Fact check · trading agents · checked 2026-10-16
Did Grok turn $170 into $2,154 overnight by copying eight traders?
What was claimed?
“I’M GENUINELY SHOOK, 9 HOURS OF SLEEP, 97 COPIED TRADES, 37 LOSSES, AND GROK STILL TURNED MY $170 INTO $2,154 BEFORE MY F*CKING ALARM WENT OFF”
The post goes on: “Grok didn’t pick one trader. It picked eight, all from the top of the 30-day board, and copied them at the same time through CopyFomo.” It says every trade was mirrored “with about a third of a second of lag”, and “When a trader sizes in too big for my account, the bot just skips it.” It says “The losses are real. 37 of them”, “88 wins, 70% hit rate”, “3am the balance was $586”, and that the top trader “made me $436, the eighth made $99”. It ends: “The full overnight setup is below, free.”
In numbers: a start of $170, an end of $2,154 (12.7 times), 97 copied trades, 37 losses, 88 wins, nine hours from 11pm, with no time zone given. The post has no links. Its evidence is a 33-second video. It quotes a post by @0xNevsky, which is the setup it calls “below”.
What did we check, and how?
- The post and its replies, read through X’s public data (api.fxtwitter.com) on 2026-10-16: the text, the time from the post’s ID, views, and 35 of its about 80 replies (the rest did not load).
- The video, downloaded and read frame by frame: every number, label and log line on screen.
- The quoted setup: @0xNevsky’s post and the X Article it links.
- CopyFomo’s own pages: what the product is (a Telegram bot that copies traders on Robinhood Chain, Base, BNB and Solana), and its fees and referral terms.
- The author’s recent posts: the 67 posts from 2026-07-22 to 2026-10-15 that loaded (the profile shows about 212, replies included).
- Binance’s public hourly SOL/USDT record for 2026-10-14 and 10-15, to test whether the buys on screen fit the balance on screen.
We looked for a wallet address, an account name or a transaction anywhere in this material. We found none, so we could not pull a trading record. On 2026-10-16 we asked @A_Pirate_I publicly on X for the wallet address, with a link to this page, and we are waiting for a reply. If one comes, we will pull the record and amend this page visibly. Two of the replies we could read had already asked for proof or for the wallet’s history; no answer from the author appears among the replies that loaded.
What does the record show?
| Question | The claim says | What we found |
|---|---|---|
| Is there a record to check? | “The losses are real.” | Unverifiable from any public record: no wallet address, CopyFomo account or transaction appears in the post, the replies that loaded, the video, or the linked setup. CopyFomo trades from a wallet the user owns, on public chains, so a record would exist if the author gave its address. That is why the status is not unverifiable by nature. Nothing here lets anyone check that the author’s account made a trade. |
| What is the evidence? | A 33-second video | Recomputable from the video. A web dashboard titled “Grok Bot · TRADING ON ∞ COPYFOMO”, labelled “LIVE”. CopyFomo itself works inside Telegram. The on-screen clock runs from 23:03 to 08:00 in 33 seconds, about 16 minutes of night per second of video. No time zone, wallet or transaction is shown. Nothing on screen ties it to an account. Its figures behave like a sequence built toward set end values. In the first frame, all eight trader cards already show copied gains, though only 2 copies had been made and 0 had closed. At on-screen 02:09, the eight card figures add up to $274.04 against a total of $270.95. At the last frame they add up to the cent. We cannot prove how the video was made. |
| Do the counts add up? | “97 COPIED TRADES, 37 LOSSES” and “88 wins, 70% hit rate” | Recomputable from the post. 97 trades with 37 losses leaves 60 wins (62%), not 88. The video has the same gap: its header reads “97 COPIES”, its log “125 RESOLVED”, and its win-rate panel “88W 37L”, which is 125. 88 of 125 is 70.4%, so the 70% fits the 125, not the 97. A dashboard could count each partial sale as its own result; the video does not say so, and the post calls the same night 97 trades with 37 losses and 88 wins. The post’s three counts cannot all be true at once. |
| Do the dollars add up? | “$170 INTO $2,154”; top trader “$436”, the eighth “$99” | Recomputable from the video. The last frame shows “BALANCE $2,154.00”, “started at $170.00” and “TOTAL P&L +$1,984.00”. The eight traders’ copied figures sum to exactly $1,984.00, and each is an exact whole percentage of that total: 22%, 17%, 15%, 13%, 11%, 9%, 8% and 5%, together 100%. For any one figure, landing on a whole percentage of the total to the cent is about a 1-in-2,000 chance; all eight did. |
| Could the account afford its first trades? | A $170 start; “the bot just skips” trades too big for the account | Recomputable from the video and a public price record. The first frame shows a balance of $170.72 and two copied buys: “BUY 0.39 SOL $BOME” and “BUY 2.28 SOL $PIPPIN”. For the second buy alone to fit inside the balance, one SOL would have had to cost less than $74.88. Binance’s public hourly SOL/USDT record for 2026-10-14 00:00 to 2026-10-15 16:00 UTC never went that low. That window covers every time zone in which an 11pm-to-8am night ends before the post. On screen, the account spends more than it holds in its first minutes, which the post says the bot does not do and the dashboard’s own “skipped” lines say it avoids. This assumes the logged buy sizes are the user’s own fills, which is how the log labels them. |
| Who is behind it? | Grok, picking eight traders “from the top of the 30-day board” | Unverifiable by nature. Nothing public shows that Grok chose or placed any trade. We did not check the eight trader names against fomo’s 30-day board, because a leaderboard read on another day would not test the night. The setup the post points to gives Python code that calls xAI’s API and a rug-check service, and tells readers to sign up through a referral link under its author’s name. |
Where does the "free setup" lead?
