Blog · 2026-10-30 · Ashutosh Agarwal

The most comprehensive short-signal skills library: What generic red-flag checklists miss

A backlog, a jump in cash flow or an auditor change can mean opposite things in different industries, so Matterfact's short-signal skills give our research agent the industry-specific tells, primary records and short-killers that a specialist short seller checks.

TL;DR

• The same red flag means different things in different industries. A record backlog can hold suspended rigs at an offshore driller, unexercised options at a government consultant, and orders that are firm only 90 days out at a truck maker.
• The best short signals in an industry appear on no generic checklist. In offshore drilling the tell is new contracts signing below the dayrate in the backlog; in staffing it is a public jobs series that turns quarters before company revenue.
• The agent reads the primary records each industry produces, from fleet status reports and mine-safety orders to utility rate dockets and auto-loan trust reports. Every case and rule in the skills was checked against such records, and what could not be verified was cut.
• Every skill names the short-killers before the thesis: consolidation bids, controlling shareholders, regulated floors, state rescues and a reopened equity window.
• The agent assists the analyst; it does not replace the borrow desk or the judgement. It has no borrow-fee data, works on delayed market data and never calls a company a fraud without a finding.

Between 2014 and 2016, two US companies that run payroll and benefits on behalf of their clients sent a generic short screen the same alarms. Each took a sharp, unexpected hit to a cost it carried for clients. Each then filed a periodic report late. Each changed auditors. At Barrett Business Services, a restatement and an SEC order followed. At TriNet, the stock closed at its low the day after its late-filing notice and rose more than tenfold over the next eight years.

A specialist would have told them apart from the filings themselves. Barrett's auditor withdrew its review of an interim quarter, citing information that an illegal act "has or may have occurred". TriNet's late-filing notice cited weaknesses in its IT controls and said it expected no restatement. A checklist that counts red flags scores the two companies the same. A short seller who covers the industry reads what each flag says.

That gap, between counting red flags and reading them the way an industry specialist does, is what Matterfact's short-signal skills close. They give our research agent a specialist short seller's playbook for every sector, with industry-level depth: what each metric means in that industry, which tells appear on no generic list, which primary records settle the argument, and which bids, rescues and regulators can kill a correct short. Every case and rule in them was checked against primary records, and whatever could not be verified was cut.

This article sets out why generic red flags misfire, what industry-specific signals look like, and which records the agent reads. It then works two historical cases end to end, one in offshore drilling and one in staffing, and closes with the short-killers the skills check first and the limits that remain.

heading: What we cover
The same red flag means different things in different industries
The best short signals in an industry appear on no generic checklist
The agent reads the primary records each industry produces
In offshore drilling, the thesis was right and the path was painful
In staffing, the same red flags led to opposite outcomes
Every skill names the short-killers before the thesis
Case study: running the agent on a live name (coming)
The agent assists the analyst; it does not replace the judgement
Questions to put to any short-research process

The same red flag means different things in different industries

Short screens are built from red flags that sound universal: revenue slowing, receivables growing faster than sales, an auditor change, insider selling, a falling backlog. Each is a fair question to ask. None is an answer, because the number behind it is defined, booked and dressed up differently in each industry.

Generic signals are also crowded. Published return predictors lose about 58% of their returns after publication,¹ and across 162 documented anomalies the long-short return nets to about zero once stock-level borrow fees are paid.² A short that rests on a factor screen alone is rarely worth the borrow. What still works is reading what a screen cannot: the definitions, footnotes, exhibits and public records specific to one industry.

A record backlog can hide the problem it seems to rule out

Take the backlog. A rising backlog usually reads as visibility, and a falling one as the start of a short. Across six industries the same headline means six different things, and in several of them it can hide the very problem it seems to rule out (Exhibit 1).

Exhibit 1
"Record backlog" means something different in each of six industries, and in several it can hide the short.

