bStocks - The 3 a.m. stock market

The busiest hour in Binance's tokenized equity market is 07:00 UTC. In New York that is three in the morning. The NYSE closed eleven hours earlier and will not ring the opening bell for another six and a half. In Hong Kong it is mid-afternoon, and traders there are working through a normal day.

By 21:00 UTC, an hour after the American closing bell, volume has collapsed to a seventeenth of that peak. The quietest hour of the day for tokenized US stocks is early evening in the city where those stocks are listed.

We indexed every bStock trade on BNB Smart Chain between 30 June and 29 July: 4.42 million fills, 17 contracts we authenticated one by one — 16 of which traded inside the window — $2.46 billion in notional. The single clearest number to come out of it is this. 91.2% of on-chain bStock volume prints while US cash equity markets are shut. Only $216.6M of $2.46B changed hands during the 9:30 to 16:00 session.

Binance launched bStocks on 11 June and has marketed them on continuous access: hold Nvidia or Tesla as a BEP-8056 token, redeem into the underlying share, trade whenever you want. Traders took that literally. The overnight window carries very nearly the entire venue.

Which raises the question we spent most of our time on. If nine dollars in ten trade with no reference market open, what is the price actually doing?

Section onePeak hour is three in the morning on Wall Street

BNB Smart Chain settles a block roughly every three quarters of a second and the pools accept orders continuously, so the question was never whether you can buy the Nasdaq at 3 a.m. You can. It is who does, and what the tape looks like when they do.

Section twoNinety-six percent of this market is one ticker

Before the price question, a correction to the mental image. When Binance talks about bStocks it talks about Nvidia, Tesla, Apple, Meta, SpaceX. When you sort the tape by volume, almost none of that is there.

QQQB, the wrapper on the Invesco QQQ Trust, is 96.2% of all on-chain bStock volume. Every other bStock that traded put together is the remaining 3.8%.

Tokenized Tesla, the name that leads most of the press coverage, did $42,614 across the entire month. Four hundred and eighty-eight wallets touched it. The median TSLAB fill was ten dollars. Tokenized Meta did $1,186. Robinhood did $565 across nine wallets. Tokenized AMD has 1,153 holders and traded $778, which works out to about sixty-seven cents per holder for the month.

Section threeA treadmill running under a fixed price

Here is where the data stopped agreeing with what we expected.

The obvious reading of a market that trades 91% of its volume after hours is that after hours must be where the action is. Thin books, no reference price, retail gambling on gaps. We measured realized volatility of hourly VWAP for each ticker, split by session, expecting the closed-market bars to tower over the open-market ones.

They do the opposite. QQQB moves 27% less per hour when US markets are closed. SPYB moves 43% less. The overnight and weekend market, which carries the overwhelming majority of the flow, produces less price movement per hour than the sliver of volume that trades during the session.

The weekend numbers make it starker. Over a full Saturday and Sunday, with every equity market on the planet shut, we measured QQQB's entire hourly price range at between 0.36% and 1.24%. SPYB never exceeded 0.71%. Millions of dollars crossing every hour, and the price sits almost perfectly still. The QQQB/USDT book shows the same flatness live, if you want to watch it happen over a weekend.

Two tickers invert the pattern, and they are the interesting ones. SPCXB and CRCLB are roughly two and a quarter times more volatile when markets are closed. CRCLB is a small, thinly arbitraged book. SPCXB is SpaceX, which has no public market at any hour of any day. Nothing pins it, so it wanders, and it wanders hardest when liquidity elsewhere is at its thinnest.

Those two exceptions tell you how the rest of it works. Where a live reference price exists, the after-hours bStock market is a treadmill: real money changing hands around a number that Binance's own order book is holding in place. Where no reference price exists, the token has to find its own level, and the result is visibly messier.

Section fourThe one thing people actually want is not a stock

Holder counts and volume tell two different stories, and the gap between them is the most commercially interesting thing in this dataset.

SPCXB has 64,742 holders, which is 73.9% of every bStock holder position across all seventeen contracts. Tokenized SpaceX, a company that has never listed, is what three quarters of this user base owns.

The trade sizes fit that. SPCXB's median fill is $12. QQQB's is $446. One of those looks like a bot working an index book; the other looks like people buying a small piece of something they have wanted for years and cannot get through a broker.

