One account called a leveraged Korea ETF more often over the last 90 days than the whole of our index called bitcoin. The post announcing it carries a hash, a slash and a USDT pair, exactly like every other signal in the feed.

What is not a cryptocurrency in a crypto signal feed?

More than you would guess from the tickers. Of the 4,030 signals our parser has read from the channels in our index, 235 name an instrument that is not a cryptocurrency: 131 calls on individual shares, 94 on index and leveraged ETFs, and 10 on metals. That is 5.8% of the corpus, published by seven of the 22 accounts that post anything a parser can read.

The count is deliberately conservative. A wider list of equity-looking tickers takes the figure to 336 signals, or 8.3%. We leave those out. Several of the strings also name real crypto tokens, APEX being ApeX Protocol and NOW being ChangeNOW, and a wrong identification would be worse here than an undercount.

Over the last 90 days the most-called of these instruments was KORU, with 45 signals, and every one of them came from a single account. Bitcoin drew 23 over the same window from four separate accounts. Bitcoin, ether, solana and XRP together drew 99 from nine accounts. The 235 calls on things that trade on stock and commodity exchanges came from seven.

So the sharpest version of the comparison is also the narrowest one: one publisher called a leveraged Korea fund more often than our whole index called bitcoin. That is a fact about one publisher's habit, and it is still the most surprising line in this data. Not everything in the set is one account, either. SOXL drew 25 signals from six accounts and SOXS 18 from six.

None of this existed a year ago, or even six months ago. The first non-crypto signal anywhere in our archive is NVDAUSDT, dated 17 June 2026. June produced 17 of them, July 141, and August 77 by the 18th. The archive runs back to November 2021 and contains nothing of the kind before this summer.

Who is actually publishing them?

One account, mostly, and the concentration matters more than the headline share. Seven accounts published the 235 signals, and the distribution across them is not remotely even.

Account (masked) Non-crypto signals Total signals Share of its own output
#1 181 1,949 9.3%
#2 14 339 4.1%
#3 14 339 4.1%
#4 14 339 4.1%
#5 5 69 7.2%
#6 5 76 6.6%
#7 2 35 5.7%

Account #1 supplies 181 of the 235, which is 77% of the phenomenon. Accounts #2, #3 and #4 carry identical counts down every column because they are one feed republished under three brands, a pattern we documented separately in Clone signal channels. Counting them as three independent publishers triples one operation's weight in any average.

So the honest version of the finding is narrower than "signal channels have started calling stocks". A handful of accounts have, one of them heavily, and the rest of the index has not moved. That distinction is the same one that decided the meaning of our cancelled-outcome figures in Half the trades never happen, where a single publisher carried 1,501 of 1,958 cases.

Can a subscriber tell from the post?

No, and the clearest demonstration comes from two messages by the same channel a few days apart. Here is the first, from spnchl on 3 August 2026:

#BTC/USDT Direction: 🔐 Entry: 🔐 Stop Loss: 🔐

Target: 🔐 Leverage: x🔐

Details Available on VIP Channel

And the second, from 31 July:

#NVDA/USDT Direction: 🔐 Entry: 🔐 Stop Loss: 🔐

Target: 🔐 Leverage: x🔐

Details Available on VIP Channel

Swap the ticker and the two messages are character for character the same. One is bitcoin. The other is NVIDIA.

Filled-in calls look no different. The same channel posts #QQQ/USDT - Long🟢 at an entry of 679.4 with three targets and Leverage: x23. A Spanish-language channel, Libre_Indicadores_Futuros, posts #GOOGL/USDT - Corto🔴 at 356.03 with three targets and Apalancamiento: x14, in the same layout translated.

Those two are one feed in two languages, which is worth saying in an article about counting. On 3 August both published a redacted #TQQQ/USDT teaser in their own language eight seconds apart, at 07:47:03 and 07:47:11 — the same simultaneity test we use in Clone signal channels. Read them as two independent channels and you have counted one operation twice, which is the trap the table above is built to avoid.

Nothing in the format is doing any work here. Our separate count of ticker conventions across the corpus found the hash-and-slash form overwhelmingly dominant and the exchange's own string for a contract almost absent from the messages. A reader has learned to see #SOMETHING/USDT and read "a coin", because until this summer that inference was almost always right.

What are these tickers, then?