"Record backlog": the same headline, read by six industry skills

Note: Examples are historical disclosures, shown to illustrate how each industry defines backlog; they are not views on the companies today.
Source: Company 10-K filings and earnings releases as cited; Matterfact short-signal skills

The drilling row shows why the definition matters to a short. A suspension stops revenue while the contract stays in backlog. A short that waits for backlog to fall misses the revenue gap, and a short on the gap can be squeezed when the customer reactivates the rigs and adds the suspended days to their terms. The skill separates suspended rigs from working ones before it scores either.

Cash flow, auditor changes and inventory read in opposite directions too

The pattern repeats across the other staples of a forensic checklist (Exhibit 2).

Rising operating cash flow is the classic quality test, and in some industries it is the warning. Staffing firms pay temporary workers weekly and are paid by clients in 30 to 60 days, so a business in decline releases cash: free cash flow and buybacks rise in the bad years, and the stock can bottom on its highest free-cash-flow yield. A homebuilder that stops buying land reports record free cash flow while it shrinks. At auto dealers, the "adjusted" operating cash flow that companies present includes floorplan borrowing.

An auditor change is a top-tier red flag in every forensic framework, and read alone it misleads both ways. A rotation between two large firms with no disagreements and no reportable events is routine. A change forced by a merger can still list a material weakness. And a letter that opens with "no disagreements" can go on to cite a reportable event, which is the part that matters.

An inventory or receivables build can be a pre-buy, or it can be invisible. Distributors sell early-buy programmes on extended terms before the season, so inventory and receivables rise ahead of a strong quarter, not a weak one. Powersports and boat makers are paid by floor-plan lenders within days of shipment, so a receivables screen reads clean while dealers overstock.

Exhibit 2
Strong cash flow, an auditor change and an inventory build each read differently by industry.

"Operating cash flow jumps": usually read as earnings quality

"The auditor changes": usually read as a top-tier red flag

"Inventory or receivables build": usually read as demand falling short

Note: Examples are historical disclosures, shown to illustrate how one signal reads in different industries; they are not views on the companies today.
Source: Company 10-K and 8-K filings as cited; Reuters (Pemex supplier payments, 2024); Matterfact short-signal skills

Insider selling has the same problem in a different form. Until March 2026, directors and officers of foreign private issuers filed no Form 4s at all, so a generic insider screen read their silence as a clean record. Two footwear brands listed in New York saw their directors and officers file their first Forms 3 on 18 March 2026, when the Holding Foreign Insiders Accountable Act brought them under Section 16 reporting. A large holder who is neither a director nor an officer still files none, so the skill follows those holders' sales through resale prospectuses, Form 144 notices and ownership filings instead.

For a short book, these misreadings cost in both directions. A screen that scores every auditor change or cash-flow jump the same way fills the pipeline with false positives, and it waves through the shorts whose problem sits inside a definition.

The best short signals in an industry appear on no generic checklist

Correcting the generic flags is half the value. The other half is the signals a generic list does not contain at all, because they live in documents and data series specific to one industry.

Offshore drilling shows both halves (Exhibit 3). Every generic check on the left has a more precise version on the right, and the precise version is usually where the short is found or killed.

Exhibit 3
In offshore drilling, every generic check has a sharper specialist version, and most sit in documents a screen never reads.

Source: Matterfact short-signal skill for oil and gas drilling

The lead signal is the fixture: the dayrate and term on each new contract, which offshore drillers publish rig by rig. When new fixtures for comparable rigs come in below the average dayrate in the backlog, revenue per day steps down as old contracts roll off, months before reported revenue turns. Land drillers work differently. Their short contracts reprice within quarters, so the skill watches guided margin per day roll over while the guided rig count holds. A land fleet, a floater fleet and a jackup fleet that depends on one national oil company share an industry label and little else, so the skill splits them before it scores anything.

