Weekend token prices called 54% of Monday opens. Guessing that every stock would open higher called 55%
Crypto never closes. We scored what Sunday prices of 121 tokenized stocks say about Monday, against guesses that use no weekend price at all.
- 54%
- 53%
- 57% / 49%
- 0.9%
- 0.3 points
01 · The Sunday priceA live price while Wall Street is shut
The US stock market shuts at 4pm on Friday and opens again at 9:30am on Monday. Crypto keeps trading all weekend. So do tokens that track shares like Nvidia and Tesla.
That gives those stocks a live price on Sunday evening, and it looks like a free preview of Monday's open. On 5 October we reported that the preview pointed the right way 56% of the time. A coin toss gets 50%.
But a coin toss is too easy a bar. Across the 17 weekends we have now scored, 55% of the Monday opens went up. So a guess that every stock will open higher, which we call always up, beats 50% without looking at any weekend price. We added the four weekends since 5 October and scored the Sunday price against that guess. The Sunday price showed no clear edge. On the size of the move, it did worse than guessing no change at all.
02 · How tokenized stocks workA share in custody, a price set on-chain
A tokenized stock is a token that tracks one company's share. The issuer, the company behind the token, says it holds the real share in custody. Because the token lives on a blockchain, it can change hands on a DEX at 3am on a Sunday.
On a weekday, arb traders keep each token pegged to its share. They buy when it trades below the share price and sell when it trades above. On a weekend the share can't trade, so there is nothing to arb against, and the token's price is whatever weekend buyers and sellers agree on.
We covered the same six issuers as before: Robinhood, Binance, Coinbase, xStocks, Ondo and Backpack. Tokenized stocks from the six largest issuers were worth about $3.2 billion at the end of October, by our market count. An earlier piece found the pools behind these tokens too thin for a $10,000 order. Our October report shows where they trade, chain by chain.
Some stocks have tokens from several issuers, and we blend them into one Sunday price per stock. For Nvidia, our piece on six tokens for one stock found four issuers' tokens within 0.12% of each other.
03 · How we scored itThree simple guesses to beat
For each stock and each weekend we compared two moves. One is the token's move from Friday afternoon to Sunday evening. The other is the stock's move from Friday's close to Monday's open. If both went the same way, the token called it.
We took the Sunday reading just before 6pm, New York time. That's when US stock futures, contracts on the S&P 500 and Nasdaq indexes, start trading again. After that, traders can simply copy the futures.
We scored the Sunday price against three guesses that need no weekend data at all. A coin toss is right half the time. Always up says every stock opens higher on Monday. No change says every stock opens at Friday's close; it has no direction, so it only competes on the size of the move.
Tiny moves are hard to call. For direction, we only scored opens that moved at least 0.5% from Friday's close. That kept two thirds of them.
04 · The result54% for the tokens, 55% for always up
Sunday token prices called 54% of Monday opens right. That share is their hit rate. Always up had a hit rate of 55%, so on this count it wins by a point.
That point means little. Results swing a lot from weekend to weekend. So each score comes with a range, where the true figure could sit. If the range covers 50%, or zero for a gap, luck alone could explain the result. Here the gap could sit anywhere from about 20 points in always up's favor to 16 in the tokens' favor.
Comparing hit rates is a blunt test with only 17 weekends. Even a true 20-point lead would show up clearly only about six times in ten. A 10-point lead would usually slip past it.
Always up has a head start when more stocks rise than fall. To take that out, we also scored rises and falls apart and averaged the two hit rates. This is called balanced accuracy. A coin toss gets 50% on it, and so does any fixed guess, such as always up.
The tokens beat a coin toss by 7 points on rises and trailed it by 1 point on falls. That averages out at a balanced accuracy of 53%. It puts the tokens 3 points ahead of always up, where the plain count had them a point behind. Both gaps sit well inside their ranges. The balanced score could be anywhere from 44% to 62%, and that range includes the 50% any fixed guess gets. So the tokens show no clear edge.
