A new research paper co-authored by professors at the MIT Sloan School of Management and Harvard Business School purports that even the most expert analysts overlook a number of off-the-income-statement items from the 10-K, and that by identifying these items investors can get a much better picture of a company’s real earnings. The paper, titled “Core Earnings: New Data and Evidence,” is discussed in an article in Institutional Investor.

Differentiating core-earnings items that come from a company’s main activities from non-core-earnings items stemming from a company’s sideline activities is the biggest challenge for analysts, the paper contends. That challenge becomes harder when accounting practices vary at public companies, such as adjusting the figures they’re required to report and directing investors to discount activities that the company thinks will be a one-off event.

Between 1998 and 2017, the average number of non-core earnings items reported in 10-Ks increased from 6 to 8. But half of those weren’t disclosed in the actual income statement, coming instead in footnotes, the cash-flow statement, or through discussions by management. Therefore, analysts and investors trying to get the whole earnings picture have to wade through a pile of information that might be sprinkled throughout the more obscure parts of the 10-K.

The professors found that 19% of net-income items stray from core-earnings activities, and created a model that adjusts net income for these items. That adjusted net income allowed for an 8.2% annualized return for the investors who purchase stocks that do well on the earnings-distortion scale and dump those with a lower performance. Therefore, the article concludes, putting in the time to read the footnotes of earnings reports and statements would be well worth the effort, and give investors who know about earnings distortion and are cognizant of the way companies manipulate earnings an edge.