Other industries have tells of their own:

• Staffing turns with a public data series. The national temporary-help employment series published by the Bureau of Labor Statistics leads payroll employment, and a staffing firm's organic growth, adjusted for billing days, tracks it. Permanent-placement fees, which are almost all gross profit and carry no backlog, turn a quarter or two before temporary hours, and clients cut hours before they cut heads. In the last cycle the series peaked in March 2022. The US professional-staffing leader's revenue then fell from $7.24bn in 2022 to $5.38bn in 2025, and its shares fell about 82% from their 2022 closing high to a low in March 2026, then doubled as the series stabilised. The series gave both the entry and the exit.
• Gold miners are tested in the mill. Each quarter's processed head grade, set against the reserve grade of the ore the mine plan scheduled, tests the reserve model in the plant. A shortfall that lasts three quarters or more, explained each time as "sequencing" or "dilution", points to a reserve cut and a higher depletion rate. Cost per ounce has to be rebuilt at a fixed gold price, fixed by-product prices and fixed exchange rates to separate operations from arithmetic. And the reserve price is a choice: raising it turns lower-grade resource into reserves, lengthens mine life and lowers depletion per ounce, so earnings rise with nothing changed underground.
• Auto dealers make their money below the revenue line. New vehicles are a large share of a dealer's revenue and a small share of its gross profit, so a revenue miss on units can sit beside a gross-profit beat from parts, service, and finance and insurance. The short lives in gross profit per unit by line, in the chargeback reserve behind finance and insurance income, and at buy-here-pay-here lenders in charge-offs that lag through loan modifications and repossessions.
• Integrated oil companies can report lower volumes because prices rose. Under production-sharing contracts the oil company recovers its costs in barrels, so when prices rise it needs fewer barrels and its reported entitlement volume falls. Chevron's 10-K names these "entitlement effects", and ExxonMobil's release carries entitlement lines in its volume bridge. The surprise also tends to arrive before the print: ExxonMobil files an 8-K of market and planned factors after each quarter end, so a short placed for results day trades what is left.
• IFRS miners can write impairments back up in a rally; US GAAP miners cannot. IAS 36 lets an IFRS filer reverse a past impairment, other than goodwill, when its long-term price rises. Kinross reversed $116.1m at one mine in the fourth quarter of 2025 on higher gold-price estimates. Comparing earnings across the two accounting regimes without stripping reversals invents a gap between peers.

None of these appears on a generic checklist, and each one changes a thesis. The bar we hold the skills to is that a short seller who specialises in the industry should find nothing generic in them and nothing wrong.

The agent reads the primary records each industry produces

Short debates are settled by records, not arguments. In 60 documented short campaigns from 1999 to early 2026 that we studied, the decisive evidence was usually a primary record that contradicted the company: a registry filing, an auditor's resignation letter, a regulator's record, or a customer that did not exist. Confirmed cases rested on about 2.6 independent types of evidence on average, against 1.8 for the shorts that failed. Evidence that checks a claim against an independent record confirmed most often: customer checks in 82% of the cases that used them, auditor signals in 83% and regulator records in 78%, against about half for filings and financial analysis alone.

Those counts are directional, because famous campaigns are mostly ones that worked. The lesson is still clear: breadth of independent evidence matters, and the most useful records differ by industry. A drilling short is settled by fleet status reports and the offshore regulator; a gold short by technical reports, mine-safety orders and the tailings register; a utility short by state commission dockets (Exhibit 4).

Exhibit 4
The record that settles a short debate is different in every industry.