- The post’s “setup below” is the post it quotes: @0xNevsky, 2026-10-12 13:26:41 UTC, about 822,638 views as fetched
2026-10-16, which links an X Article titled “Grok + copyfomo: $320 Into +$3,879 in One Month Without Picking a Single Trade (Setup Guide)”.
The article gives a referral link under its author’s name,
t.me/copyfomo_bot?start=ref_0xnevsky, and says, in capitals: “FILLS GO THROUGH t.me/copyfomo_bot?start=ref_0xnevsky AND NOWHERE ELSE.” It also says “Most people who copy trade lose money”. We did not check the article’s own claim. - CopyFomo’s documentation says people who join through a link “get 10% off their trading fees, for good, and you earn 15% of what they pay, on every trade”. Its fee is “2% per trade”. Under those terms, the setup’s author would earn a share of the fees of readers who sign up through the link.
- The post itself carries no referral link and no paid-promotion label; the link is in the article it quotes.
Is this a one-off?
- No. @A_Pirate_I opened its account on 2026-01-22. Through 2026-09-25 it posted mostly about AI-generated influencers and AI video. From 2026-09-26 to 2026-10-15 it posted 19 times about an AI trading session and its winnings, one a day except 2026-09-28, among the 67 posts we could load. Examples: “$100 AT MIDNIGHT. $5,468 BY MORNING” (2026-10-01), “I LEFT GROK $25 AND WENT TO BED. WOKE UP TO $1,777” (2026-10-04), and “I GAVE 6 GROK BOTS $150” (2026-10-14).
- 12 of the 17 posts from 2026-09-29 open on a “Grok Bot” dashboard laid out like this one: balance, profit and loss, win rate, copy log, trader cards, with other numbers. We read only their first frames. Each of the 19 quotes another account’s post or article, and most of the articles we read carry trading or copy-trading referral links.
- 17 of the 19 had under 3,000 views when we read them; this one had about 294,744. The account has had posts this size before: five earlier posts, from 2026-07-23 to 2026-09-20, passed 350,000 views.
- Two copies in Chinese, posted on 2026-10-16, reuse the same video, which still reads “TRADING ON COPYFOMO”, but swap in a referral link for a different trading bot. The larger had about 38,000 views. Searching X and the web for the post’s phrases and figures, we found no version earlier than @A_Pirate_I’s.
What does this prove, and what does it not?
- It proves that the post offers no record anyone could check: no wallet, no account, no transaction.
- It proves that the post’s counts do not reconcile with each other; that the eight per-trader figures are each an exact whole percentage of the final total, yet do not add up to the total mid-video; and that the first buys logged on screen are larger than the balance shown.
- It proves that the setup the post points to carries a referral link that, under CopyFomo’s published terms, pays its author a share of readers’ fees.
- It does not prove that no trades happened, or that the author made or lost money. Without a record, neither can be shown.
- What would change the status: a wallet address or CopyFomo trade history covering the night of 2026-10-14 to 2026-10-15. We would pull it, test the start, the end, the 97 trades and the timing, and amend this page visibly.
Recompute it yourself
- Read the post through
https://api.fxtwitter.com/A_Pirate_I/status/2110761136706527452(x.com blocks most automated readers). Its time is also in its ID: (2110761136706527452 >> 22) + 1288834974657 milliseconds since 1970. - Download the video from the post’s media link and step through it. The first frame shows the $170.72 balance, two copies and the two buys. The frame at on-screen 02:09 shows the eight card figures summing to $274.04 against $270.95. The last frame shows “8 WALLETS · 97 COPIES”, “WIN RATE 70% 88W 37L”, “125 RESOLVED” and the eight cards.
- Add the eight “COPIED” figures on the last frame: $436.48 + $337.28 + $297.60 + $257.92 + $218.24 + $178.56 + $158.72 + $99.20 = $1,984.00. Divide each by $1,984.00.
- Divide the first frame’s $170.72 balance by 2.28 SOL. Then read SOL/USDT hourly candles:
https://api.binance.com/api/v3/klines?symbol=SOLUSDT&interval=1h&startTime=1791936000000&endTime=1792080000000(2026-10-14 00:00 to 2026-10-15 16:00 UTC) and compare the lowest low with that figure. - Read @0xNevsky’s post and article through api.fxtwitter.com, and CopyFomo’s terms at copyfomo.com/docs.
- For the posting history, read each status ID listed in this page’s data file through api.fxtwitter.com, and the two copies listed there.
Statuses used on this site are defined on the fact-check index. This page
carries ClaimReview structured data that matches its text.