Millions of people contribute meaningful information to online platforms on a daily basis, including reviews on Amazon, reviews on Wikipedia or even financial estimates on Estimize. The truth of the matter is the masses can be surprisingly more accurate than individual experts in almost any arena.
The notion that group judgement informs better decision making underpins the wisdom of crowds theory and the core philosophy of Estimize. Despite oodles of empirical evidence to support the efficacy of the framework, we consistently get asked why our users, comprising of buyside analysts, independent investors and academics, contribute estimates to the platform.
Answering the question involves dissecting two distinct issues. 1) the reason why market participants contribute; and, 2) the contributors’ rationale for providing reliable estimates and how Estimize ensures their quality.
In the first part of the series we look at the core question of why people contribute and some basic psychology behind crowdsourcing. That leads into a broader discussion about why the Estimize data persistently outperforms I/B/E/S with each passing quarter. We conclude the series with a look at how we curate the consistently accurate data to be more precise.
Why Do They Contribute
It’s well documented that certain biases influence the behavior of sell-side brokers and analysts publishing quarterly estimates. Instead of pursuing accuracy, analysts tend to sandbag estimates in order to maintain corporate access relationships for the firm’s investment banking arm. In other words, the directionality and magnitude of revisions follow a biased pattern throughout an average quarter, starting with far too high and ending far too low, to juice the “beat rate” relative to the consensus. The problem escalates when analysts exhibit herding behavior, thereby releasing forecast estimates similar to ones previously published by other analysts. After all, humans tend to mimic larger groups regardless of how woefully wrong or irrational it can be at times. Hence, the I/B/E/S data effectively misrepresents the market’s true expectations for both consensus estimates and dispersion on a quarterly basis.
Public goods markets for information, e.g. Wikipedia versus Britannica, recently supplanted single source information providers across a range of industries. Initially, many questioned the reliability and accuracy of Wikipedia given a lack of monetary incentives for contributing; however, empirical research suggests that Wikipedia exceeds Britannica in terms of timeliness and accuracy. Ongoing academic research on the topic proved that our growing list of participants at Estimize behave consistently to other crowdsourced groups like Wikipedia. Inherently, humans are driven by various non monetary incentives, many of which can be leveraged more effectively to build crowdsourced content and data sets.
Buy side analysts, portfolio managers and traders need to understand where their fundamental expectations fall in relation to their peers in order to know whether they have an outside view. The delta between their view and the market’s view, if correct, is their alpha. But to know that delta, you need to know what is currently priced into the market. This is why these professionals will call around to each other before an earnings report, in order to triangulate the “real” consensus from a non biased crowd. This behavior took place because the I/B/E/S data was an unreliable measure of the market’s true expectations. Before Estimize existed, these professionals would share Google Docs online, or attend idea dinners where this information was shared among a small group. In these cases of direct reciprocity there was little or no free ridership. But obtaining this information was arduous, you needed to have access and relationships with your peers, and you needed to build trust over a long period of time in their expectations. It was inefficient.
These professionals come to Estimize.com and are able to obtain the estimates and consensus from their peers for free only after they have entered their own estimate for a single quarter of a specific stock. The platform operates on full reciprocity, contributors give an estimate, they get estimates for that quarter. This significantly limits free ridership in the public goods market.
Non professional investors, industry experts, students and academics contribute to our data set for many of the same reasons, specifically their limited access to quality expectations. They also have limited access to buy side networks.
Estimize is a public goods market, which routinely performs surveys of its 40,000+ community of contributors. Their answers confirm each individual is incentivized to contribute to this market in return for peer data. This is a classic game theory exercise, and in the Estimize system, no individual is contributing to their detriment. Contributors are not at risk of being arbitraged, even if they are aggressive and eventually correct, because they can use a pseudonymous handle i.e. not providing their real name or name of their firm.
In the beginning a small percentage of people were willing to contribute to a fledgling data set that did not have much data. However, critical mass has been reached as the contribution rate of our community has increased more than 4X over the past two years. Participants find their contributions are rewarded with a commensurate amount of value. In essence, we have an incredibly healthy public goods market.
Members also participate to judge their own accuracy versus their peers. Ego is a heuristic that market participants operate under for many reasons, including internal validation of their accuracy or external recognition of their abilities. PMs will hold on to trades against their better judgement simply to “be right” even though the expected value of that decision is negative in dollars. Estimize scores and ranks analysts for each report and overall to facilitate contribution.
And if we want to get really philosophical here, we believe that some contributors simply fall into the bucket which drives much of the content creation on the internet. Humans simply find it irresistible to share their own views when they believe someone has said something on the internet that they don’t agree with. Laugh, yes, but we believe it drives a significant amount of high quality data for us and we continue to develop the platform in a way that promotes discord amongst the community in order to bring out this behavior.
Institutions can buy access to view our data set through Estimize.com if they want full access to all stocks and all quarters without contributing. Quantitative institutional traders can purchase access to our full data set for systematic trading without contributing. The community does not find this to be an issue as these paying institutions enable the platform to be free for contributors. Ultimately, Estimize, Inc. does not exist without paying clients.
Estimize has built a profitable business challenging incumbents such as Thomson I/B/E/S by building a market for information, not only in earnings estimates, but economic estimates and other forward looking expectations data sets under the same structure.
Photo Credit: Giulia Forsythe