Each of the following was checked against a public source before it went in this table. Where a ticker could not be pinned down, we left it out entirely.

Posted as What it actually is
KORU Direxion Daily MSCI South Korea Bull 3X ETF, listed on NYSE Arca, seeking 300% of the daily move of the MSCI Korea 25/50 Index
SOXL Direxion Daily Semiconductor Bull 3X ETF, NYSE Arca, daily target +300%
SOXS Direxion Daily Semiconductor Bear 3X ETF, NYSE Arca, daily target −300%
SPY SPDR S&P 500 ETF Trust, benchmarked to the S&P 500, trading since January 1993
QQQ Invesco QQQ Trust, Series 1, tracking the Nasdaq-100
NVDA NVIDIA Corporation
GOOGL Alphabet Inc.
CRCL Circle Internet Group, Inc.
NBIS Nebius Group N.V.
ALAB Astera Labs, Inc.
SKHYNIX, also SKHY SK hynix Inc., Korea Exchange, code 000660. The corpus carries two strings for the one company, 16 calls and 7
ZHIPU Zhipu AI (now Z.ai, 02513.HK), the Chinese model developer that listed in Hong Kong in January 2026
XAU, XAG gold and silver, which both exchanges classify as commodity contracts

Two of the five most-called instruments in the set are Korean exposure. KORU is triple-leveraged Korea, and SKHYNIX is one of the two largest companies on the Korean market by value. Samsung Electronics and Hyundai appear further down the list under their own names.

How do the exchanges label these contracts?

Unambiguously, in machine-readable fields that never reach the Telegram post. Both venues carry a separate instrument class for them, and both say so in the same public endpoints anyone can query.

Field BTCUSDT KORUUSDT NVDAUSDT
Bybit symbolType empty ETF stock
Bybit max leverage 100x 25x 50x
Bybit copyTrading both none none
Bybit price limit ratio 1% / 2% 3% / 5% 3% / 5%
Binance contractType PERPETUAL TRADIFI_PERPETUAL TRADIFI_PERPETUAL
Binance underlyingType COIN EQUITY EQUITY

Read on 18 August 2026, Bybit had 716 USDT perpetual contracts in trading status, of which 155 are tagged stock, 41 ETF and 4 commodity. Binance had 169 contracts of type TRADIFI_PERPETUAL, 168 of them USDT-quoted, covering US, Korean, Hong Kong and Chinese equities alongside commodities. Two hundred instruments on one venue and 169 on the other add up to a product line rather than an experiment. It is also a recent one: the earliest launchTime across those 200 Bybit contracts is 9 March 2026, so none of this existed at the start of the year.

The copyTrading field is the exchange's own risk view written into metadata. Bybit runs its copy-trading product on bitcoin and does not offer these contracts inside it. Nobody acting on a Telegram post was in that product to begin with, so no rule is being circumvented; the flag records that the venue thinks this instrument sits badly with people who follow somebody else's positions.

What is different about the instrument itself?

Bybit publishes the answer in its help centre, and the parts a follower needs are stated plainly. The product is called a TradFi Perpetual Contract, and the exchange defines it this way:

"Bybit TradFi Perpetual Contracts are USDT-denominated and USDT-settled Perpetual Contracts that track the price of traditional financial (TradFi) assets. These contracts allow traders to gain exposure to the price movements of traditional assets without holding the underlying asset." — Bybit Help Center, Introduction to TradFi Perpetual Contracts

The margin, funding and liquidation machinery is the same as any USDT perpetual, which is the machinery described in Spot, futures and perpetuals. What changes is everything downstream of the underlying market being a stock exchange.

You own no share. The same page states that holders have no shareholder rights, no dividend entitlement and no voting rights, and that "unless explicitly reflected by Bybit through pricing or contract adjustments, users will not benefit from any corporate actions". Corporate actions are not hypothetical on this list: Direxion announced a reverse split of KORU in January 2025 and a forward split in June 2026.

The exchange warns about gaps. Its own risk disclaimer names two hazards with no crypto equivalent. The first: "Events such as earnings releases or major news may cause significant price gaps, which may trigger liquidation." The second applies while the shares are not trading, when "contract liquidity may be significantly reduced, resulting in wider bid-ask spreads". An earnings release has no equivalent in a bitcoin chart.

What happens while the underlying market is shut?