Primary records Matterfact's agent reads, by industry

• Oil and gas drilling
  • Fleet status reports: Rig-by-rig customer, dayrate, term, options and survey downtime: the pricing record.
  • Offshore regulator data: Incidents of non-compliance, issued to contractors as well as operators since October 2011.
  • State drilling records: North Dakota's daily rig list names each rig, its contractor, operator and well.
• Pipelines and midstream
  • FERC dockets: Certificate orders, rate cases, complaints, capacity turned back, tariff filings.
  • FERC Forms 2 and 6: Cost of service and earned return for each regulated pipeline subsidiary.
  • PHMSA: Incidents, notices of probable violation and corrective action orders.
• Gold miners
  • Technical reports: Reserve price, grade and mine plan, filed as SEC exhibits.
  • MSHA mine data: Inspections, citations, withdrawal orders and accidents at US mines, by mine.
  • Global Tailings Portal and ICSID: Each tailings facility's consequence class; arbitrations against host states.
• Electric utilities
  • State commission dockets: Rate cases, testimony, settlements, wildfire-mitigation plans, arrears reports.
  • FERC Form 1: Plant, operating cost and rate-base inputs by operating company.
  • NRC records: Daily reactor power, event notifications and oversight findings.
• Defense contractors
  • Daily contract announcements: Every Defense Department award of $7.5m or more, with obligated amounts.
  • USAspending: Obligations and de-obligations by recipient.
  • Program cost reports: Selected Acquisition Reports and GAO assessments of cost growth against baseline.
• Airlines
  • DOT consumer reports: Complaints, mishandled bags and enforcement penalties by carrier.
  • Airworthiness directives: FAA and EASA inspections and groundings by engine or aircraft type.
  • BTS traffic and Form 41: Route-level traffic and capacity, and financials by carrier.
• Auto retail
  • Auto-loan trust reports: Servicer reports and loan-level data on how each lender's loans perform.
  • State dealer-licensing records: Licence suspensions over title and registration delays.
  • Used-vehicle value index: Wholesale prices behind used gross profit and inventory marks.
• Restaurants
  • Franchise Disclosure Documents: Performance claims, outlet openings and closings, departed franchisees, litigation.
  • Outbreak notices: CDC and FDA multi-state outbreaks and the chains implicated.
  • City health inspections: Critical violations, unit by unit.
• Staffing
  • BLS temporary-help series: National temporary employment and hours: the leading series.
  • Wage and Hour enforcement: Back-wage findings against staffing firms and professional employer organisations.
  • IRS certified PEO listings: Active, suspended and revoked certifications of professional employer organisations (PEOs).

Source: Public agency, regulator and company records named in Matterfact's short-signal skills

Three rules govern how the agent uses these records.

It pulls live data at run time. The skills carry no current figures that go stale: no dayrates, margins, multiples, short-interest levels or prices. Their thresholds are relative, set against the company's own history, its peers or a turn in the second derivative, and every level is fetched when the question is asked, with its as-of date.

Every case and rule in the skills was checked against primary records. That means SEC filings and exhibits, regulator databases and court dockets, or a major outlet where no record exists. Historical cases that could not be verified were left out rather than marked, and data sources that could not be confirmed are flagged as unverified rather than presented as fact.

It grades evidence and labels claims. Each item in a short case is marked as a primary record, a figure computed from one, third-party reporting, or an unverified allegation. A published short report counts as a list of allegations until each claim is checked against a record, and ten articles repeating one claim count once. The agent never calls a company a fraud, or a person a fraudster, unless a court, a regulator or the company itself has established it, and every allegation carries who made it, when, the company's response and its current legal status.

In offshore drilling, the thesis was right and the path was painful

The first worked case is Seadrill, an offshore driller, from 2014 to 2022 (Exhibit 5). It is the industry's signature financing short, and it shows why the skill's work does not stop when the evidence is in.

Exhibit 5
At Seadrill, the financing evidence was public by 2014, yet the shares jumped more than 55% in a morning in 2016 before the first bankruptcy in 2017.

Seadrill, 2014 to 2022: an offshore driller's financing short

Note: A historical case from public filings and reporting, shown to illustrate how the skill reasons. Not investment advice.
Source: Seadrill 6-K filings (2017, 2022); Maritime Executive and The Motley Fool (26 November 2014, 4 March 2016); gCaptain (2018)

Three things in the trace are specific to the industry, and a generic process would have missed each one.

