Whisper numbers first came about in the early 1990s as a reference to earnings expectations passed along to large investors and institutional traders. During a period of exceptionally high earnings growth, the combination of naturally conservative analysts, especially on the sell-side, and the expanded research departments by institutions primarily on the buy-side, resulted in whispered expectations that were consistently more accurate than consensus earnings estimates. Consequently, by 1997, whisper numbers had replaced consensus estimates as the market's true expectations for earnings and Michael Moe at Montgomery Securities said at the time "[t]he whisper... becomes in effect more important than the published number. The whisper is increasingly becoming the true expectation."

There are some studies, such as one published by professors at MIT, Northwestern, and Duke that imply analysts publish their more conservative estimates to protect their jobs and verbally pass along or cite the more aggressive whisper numbers so they cannot be tied to them if they turn out to be wrong. This applies to negative expectations as well and other studies indicate analysts are more timid with their negative expectations for fear of alienating management. Therefore, contrary to many beliefs, conservative estimates do not always mean they are low for an easy hurdle and, in fact, research published by professors at Michigan, Indiana, and Purdue found "whisper forecasts are, on average, more accurate than First Call forecasts and that their superior accuracy does not depend on whether they are greater or less than their First Call counterparts".

Besides the inherent conservatism among financial professionals - including analysts - another reason whisper numbers are a better proxy of the market's expectations is simply timing. A lot of calculations go into determining an earnings estimate and much of that data comes in late in the quarter. It is unreasonable to expect an analyst to repopulate the data in a 25-page spreadsheet with each new data point for each company he or she covers, not to mention issue a new report and submit a revised estimate to the reporting agencies. However, it is common to see the same analyst mention in a report or in another form that there is upside or downside risk to estimates. For example, a July 9, 2010 research report by Bank of America-Merrill Lynch, analyst Kuni M. Chen said the consensus estimate was $0.15 per share Alcoa (AA) but that he believed estimates would move closer to breakeven. It is equally common for analysts to tell clients expectations on earnings that differ from their published estimate after new data points come in. Contrary to some claims, this is not a violation of any disclosure or securities laws and, in fact, it is their job to tell their clients their expectations.

This timing issue is also the reason why whisper numbers do not materialize until a few weeks or closer to a company's earnings release. Still, the previously mentioned report by professors at the Universities of Michigan, Indiana, and Purdue stated that timing only partially explains the accuracy and the market's reliance on whisper numbers.

As the Internet gained in popularity as a medium for information while high-growth stocks rallied during the dot.com bubble of the late 1990s, whisper numbers gained household recognition. Unfortunately, since whisper numbers were not defined in the financial text books, less informed individuals and neophyte financial reporters that did not understand whisper numbers began looking for different, often easier, explanations for why stocks moved on expectations other than published estimates. The most popular explanation was that it was inside information leaked to analysts from corporate insiders and then only given to a research firm's favorite clients. This explanation even became so widespread that Regulation FD was passed to prevent company management from giving information to analysts that they didn't also give to the public. Still, it probably isn't a coincidence that there are corporate insiders arrested nearly every year for insider trading or selling information to a hedge fund, but we've never heard of a single insider arrested for giving secret information to an analyst. Likewise, even after Regulation FD, every earnings season it is hard not to hear analysts and traders discussing whisper numbers.

Another misconception promulgated by some is that a whisper number is the sentiment of the individual investor. Since the day EarningsWhispers went online in 1998 we have taken polls of our visitors' earnings expectations but we have specifically stated it was not used for gathering a whisper number for the simple reason that sentiment is not a whisper number. Shortly after our launch, the venture capital firm Internet Financial Network inquired about taking an equity stake in EarningsWhispers. However, when they informed us that their intentions for the capital were not to expand our research but rather to create a community for individual investors to provide their expectations, we declined the offer. The firm subsequently bought another ".com" company, but several years later after failing to produce the same results, Internet Financial Network then tried to sell the company to us. They have since cut their losses and moved on.

In fact, every single research we've ever seen about sentiment, especially of the individual investor, is that it is only valuable as a contrarian indicator. Furthermore, research published by the universities of San Jose State and Pepperdine found polling of individual investors resulted in estimates that were less accurate than consensus forecasts, which is the opposite of all studies of real whisper numbers, and that trading on polled expectations did not provide the positive returns that other studies found for real whisper numbers.

The simple reality is that whisper numbers come from research analysts, either on the sell-side or the buy-side, and they are used by professional traders, institutional investors, and hedge funds. They most commonly represent the expectations of the large investor, or the "smart money", and typically are the number that a company needs to exceed for these investors' models to change and buy more of the stock or begin selling on a miss. Because of the misuse of the term "whisper number" by so many on the Internet and in the media, the U.S. Patent office has awarded us a registered trademark to indicate that an Earnings Whisper ® number is a real whisper number.