The contract keeps trading. That is the entire premise of the product, and Bybit states it as such: traditional markets "operate within fixed trading hours", and the exchange built these contracts "to provide 24/7 access to Futures on traditional assets".

NYSE's core session runs 9:30 a.m. to 4:00 p.m. Eastern, Monday to Friday, and NYSE Arca — where KORU, SOXL, SOXS, SPY and QQQ list — extends that with an early session from 4:00 a.m. and a late one to 8:00 p.m. Bybit counts those extended hours as market-open too, defining its open period as "Regular trading hours + Pre-market/After-hours + Overnight trading hours". The genuine gap is what falls outside all of it: the small hours, the published holidays, and every weekend.

Bybit describes what its index does in those hours. During market closure, if component prices "stop updating, the system may temporarily exclude those components from the index weighted-average calculation to reduce the impact of stale prices". The exchange also reserves the right, "at its sole discretion", to decide what counts as an opening period and a closed period. Those are reasonable engineering choices for a hard problem, and the exchange publishes them openly, which is where we read them. The post is the part that says nothing. A follower acting on #NVDA/USDT on a Saturday is inside a pricing regime the message never mentions.

The practical shape of it: a signal posted on Saturday afternoon names a price for something whose market last printed on Friday and will not print again until Monday.

What does a 3x ETF do inside a 25x contract?

It multiplies, for one day at a time. KORU is built to deliver a daily result, and Direxion says so on the fund page in both directions:

"This leveraged ETF seeks a return that is 300% of the return of its benchmark index for a single day. The fund should not be expected to provide three times the return of the benchmark's cumulative return for periods greater than a day." — Direxion, Daily MSCI South Korea Bull 3X ETF

Now put the contract on top. Bybit allows up to 25x on KORUUSDT, so a maximum-leverage position on a fund targeting 300% of the daily index move would be roughly 75 times a single day in the MSCI Korea 25/50 index. That figure describes what the contract permits, and our archive shows nobody near it: all 45 KORU calls asked for 10x, and the highest leverage anywhere in this set is the 23x on that #QQQ/USDT post. Costs stack more quietly, with the fund charging a 1.32% annual expense ratio internally while the perpetual charges funding every eight hours. What channels ask for across the whole corpus is in What leverage do signal channels actually ask for.

Direction is where the layering turns genuinely confusing, and the corpus shows it happening. Of the 45 KORU calls, 43 are shorts: a fund built to return three times a rising Korean market, sold short. SOXS splits 13 short against 5 long, and because SOXS is itself a tripled bear fund on semiconductors, a short on it is a tripled long on semiconductors two layers down. SOXL runs 15 long to 10 short. Working out which way a follower ends up facing takes the direction in the post, the sign of the fund and the fund's multiplier, all three at once.

Is any of this a scam?

Nothing here supports that word. A perpetual on Nvidia is a real product from two of the largest exchanges in the market, listed openly and documented at length. A channel calling it is trading something its exchange offers.

What we can say is narrower and still uncomfortable. Somebody who subscribed to a crypto signal channel is being handed exposure to Korean semiconductor manufacturing, and the post gives them no way to notice. This is a disclosure question and a suitability question. It sits alongside our standing position that a paid channel is not a fraudulent one and that a subscription price proves nothing either way, set out in more detail in How to verify a crypto signal channel before you pay.

Intent is also outside what we measure. A channel might be following the liquidity, might be chasing an AI-shares news cycle, or might simply post whatever its charting tool flags. Our archive records what was published.

What this does not prove

The index is not the market. We parse 22 accounts out of the 47 in our index, and the 235 come from seven of those 22. Channels that publish screenshots or free-form commentary contribute nothing to these counts and are not a random sample.

One account decides the headline. At 181 of 235, account #1's habits are three quarters of the finding. The clone cluster distorts the rest: three of the seven publishers are one feed, so the number of independent operations behind this behaviour is five at most.

The ticker list is conservative and therefore incomplete. Strings such as SNXX, MVLL, MUU, INTW and SHAZ are excluded because they are ambiguous, which means the true figure sits somewhere between 5.8% and 8.3% and we cannot narrow it further from a symbol string alone.

Exchange listings are today's. The instrument metadata, the leverage caps and the counts of TradFi contracts were read on 18 August 2026 and change as venues list and delist. A symbol absent today may have been listed when a signal was published, so nothing above is a claim about what any venue offered in the past.