The signal was a pattern, not a ratio. Payouts kept up at peak dayrates while a shipyard orderbook was still unpaid is the offshore driller's signature financing short. A generic debt screen sees debt. The skill sees debt plus yard instalments plus a payout, set against when the backlog's cash arrives, and it dates the maturity that forces the issue.

The squeeze came from the structure that later failed. A controlling shareholder's rescue talk lifted a heavily shorted stock by more than half in one morning, and the company still filed the next year. A majority holder is no guarantee of a rescue either: Diamond Offshore filed for Chapter 11 in April 2020 while its largest shareholder held about 53% of its shares. The skill does not treat a large holder as a floor or a reason to cover. It reads what the holder has committed in writing and sizes the short to survive the squeeze while it waits for the forcing date.

The industry has a takeout problem. The listed offshore fleet has been consolidating through mergers since 2018, many of them all-stock, and bids arrive in downturns and recoveries alike. Transocean agreed in February 2026 to buy Valaris in an all-stock deal at a 31.6% premium to the prior close; the deal is pending as of 30 October 2026. FINRA data show Valaris short interest falling from about 11.0m shares at the end of September 2025 to about 6.1m in mid-January 2026, so much of the short base had covered before the bid. A merger of two levered fleets is no floor either: Ensco and Rowan combined in an all-stock deal in April 2019, and the merged company, by then named Valaris, filed for Chapter 11 in August 2020.

The skill's conclusion in this industry is often not "short" but "not yet" or "not like this". When a Matterfact agent swarm researched Borr Drilling, a levered jackup owner, it valued the shares at less than half their price and still advised against a short, because the same model valued four listed peers below their prices as well, which said more about the model than about Borr. The case is in An Agent Swarm on Borr Drilling. For expression, the skill times financing shorts to a maturity or yard payment, and pairs fixture shorts within one rig class to strip out oil prices and sentiment, never a jackup fleet against a floater fleet.

In staffing, the same red flags led to opposite outcomes

The second worked case returns to the two companies in the opening. Barrett Business Services and TriNet both run payroll and benefits for client businesses as professional employer organisations, or PEOs. Between 2014 and 2016 each gave a generic screen an earnings break, a late filing and an auditor change (Exhibit 6).

Exhibit 6
Barrett and TriNet both filed late and changed auditors; Barrett restated and settled with the SEC, while TriNet's stock rose more than tenfold from the day after its late-filing notice.

Note: The claims break at TriNet paid shorts at once; the short failed for those who held past the late filing. Historical cases from public filings, shown to illustrate how the skill reasons.
Source: SEC Release 33-10557 (2018); Barrett 8-K filings (29 October 2014, 9 November 2015, 9 March 2016); TriNet 8-K filings (3 August 2015, 17 March 2016, 10 May 2016) and late-filing notice (29 February 2016)

The difference was in two documents a generic screen does not read. The first is the late-filing notice itself, whose narrative section says why the report is late: a notice that cites errors in prior-period statements is a different signal from one that cites control weaknesses and expects no restatement. The second is the auditor's letter on any change, where the paragraph on reportable events matters more than the line on disagreements. The staffing skill reads both, and it scores a controls-only delay at a large PEO lower than one that cites prior-period errors.

It also knows what drives the margin in this business model. A PEO bills its clients a fixed rate for workers' compensation and health benefits and carries the claims itself through large-deductible and minimum-premium arrangements. Under-reserving flatters margin until an actuarial review or the auditor forces a charge. So the skill tracks workers' compensation expense against payroll as the client mix changes, favourable prior-year reserve development as a share of pre-tax income, and the carrier agreements filed as exhibits. That is where Barrett's problem sat, and why its first charge was not its last: each step repriced the stock, and the restatement came sixteen months after the reserve charge.

