Equity Strategy · 8 min read · 2026-04-28
The 9 red flags that predict an earnings restatement.
Restatements are not random. They are the tip of an iceberg of accounting-quality decay that is typically visible in the filings two to four quarters earlier.
Restatements are the longest predictable shock in equities.
Material earnings restatements produce a median same-day drop of 12 percent and continue to underperform for the next twelve months. The accounting-quality decay that produces the restatement is typically observable in the filings two to four quarters before the event. The signals are public; the assembly is the work.
The nine indicators
The nine red flags that predict a restatement.
Each is verifiable in the 10-K, 10-Q, proxy, and the 8-K Item 4.02 disclosures. The screen runs quarterly on every actively held position.
SOX 404 material weakness disclosure
Source: 10-K Item 9A
Material weakness in internal control over financial reporting is disclosed under SOX. The disclosure is the most direct single tell of accounting-quality decay.
Beneish M-Score above -2.0
Threshold: M > -2.0
Messod Beneish's M-Score combines eight ratios into a single fraud-probability metric. M > -2.0 historically captures roughly 70% of subsequent restating firms.
Accruals-to-assets ratio in top quintile
Threshold: > 80th percentile
High accruals relative to assets indicate aggressive revenue or expense recognition. The Sloan accrual anomaly remains one of the most-replicated signals in academic finance.
Audit committee chair turnover in trailing 24 months
Pattern: chair departure
Audit committee chairs leave companies that are heading toward problems. The departure precedes the restatement by 12–24 months in the historical sample.
External auditor tenure < 3 years OR > 25 years
Threshold: out-of-norm tenure
Very short tenure suggests recent disagreement with prior auditor. Very long tenure suggests entrenched relationships and reduced skepticism. The middle is the safer band.
Late filing of 10-K or 10-Q in last 24 months
Pattern: NT-10K or NT-10Q
Late filings (Form NT) signal accounting issues that prevented timely close. Repeat NT filings are a strong restatement precursor.
Free cash flow < net income for 4 of last 8 quarters
Threshold: 4/8 quarters
Persistent gap between earnings and cash flow indicates accruals-driven earnings. The longer the gap persists, the more it tends to reverse.
CFO turnover in trailing 24 months
Frequency in restating firms: 41%
Forty-one percent of restating S&P firms had CFO turnover in the 24 months prior. New CFOs sometimes restate to clear inherited issues; sometimes they create them by changing methods.
Inventory growth materially outpacing sales growth
Threshold: 2x sales growth
Inventory growth significantly above sales growth signals demand softness or accounting problems with cost-of-goods-sold timing. A 2x gap for 3 quarters is the threshold.
Why restatements are predictable but rarely predicted
Material earnings restatements affect roughly 1 percent of U.S. listed companies per year, and the average restating firm has been issuing problematic filings for 18 to 30 months before the announcement. Despite the long lead time, the restatement is almost always a surprise to the holder base. The signals exist; the discipline to assemble them does not.
The signals fall into three broad categories: structural (auditor and audit-committee composition), behavioral (filing timeliness, executive turnover), and statistical (accruals quality, FCF-to-NI gap, inventory growth). The composite across the three categories produces a meaningful predictive signal. Any single signal in isolation is too noisy to act on; the joint distribution carries the information.
The Beneish M-Score, properly applied
Messod Beneish's 1999 model combines eight financial ratios into a single fraud-probability metric. The model was trained on a sample of known accounting-manipulation cases and has been replicated extensively in subsequent literature. An M-Score above -1.78 (Beneish's original threshold) flags approximately 76 percent of subsequent earnings restaters in out-of-sample tests, with a false-positive rate around 17 percent.
The model is simple to compute. The eight inputs — DSRI, GMI, AQI, SGI, DEPI, SGAI, LVGI, TATA — are all in the 10-K. A spreadsheet implementation takes 30 minutes to build once and produces a quarterly score per position thereafter. The discipline is to score every active long position quarterly and flag scores above the threshold for additional review.
Accruals quality and the Sloan anomaly
Richard Sloan's 1996 paper on the accrual anomaly remains one of the most-replicated findings in finance. Companies in the top quintile of accruals-to-assets underperform companies in the bottom quintile by approximately 10 percentage points per year on a risk-adjusted basis. The mechanism is that high accruals indicate aggressive earnings — earnings that are produced by accounting choices rather than cash flow — and those earnings tend to reverse.
Restatements concentrate disproportionately in the top accruals quintile. The screen overlaps with the M-Score; using both produces tighter convergence than either alone.
Internal-control disclosures and audit-committee turnover
SOX Section 404 requires management to assess internal controls over financial reporting. Material weaknesses are disclosed in the 10-K and 10-Q. A material weakness disclosure is the single most direct restatement precursor in the public-filings universe — companies disclosing material weaknesses subsequently restate at roughly 6x the base rate of the broader universe.
Audit-committee chair turnover is a softer but still meaningful signal. The chair has the most direct view into accounting-quality issues and is the executive most likely to leave when issues are not being resolved. Chair departures within 24 months of the eventual restatement appear in approximately 30 percent of cases.
What to do with the screen
Names firing five or more of the nine signals deserve a focused review of the latest 10-K and 10-Q. The review focuses on the SOX 404 disclosure, the auditor's opinion, the CFO's qualifications and tenure, and the accruals-quality ratios. If the focused review confirms the signal cluster, the position should be reduced or exited.
The strategy is risk management, not alpha generation. The base rate of restatements is low; the screen is designed to avoid the worst tail outcomes rather than to identify high-return positions. Avoiding a single 35% restatement-driven drawdown across a portfolio over a five-year holding period contributes meaningfully to net-of-risk returns.
Get the nine, every Monday.
Free weekly digest. The only U.S. equity letter that publishes a name only when nine independent signals align.
Common questions
Questions.
How does a restatement differ from a revision?
A revision is a correction to immaterial errors. A restatement is a material correction that requires re-filing the affected periods (the 'Big R' restatement). Big R restatements produce the price action; revisions usually do not.
Are all material weakness disclosures bad?
Most resolve without restatement, but they significantly elevate the risk. A first-time material weakness in an established company is a more concerning signal than a longstanding small-company control gap.
Can I run the M-Score myself?
Yes. The eight inputs are public in the 10-K. Spreadsheet implementations are widely available. The discipline is to compute it consistently across positions, not to trust occasional ad hoc scoring.
How often do clean firms restate?
About 1 percent of U.S. listed companies issue a material restatement in any given year. The screen targets the 5–10 percent of names that produce most of the restatements, dramatically reducing exposure to the tail.
Does the screen work for non-U.S. companies?
The accruals signals work universally. The U.S.-specific signals (SOX 404, NT filings, 8-K Item 4.02) require local equivalents. Restatements outside the U.S. are less timely and produce different price dynamics.
Should I short names that fire the screen?
Generally not. Borrow costs are high on candidate names, and timing is uncertain. The screen is most powerful as a defensive overlay, removing or reducing at-risk positions before the event.
One name. Sometimes weeks of silence. Always with conviction.
Free weekly digest. No credit card. Unsubscribe anytime.