Introducing the Signals Feed: Algorithmic SEC Filing Alerts
Track cybersecurity incidents, bankruptcies, financial restatements, and insider buying clusters in one real-time feed. No more digging through EDGAR.
Every day, the SEC receives thousands of filings. Most are routine. But buried in that stream are the material events that actually move stocks: a CEO getting fired, a company disclosing a data breach, or five insiders buying shares in the same week.
Finding those needles takes hours. I know because I've spent years doing it manually.
Today I'm launching the Signals Feed, a real-time stream of algorithmically-detected patterns from SEC filings. No more skimming 8-Ks hoping to spot something important. No more building spreadsheets to track insider clusters. The feed does the pattern matching for you.
What Signals Detect
The feed currently tracks five signal types across two categories.
8-K Material Events
Companies must file 8-Ks within four business days of certain events. We flag the three that historically cause the biggest price moves:
Cybersecurity Incidents (Item 1.05) Companies must disclose material cybersecurity incidents. On December 29, 2025, Coupang filed an Item 1.05 disclosing a security breach. These filings tend to drop after hours and move stocks before most investors know there's a problem.
Bankruptcy (Item 1.03) Chapter 7 or Chapter 11 filings. Dynatronics Corp filed Chapter 7 on January 12, 2026, and its entire board resigned the same day. The stock craters, but the bond implications are where it gets interesting. These are time-sensitive.
Financial Restatements (Item 4.02) When prior financials turn out to be wrong, that's Item 4.02. Southwest Gas Holdings filed one on January 9, 2026 to correct quarterly tax expenses. Restatements range from minor (accounting reclassification) to devastating (revenue recognition fraud). The signal gets you to the filing fast so you can figure out which.
Insider Transaction Clusters
Form 4 filings show when executives and directors buy or sell company stock. A single purchase tells you little. But when multiple insiders buy within the same week, that's unusual.
Cluster Buying We flag when 3 or more insiders purchase shares within a 7-day window, with total value between $500K and $20M. On January 5, 2026, 12 insiders at TriplePoint Venture Growth (TPVG) bought $2.13M in stock. Coordinated buying like this stands out. The $500K floor filters out routine director stock purchases. The $20M ceiling filters out planned transactions tied to equity grants.
Cluster Selling Same logic, but for sales. Multiple insiders selling in the same week suggests they know something, though selling is harder to interpret. Insiders sell for taxes, to spread out their holdings, and dozens of other reasons. Still worth flagging.
Why I Built This
I've watched the wrong things for years.
I used to set up RSS feeds for every company in my portfolio and scan headlines for material events. The problem is signal-to-noise ratio. For every meaningful 8-K, there are dozens of routine filings: earnings date confirmations, exhibit amendments, director committee assignments.
The real signals are buried. A cybersecurity disclosure lands at 4:47pm on a Friday. An insider cluster buy shows up across four separate Form 4s filed over a week. You'd have to be watching constantly and cross-referencing filings to catch it.
That's not realistic. So I built detection for the patterns that actually matter.
What the Feed Looks Like
Each signal includes:
- Company and ticker with a link to their full filing history
- Signal type badge so you can visually scan for what you care about
- Date the event occurred (or window end date for clusters)
- AI summary explaining what happened, for supported signal types
- Key stats like insider count and total dollar value for clusters
You can filter by signal type. If you only care about cluster buys, show only those. If you want to monitor all material 8-K events, filter to those three.
The Filtering Problem
Filtering matters more than speed.
Yes, the signals appear within minutes of filings hitting EDGAR. But the real value isn't speed. It's the filtering. I'm not building a Bloomberg terminal. I'm building a tool that shows you the 5 events that matter out of the 3,000 filings that hit today.
The cluster buy filter is a good example. There are roughly 5,000 Form 4 filings per week. Running the cluster detection algorithm against that stream produces maybe 10-15 signals per week. That's a 99.7% noise reduction.
Same with 8-K events. The SEC gets 50-100 8-Ks per day. Three items (1.03, 1.05, 4.02) account for maybe 2% of those. The filter lets me focus on the filings where something actually went wrong.
What's Next
The five signal types I launched with are the ones I've found most useful for my own research. But there's more to detect:
- Activist positions: 13D filings where an investor crosses 5% and discloses activist intent
- Large insider sales by individual: When a CEO sells $50M of stock, that's different from routine selling to spread out holdings
- Earnings guidance changes: 8-K Item 7.01 voluntary disclosures that update guidance
- Auditor changes: Item 4.01 filings where a company fires their auditor
If you have signal ideas, I want to hear them. What patterns do you watch for manually that could be automated?
Try It Now
The Signals Feed is live and free to use.
Signals are algorithmic pattern detection, not investment advice. The SEC filing is always the authoritative source. Use signals to know when to pay attention, then read the actual document.