & Bradstreet traces its origin to Lewis Tappan, who in 1841 left Arthur Tappan & Company (a New York silk trading firm) to found a credit information bureau called the Mercantile Agency. Tappan had long been aware of the need for better credit reporting. As the borders of the United States expanded westward, traders were moving beyond the easy view of the East Coast merchants and bankers who kept them supplied and capitalized. Information on the creditworthiness of these far-flung businesses was collected by individual trading houses and banks in a scattershot fashion, and Tappan saw that centralizing the process of collecting information would result in greater efficiency. Accordingly, he took out an advertisement in the New York Commercial Advertiser on July 20, 1841, and opened a shop 11 days later on the corner of Hanover and Exchange streets in Manhattan.
The Mercantile Agency operated by gathering information through a network of correspondents and selling it to subscribers. The agents were attorneys, cashiers of banks and merchants. As the nation grew and commerce boomed in the decades following the Civil War, Dun had to keep up with it by establishing new branch offices. In 1933, at the trough of the Great Depression, R.G. Dun merged with one of its main competitors, the Bradstreet Company. Since the two companies overlapped each other in many activities and resources, an amalgamation at that time made sense. Bradstreet was founded in Cincinnati, Ohio, in 1849 by John Bradstreet, a lawyer and merchant.
In the subsequent decades the company expanded rapidly by numerous acquisitions. In 2000, then, the Dun & Bradstreet Corporation split in two again, spinning off Moody’s Investors Service as an independent, publicly traded company. It was then incumbent on the Dun and Bradstreet operating company to spearhead improved earnings, particularly through the aggressive expansion of its Internet presence.
Inter-correlation Testing D&B vs. S&P500 Index
Our tests account for market cycles. The optimization tests are performed on weekly data with entry long from 0 to 10 bars ago. Period tested 2002 to 2013.
- 4185 runs without stop loss and profit target adjustments:
Results without stop loss and profit optimization show non satisfactory results. The Monte Carlos Simulation reveals that most of the trades no matter where the starting point is need an average of 50 trades and more to break even.
- After 121 runs with stop loss and profit target adjustments:
At the first glance the results seem more promising. Nevertheless a second look reveals the deficiencies of this strategy. Starting at the beginning of 2008 until the first half of 2009 there are no trade signals generated. Another such period occurs in 2011 and 2012 with not a single trade.
Approximately 1/3 of all 99 Monte Carlo simulations need 50 trades and more before they break even.
Conclusion: Based on our results there are numerous drawbacks trading this strategy. However we think their is a predictive power to a certain degree but further research and additional data is required. Surely a look under the hood and extracting data available in the D&B database could definitely produce far superior results but such research would go beyond the scope of this article. For those interested in extracting data and using it for hyper contextual trading (HTC) here is a link to dive deeper into this topic Text Mining with R.