TriNet's case carries the opposite lesson. The claims break paid at once. The late filing, the material weaknesses and the auditor change then looked like Barrett's on a screen, but they cited controls, not prior-period errors. A short pressed on the second set of flags faced a stock that went on to rise more than tenfold.

Every skill names the short-killers before the thesis

A correct short can still lose money, and each industry has its own ways of making that happen. The skills list them before they score the thesis, because a short that ignores its industry's killers is not a thesis (Exhibit 7).

Exhibit 7
Each industry has its own way of killing a correct short: a bid, a rescue, a regulator or a reopened equity window.

Short-killers the skills check first, by industry

• Oil and gas drilling
  • Consolidation bids: Offered in downturns and recoveries alike; Transocean agreed to buy Valaris at a 31.6% premium in February 2026.
  • Controlling shareholders: Rescue talk can squeeze a crowded short even when no rescue follows.
  • The check: Size as if a bid can arrive in any quarter; after an all-stock bid the short is a view on the acquirer.
• Gold miners
  • The gold price: A rally lifts the weakest producers most and overwhelms cost, grade or reserve shorts.
  • Takeout floors and parents: A stressed single-asset miner is often a target, not a zero; a majority parent can buy out minorities on its own timetable.
  • The check: Express the short against gold or a peer basket, and size for the beta.
• Electric utilities
  • The regulated floor: A commission must allow a return on prudent investment.
  • State rescues: California's wildfire fund and Hawaii's 2025 legislation came within months to two years of the losses.
  • The check: A short that needs the equity to go to zero must explain why the state will not act.
• Airlines
  • Takeovers: Small carriers are bought for slots, gates, routes and loyalty members, often at several times the prior price.
  • Government support: In 2020 US carriers received payroll support and loans with warrants issued to the Treasury.
  • The check: A cheap carrier with assets a network carrier wants is a dangerous short.
• Homebuilders
  • Takeout floor: Japanese housing groups and a US conglomerate paid cash premiums for listed builders three times between 2024 and 2026.
  • Rates: A fall in the 10-year yield lifts the whole group in a day.
  • The check: Pair a land-heavy, levered builder against a net-cash builder to strip out rates.
• Auto retail
  • The equity window: In 2023 an online retailer swapped unsecured notes for secured pay-in-kind notes and filed a $1bn at-the-market programme on the same day; the equity recovered.
  • Regulators who change course: The FTC's dealer fee rule was vacated in January 2025 and withdrawn in February 2026.
  • The check: A cash-burn short needs the equity window shut, not just the burn.
• Staffing and HR software
  • Takeouts: Five listed HR software, executive-search and screening companies were bought or taken private between 2024 and February 2026.
  • Counter-cyclical buybacks: Receivables released in a downturn fund buybacks just as revenue falls.
  • The check: Count buybacks as a counter-signal only when earnings, not working capital, fund them.
• Integrated oil
  • The write-down marks the low: ExxonMobil's fourth-quarter 2020 impairment came after a fall of about 41% in the year, at the turn of the cycle.
  • Supply shocks: A shock can cut a major's own barrels and still lift the stock.
  • The check: Score impairment risk before the plan update that resets the price deck, not after the write-down.

Source: Company filings, FTC and court records, and press reports as cited in Matterfact's short-signal skills

Two habits run through every skill:

• Crowding is risk, not evidence. High short interest, the daily short-volume ratio and fails-to-deliver never add conviction. They feed sizing and squeeze risk. A drilling short opened when about 30% of a large driller's float was sold short, in October 2016, had to survive a rally after the election and OPEC's output deal before the stock set lower lows in 2017.
• Insider buying and a record that rebuts the thesis count most against a short. Open-market purchases by the chief executive or finance chief, or a primary record that contradicts a key claim, weigh most. A stake-building strategic holder or takeover interest, buybacks funded by free cash flow and a stronger replacement auditor or finance chief also count against it, and the skill searches for them as carefully as for red flags.

