Investing Engines Dashboard

Peterson's Investing Engine

An On-going Investing & FinTech Experiment

Also: Notes on working with AI, a short blog of tips and ideas from day-to-day work with AI.

In mid-2025 I began using AI to help with investing. After twenty years of playing the market myself, the frustration was always the gaps: weeks or months with nothing worth acting on. My first tools flagged signals, but mostly produced noise. It took about a year to build a system that researches the market and finds meaningful opportunities on its own. I refer to these as investing engines, and my current focus is making them respond well to rising and falling markets alike. Simulated trading began in early 2026; real money in June 2026. The results are early, but the numbers are accurate. White sections update daily; teal sections I update by hand. Founders profile.

Data points
56,068,340
that run our engines · updated daily
Proprietary data points
2.174M
non-public · produced by our engines
Annualized return
22.37%
realized trades · invested capital · as of Sep-2026
Sharpe
1.35
vs. SPY · as of Sep-2026
Beta
0.26
vs. SPY · as of Sep-2026
Alpha (ann.)
+8.00%
vs. SPY · as of Sep-2026
R²
0.22
vs. SPY · as of Sep-2026
Max drawdown
-3.04%
as of Sep-2026

Normalized equity curve

indexed to 100 on the first day · as of Sep 25, 2026
949698100102104106JunJulAugSep
Portfolio 103.0 SPY 101.9 Russell 2000 96.6

The data below is current and pulled live daily, a portion of the millions of data points the engines track. Not all data points are necessarily utilized by the engines in all trading scenarios.

AI Engine Index

AI risk appetite · as of Oct 5
48 / 100
Balanced
+6.4 pts vs. the prior week

Every morning our AI is handed that day's fresh set of data and asked, opportunity by opportunity, whether it would act or pass. The Index is the share that got a yes. Broadly put, above 50 the engine is finding a fair number of opportunities worth acting on. Above 70 it is saying yes to most of what it sees; below 30 it is mostly sitting on its hands. The reading shown is always yesterday's, on purpose, and a day with too few meaningful data points, or a bad-data day, is skipped. The line shows the last twenty readings. The small figure beneath the number compares the newest reading with the average of the prior five trading days.

Market nervousness

as of Oct 6
56%Nervousof which 24% acute

Every stock has two kinds of turbulence. One is how much it actually bounced around over the past year. The other is how much traders are betting it will bounce around in the weeks ahead. When traders expect bigger swings than the stock has actually been making, they are bracing for something. This is the percent of stocks where traders are gripping the wheel harder; the acute number is the percent where they are white-knuckling it.

Market breadth

as of Oct 5
42%Weak
  • Above their long-term trend42%
  • Above their short-term trend27%
  • Near a one-year high11%
  • Near a one-year low21%

How many companies are participating in the market's move, not just the biggest names. Broad participation is a healthier market than a narrow one.

Sector momentum

as of Oct 5
Sector1w1m3m
Industrials+10.0%+7.5%+23.1%
Materials+2.4%+5.6%+3.8%
Information Technology+0.2%+5.4%-1.9%
Consumer Defensive-1.3%+3.3%+4.5%
Health Care-3.4%-2.1%+5.5%
Consumer Cyclical+3.2%-2.3%+3.0%
Communication Services-0.7%-2.3%+5.1%
Real Estate-0.6%-4.9%-18.1%
Energy+1.9%-5.7%-10.6%
Financials-1.7%-6.1%-1.8%
Utilities-2.9%-10.9%-18.0%

Which parts of the market have been leading and lagging over the past week, month, and quarter.

Insider buying vs. selling

trailing 30 days
15%of insider trades were purchases
  • Healthcare13% buying
  • Technology2% buying
  • Financial Services20% buying
  • Consumer Cyclical44% buying
  • Industrials13% buying
  • Communication Services10% buying
  • Energy19% buying
  • Basic Materials36% buying

When company executives and directors trade their own shares, how often they were buying rather than selling, by sector. Selling is normal; a rise in buying is the interesting signal.

Agency attention by sector

trailing 4 weeks
  • Healthcare22%
  • Technology18%
  • Financial Services15%
  • Consumer Cyclical15%
  • Industrials8%
  • Consumer Defensive6%
  • Communication Services5%
  • Basic Materials5%

Where government and agency actions, to include approvals, enforcement, filings, and rulings, have been concentrating over the past month, by sector.

