Skip to content

The prompt from the video

I Tested 32,248 Insider Buys

10 chapters · Video coming

The prompt

Get the insider strategy prompt

One prompt for Claude Code: it proves the CEO and CFO strategy on the SEC’s own files, then sets up a daily scanner for each new buy, with a menu of settings to try. No trades unless you choose.

Get the prompt

Your name and email, and the prompt appears straight away.

Your email goes on my list for notes on this test and the next videos. Unsubscribe in one click. Privacy policy.

insider-prompt.md · the only thing you paste

The one-shot prompt

One file · 22 KB · Runs on your computer

The prompt

Fill in the form to unlock it.

A projected table of insider buy filings with one CEO buy lit up green, beside a rising lineVideo coming

Become an AI trading wizard

“When a company’s boss buys their own company’s stock, with their own money, it feels like a tip from the one person who really knows.”

I tested every insider buy of $25,000 or more since 2019: 32,248 buys in 3,201 stocks, from the SEC’s own files. I picked rules on 2019 to 2023 and locked 2024 onwards away. Once you could copy it, the typical buy trailed the S&P. One kind held up: buys by a CEO or CFO, held for 60 trading days, turned $10,000 into about $17,200 on years the test had never seen, against about $16,100 in the S&P. One prompt installs that strategy in Claude Code and finds each new CEO and CFO buy.

Your host

Lewis JacksonTests trading ideas with Claude. Founder of Part-Time Quant Academy.

Inside the test

What 32,248 buys
showed.

4 charts

2019–2023, rules picked here2024 to 5 Nov 2026, lockedS&P 500S&P 500, 2019–2023, rules picked here: $10,000 became $20.6K$20.6KS&P 500, 2024 to 5 Nov 2026, locked: $10,000 became $16.1K$16.1KSmall companiesSmall companies, 2019–2023, rules picked here: $10,000 became $16.0K$16.0KSmall companies, 2024 to 5 Nov 2026, locked: $10,000 became $14.2K$14.2KAll buys, 5 daysAll buys, 5 days, 2019–2023, rules picked here: $10,000 became $20.2K$20.2KAll buys, 5 days, 2024 to 5 Nov 2026, locked: $10,000 became $13.9K$13.9KAfter a 30% fall, 5 daysAfter a 30% fall, 5 days, 2019–2023, rules picked here: $10,000 became $104K$104KAfter a 30% fall, 5 days, 2024 to 5 Nov 2026, locked: $10,000 became $14.7K$14.7KSame, $1M+ a daySame, $1M+ a day, 2019–2023, rules picked here: $10,000 became $27.9K$27.9KSame, $1M+ a day, 2024 to 5 Nov 2026, locked: $10,000 became $9.1K$9.1KCFO buys, 5 daysCFO buys, 5 days, 2019–2023, rules picked here: $10,000 became $52.5K$52.5KCFO buys, 5 days, 2024 to 5 Nov 2026, locked: $10,000 became $17.9K$17.9KCEO or CFO buys, 60 daysCEO or CFO buys, 60 days, 2019–2023, rules picked here: $10,000 became $26.0K$26.0KCEO or CFO buys, 60 days, 2024 to 5 Nov 2026, locked: $10,000 became $17.2K$17.2Kdotted line: the $10,000 you started with
01Five rules, two periodsEvery rule was picked on 2019 to 2023, then run once on 2024 onwards, unchanged. Buying after a 30% fall turned $10,000 into about $104,000 early on, then lost money on the locked years in stocks people actually trade. CEO and CFO buys held for 60 trading days grew at the same rate in both.Yearly growth after costs. Backtest. Yahoo has no prices for delisted stocks, so 2019–2023 looks better than it was; 2024–2026 is the fairer read. Backtest results are history, not a forecast.
$10K$15K202420252026$17,200 CEO or CFO buys$16,100 S&P 500$14,200 Small companies
02The one kind that held up$10,000 into CEO and CFO buys at the start of 2024 would have been about $17,200 by 5 November 2026. The S&P 500 made about $16,100, small companies about $14,200. It fell further on the way: its worst drop was 24.7%, against 18.8% for the S&P.Each buy held 60 trading days, about 135 stocks at once. The SEC’s files end in March 2026, so the last trades closed in June and the line is flat since. Backtest. Yahoo has no prices for delisted stocks, so 2019–2023 looks better than it was; 2024–2026 is the fairer read. Backtest results are history, not a forecast.
−7%The stock’s move in the three months beforethey buy. The S&P went up 3.4%.69%of buys land 1 to 2½ months into the quarter,after results and before the blackout.$129KThe middle buy. It adds 7% to whatthey already own.43%have another insider at the same companybuying within 30 days.1 dayfrom the trade to the filing, so the publicsees most buys within a day or two.
03What insiders actually doThey buy when the stock’s falling, at a set time in the quarter, a little at a time, often together, and they tell everyone within a day or two. That last one is the problem: by the time you can see the buy, so can everyone else.Across all 32,248 buys of $25,000 or more since 2019.
Every buyEvery buy: 45% beat the S&P 50045%Their third buy or more in three yearsTheir third buy or more in three years: 43% beat the S&P 50043%Routine buyers, same month every yearRoutine buyers, same month every year: 42% beat the S&P 50042%Buys under a 10b5-1 plan (57 of them)Buys under a 10b5-1 plan (57 of them): 16% beat the S&P 50016%
04The ones to skipThe typical buy already trailed the S&P. These did worse. The 10b5-1 plan buys were the worst I found, though there were only 57 of them.Share of buys that beat the S&P 500 over the next 60 trading days, 2019–2023.

From the video

“Who’s buying, and how long you hold, made that whole difference.”
Lewis Jackson
Share this video

Questions

Before you run it.

01What does the prompt do?

It downloads the SEC’s insider files and daily prices from Yahoo, proves the CEO and CFO strategy on years it was picked on and years kept locked away, then sets up a scanner that reads each day’s new filings and keeps a practice portfolio. It runs in Claude Code on your own computer.

02Why does it ask for my name and email?

The SEC asks every script that downloads its files to say who’s asking. The prompt puts your name and email in that one header and nowhere else. Without them the SEC refuses the download.

03Will I get the same numbers?

Close, not identical. Yahoo revises its prices, and if you choose to run it to today, your locked years are longer than mine. If yours differ a lot, the prompt tells Claude to show you which step they split at.

04Does it place any trades?

Not unless you choose to. The scanner writes down the trades it would make and places none. Trading it for real is your call, with your own broker key.

05Should I trade the CEO and CFO rule?

That’s your call, and this isn’t advice. Know what it takes first. It fell further than the S&P on the way (24.7% against 18.8%), it holds about 135 stocks at once and makes about two trades every day the market’s open, and 2026 so far has been weak. That’s a job for a bot, on paper first.

06What can’t the test see?

Companies that went bust or were taken off the market, because Yahoo has no prices for them. In 2019 I had prices for 61% of the buys, by 2026 96%. The missing ones lean towards losers, so the early years look better than they really were. It also buys at the opening price with a flat cost, and thin stocks cost more in real life.

More videos · 5

LewisInvestingbyJackson

Become an AI trading wizard.

Legal

© 2026 Jackson. All rights reserved.Privacy · Terms · Contact