We measured strings, not executions. Our record holds what the channel wrote. It does not show which venue any follower used, what they filled, or whether they knew what they were trading, and no claim above should be read as measuring any of that.

The counts move. Prod figures are dated 18 August 2026 and shift as signals resolve and channels publish.

The practical read

A ticker you do not recognise is now a question rather than a shrug, and answering it takes under a minute.

Scroll the last thirty calls in a feed you pay for and search every string that means nothing to you. One that comes back as a fund prospectus or an investor relations page has told you something the post withheld.

If it turns out to be a share or a fund, read the contract before the trade. Both venues answer in one public request and neither asks for an account. Bybit's is api.bybit.com/v5/market/instruments-info?category=linear&symbol=KORUUSDT, where a symbolType of stock or ETF settles it outright. Binance's is fapi.binance.com/fapi/v1/exchangeInfo, where the same instrument comes back as contractType: TRADIFI_PERPETUAL with underlyingType: EQUITY.

Then ask what the calendar does to the call. A weekend signal on an equity contract quotes a price for a market that last printed on Friday, and an earnings date inside the holding period is a gap no chart can warn you about.

One last illustration of how quietly this arrives. KORUUSDT already sits in the published top-fifteen symbol table of Bitcoin is 0.8% of the signals, listed at 44 calls from a single channel, reading like one more obscure altcoin. It is a triple-leveraged fund on the Korean stock market, and this article is the footnote that table needed.

What a channel's advertised accuracy figure survives when replayed is in How accurate are crypto signals. Current ratings for the channels we can score are at Signal Providers. Nothing here recommends any channel, any asset, or any trade.

Sources

  • Corpus figures — 4,030 parsed signals from 22 publishing accounts, the 235 non-crypto calls and their 131/94/10 split, the seven-account concentration table, the conservative monthly split of 17/141/77, the first-dated NVDAUSDT signal of 17 June 2026, the 90-day per-symbol counts with their publisher bases, and the direction and leverage splits for KORU, SOXL, SOXS and QQQ: work/_batch10-shared-research.md, section 4, measured read-only against the production index on 18 August 2026. Ticker-writing conventions across the corpus are counted in the same file, section 1.
  • Quoted messages — the byte-identical #BTC/USDT and #NVDA/USDT teasers from spnchl (3 August and 31 July 2026), its #QQQ/USDT call of 28 July, the #GOOGL/USDT call from Libre_Indicadores_Futuros of 11 August, and the paired #TQQQ/USDT teasers of 3 August eight seconds apart: data/channel_dumps/, the repository's archive of 42 Telegram channels.
  • Bybit Help Center: Introduction to TradFi Perpetual Contracts — definition, no-ownership disclaimer, the market-open period covering pre-market and after-hours, index price during market closures, gap and liquidity risk warnings.
  • Instrument metadata, read 18 August 2026: Bybit v5/market/instruments-info for symbolType, leverage caps, copyTrading, price limit ratios and the launchTime of every stock, ETF and commodity contract; Binance fapi/v1/exchangeInfo for contractType and underlyingType.
  • Direxion: Daily MSCI South Korea Bull 3X ETF (KORU) — daily 300% objective, MSCI Korea 25/50 benchmark, NYSE Arca listing, 1.32% expense ratio, split announcements.
  • Direxion: Daily Semiconductor Bull and Bear 3X ETFs (SOXL, SOXS) — ±300% daily targets and NYSE Arca listing.
  • State Street: SPDR S&P 500 ETF Trust (SPY) — S&P 500 benchmark, inception 22 January 1993.
  • Invesco QQQ ETF — tracks the Nasdaq-100 Index.
  • SEC company tickers file — official ticker-to-registrant mapping for NVDA, GOOGL, CRCL, NBIS, ALAB, SPY and QQQ.
  • SK hynix Inc. (000660.KS) — company name, Korea Exchange listing and code.
  • Z.ai listed on the Hong Kong Stock Exchange — Zhipu AI's January 2026 listing.
  • NYSE: Hours and Calendars — core session 9:30 a.m. to 4:00 p.m. ET, the NYSE Arca early and late sessions running 4:00 a.m. to 8:00 p.m., and the published holiday schedule.