What kind of action

trailing 4 weeks
  • Lawsuits29%
  • Mergers & acquisitions27%
  • Leadership changes17%
  • Share issuance10%
  • Health approvals & rejections9%
  • Enforcement4%
  • Activist investors2%
  • Insider buying2%

The mix of notable company events on the public record this month, from deals and lawsuits to approvals and leadership changes.

Earnings season

next 4 weeks
Sectorwk of Oct 54%wk of Oct 124%wk of Oct 1924%wk of Oct 2668%
Financial Services0%2%10%12%
Industrials0%1%3%9%
Consumer Cyclical0%0%3%9%
Real Estate0%0%3%9%
Technology1%0%1%8%
Healthcare1%0%1%8%
Basic Materials1%0%1%5%
Energy0%0%1%5%

Where the coming weeks' company results are concentrated. Darker cells mean a busier week for that sector.

Economic calendar

the week ahead
  • Tue, Oct 6
    • API Crude Oil Stock Change (Oct/02)
    • Balance of Trade (Aug)
    • Exports (Aug)
    • Fed Bowman Speech
    • Fed Logan Speech
    • Fed Williams Speech
    • Imports (Aug)
  • Wed, Oct 7
    • FOMC Minutes
    • EIA Crude Oil Stocks Change (Oct/02)
    • EIA Gasoline Stocks Change (Oct/02)
    • Fed Williams Speech
    • MBA 30-Year Mortgage Rate (Oct/02)
  • Thu, Oct 8
    • Fed Musalem Speech
    • Fed Waller Speech
    • Initial Jobless Claims (Oct/03)
  • Fri, Oct 9
    • Michigan Consumer Sentiment (Oct)
    • CFTC Crude Oil speculative net positions
    • CFTC Gold Speculative net positions
    • CFTC Nasdaq 100 speculative net positions
    • CFTC S&P 500 speculative net positions
    • Fed Collins Speech
  • Mon, Oct 12
    • Budget Balance (Sep)
    • Monthly Budget Statement (Sep)

Government and central-bank releases on the calendar for the coming days. Bold items tend to move markets.

Recent deal announcements

98 within the last 30 days
  • Oct 5ALLEAllegion (ALLE) has acquired Overly Door Company, a door manufacturing company. No financial terms or deal size are disclosed in the filing.
  • Oct 5BETRBetter Home & Finance (BETR) has announced progress toward the sale of Birmingham Bank, indicating a definitive or near-definitive divestiture transaction is underway.
  • Oct 5CLHOne Rock Capital Partners has completed the sale of EnviroServe to Clean Harbors (CLH), representing a completed acquisition transaction. No deal terms or financial details are provided in the filing.
  • Oct 5CSLCSL and Alentis have announced a partnership to develop and commercialize lixudebart for rare kidney and liver diseases.
  • Oct 5GOOGGoogle has reportedly received a favorable development in its effort to acquire Spirit Airlines' data assets for AI training purposes.
  • Oct 5GRNDGrindr (GRND) has announced the acquisition of PurposeMed for $250 million, marking a strategic expansion into PrEP telehealth services. This is a definitive M&A transaction involving a material purchase price.
  • Oct 5IRTCiRhythm (IRTC) has completed its acquisition of VitalConnect, expanding its cardiac monitoring platform into ambulatory, inpatient, and hospital-to-home care segments.
  • Oct 5LDOSLeidos Security Enterprise Solutions has closed a deal with Analogic to enhance global security screening capabilities.
  • Oct 5MOBXMobix Labs (MOBX) acquisition target SPD is reportedly set to acquire Winner Water, a U.S. rare earth company backed by $53 million in federal awards. The headline is thin on details regarding deal terms or structure.
  • Oct 5PEBOPeoples Bancorp (PEBO) has announced a $728 million deal to acquire Capital Bancorp, aimed at expanding its national reach. This appears to be a definitive merger agreement of material size for a regional bank.

Mergers and acquisitions our systems flagged from the news. The ten most recent, one per company; the count shows the rest we track.