Sentiment Analysis of Financial Blog Posts: Calculated Risk and Grasping Reality

By Jett Hollister

This is part of a series. Click here to read the previous article.

Technical explanations of procedures contained in this analysis can be viewed here.

To examine the differences in sentiment between blogs, we started by gaining a deeper understanding of Calculated Risk and Grasping Reality, the two blogs that behaved most similarly. We explored the writers’ backgrounds, interests, and economic ideologies, as well as the spread of topics within each blog. We examined how these differences could have generated discrepancies within the sentiments expressed in Calculated Risk and Grasping Reality, and inferred the correlation between the blogs’ sentiments and the state of the U.S. housing market from these factors.

The discrepancies between the blogs’ behaviors could have resulted from differences in economic ideologies or topic coverage. While economic ideology is a window into how each writer would have responded to key events, it is difficult to quantify the relationship between economic ideology and sentiment. By extracting the main topics of each blog, we were able to quantify the blog’s focus on the housing market, economy, and banking industry. The consistency and scale of topics allowed us to measure how accurately sentiment reflected the state of the housing market and related industries.

James Bradford DeLong, the primary writer of Grasping Reality, is a professor of economics and finance at UC Berkeley; he has also held a variety of positions related to economic research and policy[1]. A self-proclaimed “Rubin Democrat” – in reference to former Clinton Treasury Secretary Robert Rubin – he described his views as “largely neoliberal, market-oriented, and market-regulation and tuning aimed at social democratic ends”[2]. DeLong’s interests include social democracy, financial crises, economic growth, and behavioral finance[3].

Grasping Reality’s spread of commentary reflected DeLong’s wide scope of interests, including election opinions, economic history, and market analysis. As the above graph demonstrates, the blog’s focus constantly shifted with current events. Discussions of the housing market held an inconsistent presence within DeLong’s posts until late 2007, and even at the peak of public conversation and awareness in 2008, this surge was overcome by election commentary. As a result, the relationship between DeLong’s sentiment and the state of the housing market was fairly weak. Market analysis rarely held the largest share of posts per month, widening the potential for other emotionally-charged discourse, like that of a presidential election, to skew sentiment regarding the housing market.

Calculated Risk, however, was ahead of the curve in its discussion of banks’ lending practices, the housing bubble, and the impending economic collapse[4]. As a result, nearly every post was related to the state of the U.S. economy, with a focus on the real-estate market. Bill McBride, the author of the blog since August 2005, has a background in management, finance, and economics[5]. The blog constituted the gold standard of housing market updates and projections following its conception in 2005.

The above graph reinforces the blog’s self-proclaimed focus on the real-estate market and related industries. Regardless of the number of topics extracted, none deviated from this focus. Not only did each topic encompass some aspect of the housing market, but the frequency of each topic consistently followed the overall increase in posts. As a result, Calculated Risk’s posts remained in tune with the state of the housing market and its sentiment was highly correlated with the market’s performance, projections, and shortcomings. The sentiment of McBride’s post-2007 commentary aligned not only with what we would expect to see from financial experts and the general public, but also remained consistently negative across almost every year. The focus, consistency, and post frequency of Calculated Risk could indicate that its sentiment was an indicator of the U.S. housing market and even of the impending crisis. The mid-2005 increase in housing market discussion coincided with a major spike in negative sentiment. The early, strongly-negative sentiment values could represent indications or awareness of the deterioration in subprime loans or other instabilities within the housing market.


[1] DeLong, James Bradford. “Brad DeLong’s Short Biography.” Brad DeLong’s Grasping Reality. Accessed April 22, 2020. https://www.bradford-delong.com/short-biography.html.

[2] A Clinton-era Centrist Democrat Explains Why It’s Time To Give Democratic Socialists a Chance: Zack Beauchamp – https://www.vox.com/policy-and-politics/2019/3/4/18246381/democrats-clinton-sanders-left-brad-delong

[3] Faculty Profiles: https://www.econ.berkeley.edu/faculty/812

[4] The Genius Who Invented Economics Blogging Reveals How He Got Everything Right And What’s Coming Next: Joe Weisenthal – https://www.businessinsider.com/bill-mcbride-of-calculated-risk-2012-11

[5] About Calculated Risk: https://www.calculatedriskblog.com/p/about-calculated-risk.html