05 · Why 56% became 54%One-sided Mondays move the bar
On the weekends in the 5 October piece, re-checked here, the tokens still score 56%. But always up scored only 51% on those weekends, 5 points behind them. On the four weekends added since, most stocks opened higher and always up pulled ahead: 63% against 48% for the tokens. Those four weekends flipped the lead.
Always up's score swings with the market's mood. On the weekend of 18 September, all but four of the 58 stocks we scored opened higher. A week later, only nine did. Always up went from nearly perfect to mostly wrong, while the tokens stayed near half both times.
Weekends like these are common. In six of the 16 weekends with at least eight opens scored, three quarters or more of the stocks went the same way. On weekends like that, always saying up looks clever one week and foolish the next.
06 · ChecksTwo checks on the result
US markets were shut on Monday 7 September for Labor Day, so that weekend was scored against Tuesday's open. Leaving it out changes little: the tokens' balanced score stays at 53%, with a range of 43% to 63%.
The second check waits for futures. It takes the Sunday price just before 8pm and keeps only stocks whose tokens traded $1 million or more over the weekend. It also counts an open if either the stock or its token moved at least 0.5%. On that set, the balanced score was 59%, with a range of 53% to 65%. By 8pm a token can copy the futures, so this check may measure the futures as much as the token.
07 · The size of the moveNo change was the closer guess
Direction is half the question. A trader also wants to know how big the move will be. No change, the guess that each stock opens right at Friday's close, missed by a median 0.9%. Using the token's weekend move as the forecast missed by a median 1.2%. The weekend move made the forecast worse.
The weekend moves were about as big as the moves at the open, a median 0.8% against 0.9%. They just pointed the wrong way, or went too far, too often. On 62% of opens, adding the token's move took the forecast farther from Monday's open. Friday's close alone was nearer.
On size, no change wins.
08 · What it meansA Sunday price says little about Monday
A tokenized stock's Sunday price is one you can trade at, but as a forecast of Monday's open it adds little. Once you allow for how many stocks open higher each weekend, its calls on direction could be luck. On size, it missed by more than a guess of no change.
The 5 October piece found that most of Monday's move reaches token prices only after futures reopen at 6pm on Sunday. For anyone trading a stock token on a Sunday afternoon, the price is the weekend crowd's guess. The first real hint of Monday comes when futures reopen.
09 · How we did itMethod
We used the same list of tokens, sources and price rules as the 5 October piece. Token prices come from our trade data on five chains: BNB Chain, Base, Robinhood Chain, Solana and Ethereum. Each hour's price is the median of stablecoin trades, with bigger trades counting more. Stock prices come from Yahoo Finance, checked against Nasdaq.com.
We left out tokens with less than $10,000 traded over the weekend. We also left out stocks that split or went ex-dividend on the Monday. Either one moves the open for reasons a weekend price cannot see. Stocks whose value rides on coins that trade all weekend are out too, such as Strategy (MSTR) and Coinbase's own stock (COIN). That left 121 stocks.
We fixed the guesses, the scoring and the order of reporting before pulling the October data. By then we had seen the 5 October piece's results and the counts behind them: how many stocks opened higher each weekend, and the median misses on size. We had also seen a rough tally of the first three October weekends. It came from an internal tool with a different price feed.
We also sell the full trade history behind these figures as files on the Bitquery Data Store: BNB Chain, Base, Robinhood Chain, Solana and Ethereum. Developers can pull these prices with our Crypto Price API, or ask for them in plain English through the Bitquery MCP.
