30%+ alpha in five minutes? 8-K is now live at AkkruData.

8-K event filings are now live on Akkru. Give your connected agent an investment strategy, and it can take your idea from discovery to an initial backtest in about five minutes — with the evidence and calculations ready to inspect.

Our agent found a paper by Rouen, So & Wang (JFE, 2021) while researching fundamental investment strategies. The paper shows that markets can be slow to separate one-off accounting effects from a company's underlying earnings. This is where the agent's judgment matters: it has to read the filings and decide whether a write-down reflects an old investment mistake or a business that is still getting worse. That decision determines which companies belong in the backtest.

Give the agent the whole strategy.

The agent finds, reads, and judges.

Our search returned 12 event records across 11 companies in Akkru's current dataset. The filings explain why the same “impairment” label can mean very different things:

• Citi ↗ took a $726M goodwill impairment tied to the agreed sale of a Banamex stake. Citi described the impairment as capital neutral. That makes it a plausible cleanup case, subject to a closer look at the remaining business.
• Ford ↗ disclosed ~$11.5B in non-cash charges in Item 2.06. But Item 7.01 also described ~$5.5B in expected cash costs for the wider EV reset. The surrounding disclosure changes the investment question.

Akkru supplies the sections and sources. The agent applies your criteria, selects the companies, and follows your strategy through to the return calculations.

Then it checks the returns.

Holding rule: enter on the filing date of the first 10-K or 10-Q after the 8-K; exit on the filing date of the following 10-K or 10-Q. Use the closing stock price on each filing date.

The six stocks in our illustration averaged a price return of +32.3%, versus +1.9% for the S&P 500 over matching dates: a 30.4 percentage-point difference.

These are previously audited illustrative results, not the recorded output of the prompt above. A new run may select different companies. The difference is not an estimate of risk-adjusted alpha.

Five of six returns were positive. Ford's stock fell 12.8%; Albemarle and onsemi accounted for a large share of the average gain. The result gives you a hypothesis to investigate further. Investing real money still requires deeper due diligence, but Agent + Akkru can substantially reduce research time and costs while sparking new investment ideas.

This backtest covers S&P 500 companies only. We expect greater potential alpha among smaller-cap companies, but that hypothesis has not been tested here.

Methodology, sources, and full sample

Universe. Item 2.06 filings dated July–December 2025: 12 event records across 11 companies in the current dataset. Original filings and amendments may be represented together. Two IP events share a holding window and count once. AES's January 2026 event is outside the window.

Calculation. Entry is the filing date of the first original 10-K or 10-Q after the 8-K disclosure; exit is the filing date of the following original 10-K or 10-Q. Returns use the closing stock price on each filing date. Price return = Exit close / Entry close − 1. Compare each trade with the S&P 500 price index over identical dates, then take the simple average of the unrounded returns. These returns cover different holding periods; their average is not a compounded portfolio return.

Execution. Filing-day closing prices are used for this illustration; a filing released after market close cannot be acted on at that day's closing price. Using the next trading day's opening prices for both entry and exit yields an average return of +29.9%, versus +2.1% for the S&P 500. Dividends and costs are excluded.

Selection. The six stocks form an exploratory basket. The full set of 11 companies, before any classification, averaged a return of +19.5%, versus +3.4% for the S&P 500. Removing ALB and ON from the six-stock illustration leaves +11.6% versus +1.5%; that is a sensitivity check, not a new selection rule.

Full sample. ZBRA, F, PKG, KR, ON, ALB, GM, C, FITB, IP, DOW. The download includes input prices, filing dates, event-source links, and calculation results.

Sources and timing. Filings and filing dates: Akkru. Daily stock and index prices: Yahoo Finance; entry and exit prices for F, GM, and FITB were also checked against Akkru. This retrospective illustration does not establish an out-of-sample, point-in-time classifier test. Any later amendments must be treated as available only from their public disclosure dates.

Research inspiration. Rouen, So & Wang, Core earnings: New data and evidence, Journal of Financial Economics 142(3), 2021. Their study concerns core-earnings adjustments in 10-Ks; the 8-K screen and holding rule here are our adaptation.

Download full calculation data ↓

— The Akkru Team

Technical demonstration, not investment advice. Historical results are illustrative, with substantial sector exposure, and do not establish a repeatable trading edge. “About five minutes” refers to an initial agent-assisted pass; runtime varies with the model and scope.