Skip to content

The prompt from the video

I Re-Created the $100K Grok Bot Trading Strategy (Then Made It Better)

15m 44s · 25 chapters · Lewis Jackson

The prompt

Get the Reverse-Engineered Grok Bot Strategy prompt

Surfer’s strategy as one prompt for Claude Code: the agents, the rules and the checks. It starts in scan mode, so no money moves.

Get the prompt

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

How much do you trade with?

Optional. I’ll use the answers to say who this suits in the next video.

Unsubscribe any time. Privacy policy.

grok-bot-prompt.md · the only thing you paste

The one-shot prompt

One file · 5 KB

The prompt

Fill in the form to unlock it.

Become an AI trading wizard

“I saw a Grok bot trading strategy that made $100,000. And I wondered if I could not only reverse-engineer it, but make it better.”

I pulled all 951,078 trades from Surfer’s Grok Bot wallet through Bot X-Ray. From 15 August to 27 October his bets lost about $17,000 after fees, and Polymarket paid him about $104,000 in rebates and rewards for leaving his orders waiting. I turned his strategy into one prompt for Claude Code, tried three ideas that made it worse, then found one change that beat his version in October, a month it had never seen.

Your host

Lewis JacksonReverse-engineers trading bots with Bot X-Ray. Founder of Part-Time Quant Academy.

Behind the research

What didn’t
make the video.

8 findings · 4 frames

The bets, after fees: −$17k−$17kThe bets, after feesRebates and rewards: $104k$104kRebates and rewardsWhat it made: $87k$87kWhat it madea loss
01It’s paid for waiting, not for being right.951,078 trades from 15 August to 27 October. The bets lost $17,111 after $41k in fees. Rebates and rewards paid $104,056.Surfer’s trades from Polymarket’s public data, with rebates and rewards from Polymarket’s daily P&L for his wallet.
Aug: 1.2% profit per $ traded1.2%AugSep: 0.8% profit per $ traded0.8%SepOct: 0.5% profit per $ traded0.5%Octprofit on each $ traded
02That’s also its weak spot.Profit per dollar was 1.2% in August and 0.5% in October. The rebate rate fell from 0.85% to 0.52%, and the $32k of liquidity rewards stopped.Bot X-Ray, 28 Oct 2026.
Lewis at the desk, explaining the Grok Bot
From the video · 2:05
Jul: +$14k+$14kJulAug: +$54k+$54kAugSep: +$35k+$35kSepOct to 27th: +$12k+$12kOct to 27thGrok Bot launches, 11 Aug
04August was the big month.Grok Bot launched on 11 August. The wallet traded more than three times as much as in July, and profit went from $14k to $54k.Bot X-Ray, 28 Oct 2026. Launch date from xAI’s own page.
Surfer’s versionSurfer’s version: around $1,800around $1,800My versionMy version: around $2,700around $2,700started with $1,000October, head to head
05Both bots with $1,000 in October.Replaying their real trades, his month ends at around $1,800 and mine at around $2,700. In 28% of the replays his ended below the $1,000 it started with. Mine did in 2%.1,000 replays of October’s real trades: history, not a forecast.
Lewis making a point with both hands
From the video · 14:20
Buy both sides for under $1.Up52¢+Down43¢95¢for the pairpays $1
07How one pair makes 33 cents.40 Down at 79¢, then 40 Up at 18¢ one second later. $39.67 in, $40 back, whichever side wins.Surfer’s trades on Polymarket’s public data, 14 Oct 2026. Illustrative prices in the chart.
Bot X-Ray’s Wallet screen for Surfer’s wallet: $129k since 20 March 2026
08The money’s real.Lifetime profit is $129k since 20 March, on $14.48M traded across 1,399,717 trades.Bot X-Ray on @l5zn1bwom8etsk, 28 Oct 2026.
Lewis talking to camera
From the video · 5:30
Bot X-Ray’s Skill or luck screen for Surfer’s wallet: 129 green days out of 181
10It wins on most days.129 of 181 days were green. A coin flip does that about once in 200 million tries.Bot X-Ray on @l5zn1bwom8etsk, 28 Oct 2026.
Lewis reading the trades
From the video · 10:00
Made per $100 traded
Surfer’sMine
August, from the 15th$1.01$1.29
September95¢$1.40
October42¢77¢
12Mine beat his every month.I picked the change on August and September, so October is the honest test.My test on Polymarket’s public data, 28 Oct 2026. History, not a forecast.

From the video

“I wanted more profit than his strategy, and fewer bad days as well.”
Lewis Jackson
Share this video

Questions

Before you run it.

01Does it cost anything?

No charge for the prompt. You need Claude Code to run it, and that comes with its own plan.

02Is this Surfer’s bot?

No. It builds the strategy he described in his 23 August article, as a team of agents. It isn’t his code. Anyone selling “the Grok Bot” in comments or DMs is running a scam.

03Was it really Grok Bot?

His article says he rebuilt an older bot as a team of Grok Bot agents. His wallet was already trading on 20 March, before Grok Bot launched on 11 August, so the trades can’t tell me how much Grok changed. August was his biggest month.

04Where does the money come from?

Rebates. Polymarket pays a small reward to orders that wait on the book. From 15 August to 27 October his bets lost about $17,000 after fees, and rebates and rewards paid about $104,000. Over his whole history since 20 March, his bets come out about flat: $127k of his $129k is rebates and rewards.

05Can I use it where I live?

The prompt asks where you are and checks Polymarket’s rules for your country first. If trading isn’t allowed there, it builds scan mode only and doesn’t suggest a way around the block.

06Can it trade real money?

It starts in scan mode: it logs every trade it would make and places none. Live trading stays off until you ask for it in so many words, scan mode has run for at least a week and trading is allowed where you live.

07Will it make money?

The numbers on this page are Surfer’s history and a test on real October trades. They say nothing about what you’ll make, and nothing here promises a profit. His profit per dollar fell from 1.2% in August to 0.5% in October.

08Do I need to know how to code?

No. You paste one prompt and Claude Code does the setup.

More videos · 5