The Best Tools for Investors to Detect Potential Accounting Fraud

A lonely accountant uses advanced internet based analytical models to detect fraud at a publicly-traded firm.

What are the best tools investors can use to detect potential accounting fraud at the firms in which they might invest?

That's not an easy question to answer. That's because accounting fraud is almost invariably an inside job, where most investors are on the outside looking in. Most investors simply don't have the access to information about the transactions that constitute the bulk of accounting fraud where it exists.

But for publicly-traded firms, investors can access their financial statements. If the scope and scale of potential accounting fraud at a firm is big enough, the signs of it can show up in them. Accounting professionals have developed tools to help them identify known signs of fraud within financial statements.

That brings us back to the beginning. What are the best tools investors can use to find the signs of potential accounting fraud in financial statements?

A 2021 paper by Messod Beneish and Patrick Vorst, who evaluated seven fraud prediction models. They sought to identify the tools that could successfully identify firms where real accounting fraud may be occurring, without triggering too many costly false positives in the process. Here's the list of tools they evaluated, which are identified by their creator(s) and the year the tool was introduced:

  1. Beneish (1999) M-Score
  2. Cecchini et al. (2020)
  3. Dechow et al. (2011) F-Score
  4. Amiram et al. (2015) FSD Score
  5. Alawadhi et al. (2020)
  6. Bao et al. (2020)
  7. Chakrabarty et al. (2020) ABF Score

The M-Score was developed by Messod Beneish, so as the analysis goes, we should recognize that he has skin in the game for the evaluation.

Let's cut to the chase and go straight to the conclusion to find out which tools the authors evaluated came out on top in their analysis and why they did:

We compare seven fraud prediction models that have been proposed in prior research. We find that the higher true positive rates in recent models come at the cost of higher false positive rates and that even the best models trade off false to true positives at rates exceeding 100:1. Indeed, the high number of false positives makes all seven models considered too costly for auditors to implement, even when we consider extreme subsamples where a priori firms’ management has higher incentives and/or ability to misreport. We believe this could explain audit practitioners’ apparent reluctance to use these models, despite the fact that models have nearly doubled their success at identifying fraud when compared to the initial models in Beneish (1997, 1999).

For investors, M-Score and the F-Score when used at higher cut-offs are the only models providing a net benefit when applied to the sample as a whole. We conjecture this occurs because the M-Score and the F-Score exploit fundamental signals that have been shown to predict future earnings and returns, and the main component of investors’ false positive costs is the profit foregone (or the loss avoided) by not investing in a falsely flagged firm. In addition, we find that most models are economically viable if applied to top or bottom quintiles of characteristics of firms in which managers a priori have greater incentives and/or ability to misreport.

At this point, we'll point out that we also have skin in the game, which is why the conclusion of this paper attracted our attention. Political Calculations has tool based on the F-Score fraud detection model: Using the F-Score to Detect Accounting Fraud. Meanwhile, a tool based on Beneish's M-Score model is also freely available in both spreadsheet and online tool formats.

Aside from having built a tool based on one of these potential accounting fraud prediction models, we'll recommend using either or both. It's hard enough as an investor to do proper due diligence to choose which companies you might invest in. If the potential for fraud is a concern, it's worth the time and effort to use the most effective tools to either rule it in or out of your portfolio.


Messod D. Beneish and Patrick Vorst. The Cost of Fraud Prediction Errors. The Accounting Review. DOI: 10.2308/TAR-2020-0068. [SSRN Preprint]. 30 December 2021.

Image Credit: Stable Diffusion DreamStudio beta "A lonely accountant uses advanced internet based analytical models to detect fraud at a publicly-traded firm."