Case study: running the agent on a live name (coming)

We are running the agent on a live name now, and will publish the full run here: the signals that fired, the records it pulled, the short-killers it checked and the verdict it reached, with every figure dated.

The agent assists the analyst; it does not replace the judgement

The skills give the agent a specialist's knowledge. They do not give it everything a short seller needs, and the limits are worth stating plainly.

There is no borrow-fee data. No public source gives borrow fees, utilisation or lendable supply, and the agent never estimates them as if observed. Option prices are no substitute for a quote. A thesis that passes every other test reads "actionable, pending borrow" until the analyst's prime broker confirms availability and cost, and for a hard-to-borrow name that quote can change the answer.

Market data is delayed. FINRA short interest is collected twice a month and published about seven business days after the settlement date. Option quotes are about 15 minutes old. Form 4 insider filings arrive within two business days, institutional holdings 45 days after quarter end, and some agency data lag by months or years. The agent states each lag with its figure, so a reader always knows how old a number is.

The agent produces research, not instructions. It does not tell anyone to trade, and it does not draft campaign copy that presents allegations as fact. Where the evidence can only come from a site visit, a channel check or a conversation with a former employee, it says so and leaves that work to the analyst.

Judgement stays with the analyst and the PM. The agent can read every footnote, docket and fleet table in a coverage universe and keep re-reading them as filings arrive. Deciding whether a mechanism is real, whether the catalyst will come in time and how much to risk on it remains the investor's call.

Questions to put to any short-research process

Short sellers who last have always read what others skim: the exhibit behind the headline, the paragraph after "no disagreements", the record that contradicts the filing. Matterfact's short-signal skills put that reading, industry by industry, into an agent that does not tire of footnotes. For heads of short research, three questions test whether a process reads that closely:

• Which red flags in our screen mean something different in the industries we cover? Start with backlog, cash flow and auditor changes.
• When an analyst writes "record backlog" or "strong free cash flow", who checks what the number counts?
• Before a short goes on, has someone written down the bid, the rescue, the regulator or the equity raise that could kill it?

FAQ

What is an industry-specific short signal?

A sign that a company's shares are mispriced to the downside that only makes sense in its industry's own terms: new offshore rig contracts signing below the dayrate in the backlog, a staffing firm's revenue lagging the national temporary-help series, or a gold mine's processed grade running below its reserve grade. Generic red flags such as rising receivables or an auditor change are inputs too, but their meaning depends on the industry.

Why do generic forensic accounting red flags produce false positives?

Because the same number is defined and booked differently from one industry to the next. Operating cash flow rises in a staffing downturn as receivables run off; a dealer's "adjusted" cash flow includes floorplan borrowing; a floor-plan lender pays a powersports maker within days, so its receivables look clean while dealers overstock. A screen that scores these the same way flags the wrong companies and misses the right ones.

How does Matterfact's agent build a short case?

It identifies the company's industry, applies the short-signal skill for it, and pulls the primary records that industry produces: SEC filings and exhibits, regulator databases, court dockets and public data series. It grades each piece of evidence, tests what the share price already assumes, maps the catalysts, and lists the short-killers and the borrow and squeeze checks before giving a verdict, a conviction level and a timing view.

Does Matterfact provide borrow fees or utilisation?

No. No public source carries borrow fees, utilisation or lendable supply, and the agent never estimates them. Short interest comes from FINRA with its publication lag, and a thesis that passes every other test is marked as pending borrow until the analyst's prime broker confirms availability and cost.

Will the agent call a company a fraud?

No. It describes what the records show and what they are consistent with. A company is called a fraud only when a court, a regulator or the company itself has established it, and every third-party allegation is reported with who made it, when, the company's response and its current legal status.

Historical cases above come from public filings, regulator releases and press reports, with statuses as of 30 October 2026. They illustrate how the skills reason and are not investment advice or a recommendation to buy or sell any security.