Skip to content

The prompt from the video

I Re-Created a 1,484% Day Trading Strategy With Claude Code

11 chapters · Video coming

The prompt

Get the ORB test prompt

The strategy as one prompt for Claude Code: it installs the exact settings, proves them on the minute-by-minute prices with a thousand coin flips and your own costs, and sets up practice mode. Then you get five things to change and a results card.

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.

orb-prompt.md · the only thing you paste

The one-shot prompt

One file · 23 KB · Practice mode first, no orders

The prompt

Fill in the form to unlock it.

A projected candlestick chart: the first candles boxed, then a climb to a dashed target lineVideo coming

Become an AI trading wizard

“One candle, every morning. It loses three times out of four, and it still comes out ahead.”

A 2023 paper says one candle a morning made 1,484%. I gave Claude Code the paper, had it write the rules down before it saw a single result, and tested them on nearly eight years of one-minute Nasdaq prices, including the years after the paper came out. Across 1,965 trades it lost three in four and still finished up about $24,000 at $100 risked a trade, after costs. The rules beat 998 of a thousand coin flippers. The 1,484% came from a fund that moves three times as much as the Nasdaq, on trades that cost almost nothing. Traded cheaply from January 2024 to October 2026, risking 1% a trade, the plain rules turned $10,000 into about $14,900. The authors’ upgrade took the same $10,000 down to about $800. One prompt installs it and proves it on your own computer, in Claude Code.

Your host

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

Inside the test

What one candle
a morning showed.

5 charts

Each year, at $100 risked a trade2019: $4,390$4,39020192020: $1,670$1,67020202021: $5,130$5,13020212022: $4,150$4,15020222023: $820$82020232024: $1,410$1,41020242025: $6,220$6,22020252026: $190$1902026$23,970 in total · 2026 runs to 28 OctoberThe same trades, at three costs0.7 points, what I charged0.7 points, what I charged: $23,970$23,9701.5 points1.5 points: $17,480$17,4802.5 points, a typical CFD spread2.5 points, a typical CFD spread: $9,370$9,370
01Does it make money?1,965 trades from 2019 to 28 October 2026, with 0.7 of a point charged on every one for costs. It won 24.6% of them and still finished up about $24,000 at $100 risked a trade. Every full year was up. What each trade costs decides how much you keep: at 1.5 points it made about $17,500, and at 2.5 points, a typical CFD spread, about $9,400.Nasdaq 100 one-minute prices from Dukascopy, on the days the US stock exchange was open. Backtest results are history, not a forecast.
Target, 10× the risk · about +$680 on one microStop, 34 points · about −$68 on one micro9:3011:0012:3014:0016:00the first five minutestarget hit, 11:39
02One morning, start to finishMonday 21 September 2026. The first five minutes closed up, so the rule bought at 9:35 with the stop at the bottom of that candle, 34 points away. The target was ten times that, and the price got there at 11:39. On one micro Nasdaq contract, at $2 a point, that’s about $68 risked and about $680 made.Nasdaq 100 one-minute prices from Dukascopy, shown in five-minute candles. One of the big days, picked to show the rules. Most mornings the stop gets hit.
The paper’s versionon the Nasdaq, 0.7 points a tradeThe paper’s version, on the Nasdaq, 0.7 points a trade: $10,000 became $14,890$14,890The authors’ upgrade20 stocks a day, at their own costsThe authors’ upgrade, 20 stocks a day, at their own costs: $10,000 became $3,890$3,890The authors’ upgradeplus a cent a share of slippageThe authors’ upgrade, plus a cent a share of slippage: $10,000 became $800$800dotted line: the $10,000 you started with
03The upgrade that lostLess than a year after the paper, the authors published an upgrade: the same idea on the 20 most unusually active stocks each day, with a much tighter stop. I tested it on every day since their study ended. Risking 1% of the account a trade, $10,000 became about $800. With every order filled at the exact price, it still fell to about $3,900. Over the same days, the plain version on the Nasdaq turned $10,000 into about $14,900.2 January 2024 to 28 October 2026, the 709 days the US stock market was open, with each position capped at four times the account. Alpaca prices for the stocks. Backtest results are history, not a forecast.
The targetThe target: $3,000$3,000One morning that hits its targetOne morning that hits its target: $5,000$5,000Your new finish lineYour new finish line: $10,000$10,000The loss limitThe loss limit: −$2,000−$2,000
04The rule that sinks it at a prop firmThe challenge I ran: a $3,000 target, a $2,000 loss limit and $500 risked a trade. One morning that hits its target makes $5,000. A best-day rule says no day can be more than half your profit, so that one good morning moves your finish line to $10,000. I started a new challenge on every trading day, 1,959 in all. With the rule it passed 6% of them. Without it, 31%.A target hit pays ten times the $500 risked, before costs. Pass rates are after 0.7 points a trade. Backtest results are history, not a forecast.
1,000coin flippers, each picking buy or sell at random.The real rules beat 998 of them.144versions to try to make it better.None passed on years it hadn’t seen.11,279trades in the authors’ upgrade,picked from 9,678 stocks since 2024.9,409minutes checked trade by trade, wherea one-minute chart couldn’t tell.
05What it tookI built a thousand coin flippers to see if the first candle tells you anything, and the real rules beat 998 of them. I built 144 versions to try to make it better, and none passed. Then I checked 9,409 minutes against the actual trades, because a one-minute chart couldn’t say if the upgrade’s tight stop got hit.Each coin flipper trades the same days, at the same minute, with the same kind of stop and target. Only the direction is random. No costs for anyone, because costs hurt the coin flippers more.