The product-market fit here looks like access to companies the traditional market will not sell you. Apple, which anyone with a brokerage account can already buy at will, has 44 holders and did not trade once in thirty days. That is a smaller and stranger business than "the stock market, on chain," and it is the one the data supports.

Section five1,307 impersonators, and a victim list that will surprise you

Alongside the seventeen real contracts, we found 1,307 impersonators trading under bStock ticker symbols on BNB Chain. They took $324.8M, or 11.66% of all volume transacting under those symbols.

Most are junk: BEP-20 tokens with a hundred trillion supply, twenty holders, one Uniswap pool, and a name like "Amazon Tokenized bStocks." For fifteen tickers, including COINB, AMZNB, PLTRB and MSTRB, every single dollar of on-chain activity is fake. No genuine contract for those names trades on chain at all.

Two contracts are a different class of work, and between them they account for 94% of all counterfeit volume. The SK Hynix job is worth walking through.

A near-identical contract ran the same play against QQQB on 19 July: 450,000 supply, 1,382 holders, $126.4M in a single day. Same operator, on any reasonable reading.

On the fake SK Hynix, 5,411 wallets traded. 2,468 of them finished down more than $100, for aggregate losses of $58.7M. The worst single wallet lost $148,737.

Then we checked which of those wallets had ever traded a real bStock, expecting substantial overlap. There is almost none.

Section sixThey came for a ticker symbol

That changes what kind of problem this is. The counterfeits are not harvesting confused Binance customers, because Binance's customers are barely present. A separate population of 8,124 wallets found these tokens through some other door: a screener, a symbol search, a trending list, a Telegram call. The $58.7M lost on the fake SK Hynix was lost by people who had never used bStocks in their lives.

Which puts the burden somewhere uncomfortable for us, because the doors those wallets came through are largely built by data providers. Bitquery is one. Any index, screener or API that ranks tokens by symbol will hand a user the scam first for SKHYB, COINB, PLTRB and a dozen others.

The filter that works is the token standard rather than the name: every genuine bStock is BEP-8056 and carries BscScan's bStocks and Custodial Model labels, and every fake we verified was a plain BEP-20.

Section sevenWho is on the other side of your fill

A market where 91% of volume prints overnight, the price barely moves, and one ticker is 95% of the flow raises an obvious question about counterparties.

Widen the lens and the picture holds. 184 wallets, 0.24% of the population, account for 17.2% of volume. A separate cluster of 199 addresses, identified by genuine multi-transaction round trips, moves $132.6M — 5.4% of the market — between them.

Against those counterparties, retail does about as well as you would expect. We split all 78,198 wallets by how much they traded and measured how many finished more than $100 ahead.

The gradient runs from 10.4% to 59.5% without a single step backwards. The 74 wallets that traded more than a million dollars took $10.7M between them. The 28,127 wallets that traded under a thousand dollars gave up $282,306.

Nine in ten small traders finished flat or down. We are wary of the aggregate dollar figures here, because a wallet that withdraws bStocks from Binance and sells them on chain shows up as pure profit in on-chain accounting, and that flatters the big wallets. It cannot explain a 10.4% win rate among twenty-eight thousand small ones, though. Those wallets are not withdrawing inventory to distribute. They are just losing.

Churn matches. 37.6% of all wallets appear on exactly one day and never come back, contributing 4.3% of volume. The population that actually sustains this market is 15,309 addresses alive between fifteen and twenty-eight days, and they carry 54.2% of the money.

In closingThe read

Forty-eight days in, bStocks work, though not in the way the pitch describes.

The 24/7 promise is real and heavily used, overwhelmingly by people in Asian and European time zones for whom the US session falls in the middle of the night. What they get during those hours is liquidity. The price itself is pinned by Binance's own book, and the DEX side is a well-arbitraged treadmill running underneath it, at least for the index wrappers. That is a legitimate service, and a different one from a stock market that never closes.

The demand signal that does show up clearly is SpaceX, and it points somewhere other than tokenized Apple. People are paying for access to things they cannot otherwise buy. The single-name US equities, which are the easy part of the product and the loud part of the marketing, are close to dead on chain.

And the counterfeit economy attached to all this is larger than the real thing for fifteen of the tickers, took $324.8M in a month, and is hurting a population that never touched the product. That last part is the finding we would most like other people to check.