From the video

“It all came back to the plainest version, traded in the right place.”
Lewis Jackson
Share this video

Questions

Before you run it.

01What does the prompt do?

It installs the strategy on your own computer with the exact settings from the video, then proves it on the same one-minute Nasdaq prices I used: the test, the check that catches the days the exchange was closed, a thousand coin flips and your own costs. Then it sets it up to log each day’s trade in practice mode, gives you five things to change, like the target or a prop firm challenge, and writes you a results card. It runs in Claude Code.

02Where does the paper’s 1,484% come from?

They traded QQQ, a fund that follows the Nasdaq, at five hundredths of a cent a share, and assumed every order filled at the exact price. Then, for the 1,484%, they swapped in TQQQ, a fund that moves three times as much as the Nasdaq. So it’s three times the risk, on trades that cost almost nothing, in their years, 2016 to 2023.

03Where should I trade it?

Somewhere each trade costs very little. Nasdaq futures trade on an exchange in Chicago where the gap between the buy and sell price is usually a quarter of a point, so with commission you’re at about a point a trade or less. A micro contract is $2 a point. A typical CFD spread is about 2.5 points, and in my test that took more than half of the profit. The prompt works out what your own broker costs.

04Can I pass a prop firm challenge with it?

It’s hard with a best-day rule: in my test it passed 6% of 1,959 challenges. The money comes from a few big days, and one target hit at $500 a trade makes $5,000, so the rule moves your finish line from $3,000 to $10,000. Without that rule it passed 31%. Read the rules before you pay for a challenge.

05Will I get the same numbers?

On the full test, yes. It uses the same dates and the same prices as mine, and it leaves out the days the US stock exchange was shut, so you should see 1,965 trades and 998 of 1,000 coin flips. Your costs are your own. If yours differ, ask Claude to show you which step they split at.

06Does it place any trades?

Only if you ask it to. It tests on past prices and then logs each day’s trade in practice mode, with no broker and no account. If you tell it you want to trade it yourself, it walks you through your own broker or exchange, and your key stays in a file on your computer.

07Should I trade it?

That’s your call, and this isn’t advice. Know what it’s like first. It loses about three times in four, and its longest losing run was 18 trades in a row. Its worst slide ran from November 2022 to May 2023 and cost about $3,900 at $100 a trade. You can’t pick the big days, so you take every one. Run the daily check for a month before anything else.

More videos · 5

LewisInvestingbyJackson

Become an AI trading wizard.

Legal

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