10 · The recordWeekend by weekend
| Weekend (Friday) | Opens scored | Opened higher | Token price right | Guess of up right |
|---|---|---|---|---|
| 10 Jul | 3 | 1 | 0 | 1 |
| 17 Jul | 9 | 8 | 6 | 8 |
| 24 Jul | 17 | 15 | 14 | 15 |
| 31 Jul | 18 | 8 | 7 | 8 |
| 7 Aug | 10 | 4 | 4 | 4 |
| 14 Aug | 10 | 6 | 5 | 6 |
| 21 Aug | 16 | 3 | 9 | 3 |
| 28 Aug | 20 | 6 | 9 | 6 |
| 4 Sep (Labor Day, Tuesday open) | 61 | 37 | 37 | 37 |
| 11 Sep | 70 | 27 | 46 | 27 |
| 18 Sep | 58 | 54 | 25 | 54 |
| 25 Sep | 58 | 9 | 32 | 9 |
| 2 Oct | 32 | 18 | 20 | 18 |
| 9 Oct | 31 | 10 | 10 | 10 |
| 16 Oct | 50 | 37 | 14 | 37 |
| 23 Oct | 33 | 19 | 25 | 19 |
| 30 Oct | 37 | 29 | 23 | 29 |
| All 17 | 533 | 291 | 286 | 291 |
FAQ
How does a tokenized stock's Sunday price compare with guessing up?
Over 17 weekends from July to October, Sunday token prices called the direction of 54% of Monday opens that moved 0.5% or more. Guessing that every stock would open higher called 55%. A score that allows for how many stocks rise put the tokens within what chance produces.
Why does the test stop at 6pm on Sunday?
US stock futures start trading again at 6pm New York time. After that, a token's price can simply copy the futures, so a later reading would test the futures as much as the token.
How do tokenized stocks work?
An issuer says it holds a company's share in custody and puts a token that tracks the share on a blockchain. The token can then trade on crypto exchanges and DEXs at any hour, weekends included.
What is balanced accuracy?
A score that rates rises and falls separately and takes the average of the two. Any fixed guess, such as always saying up, gets 50% on it. That makes it a test of whether a forecast knows something beyond how many stocks rise.
Why is 54% lower than the 56% reported on 5 October?
The four weekends added since gave the tokens 48%, which pulled the total down. The earlier weekends still give them 56%.
Weekends run from Friday 10 July to Friday 30 October 2026. Each is scored against the next opening price; the last was Monday 2 November. Binance and Robinhood cover every weekend, and Coinbase every weekend from 21 August. xStocks, Ondo and Backpack start on 11 September, the earliest weekend our trade data held for them when the 5 October piece was built.
Token prices are dollar-weighted medians of stablecoin trades in one-hour windows. A listing counts only if it traded in the hour before each cutoff and did at least $10,000 over the weekend. Each stock gets one value per weekend, with each issuer weighted by its weekend dollar volume. A token move of exactly zero has no direction and counts as a miss. The token's Friday price comes from its last hour of stock-market trading, which can differ from the 4pm close; in a sample of 20 stocks on October Fridays, a stock's average price in that hour sat a median 0.2% from its close.
Ranges are 95% ranges and treat each weekend as one unit, because stocks in the same weekend move with the market. Balanced accuracy uses a t-distribution across weekends, with 16 degrees of freedom (15 without Labor Day). The gap against always up and the size gap use 10,000 resamples of whole weekends. The chance of spotting an edge of a given size comes from 4,000 simulated runs per edge, each resampling whole weekends.
The weekend chart grays out weekends with fewer than eight opens scored and leaves them out of the count of which guess led.
On the re-checked daily prices used here, the 13 earlier weekends give the tokens 214 and always up 196 of 382 opens, against 213 and 197 on the prices the 5 October piece used. That piece took the first five-minute price for the opening on Monday 5 October, where the daily opening price is used here, and one stock drops out of one weekend for an ex-dividend date.
The 8pm check covers 220 opens: the tokens called 132 and always up 116. We did not compare token prices with stock futures. The 30 October weekend ran into the end of US daylight saving time; the 6pm Sunday cutoff stayed at 6pm New York time, an hour later in UTC than before.
This article is provided for informational and educational purposes only and reflects analysis of publicly available on-chain data as of the dates indicated. It does not constitute legal, financial, compliance, or investment advice, and nothing in it is a recommendation to buy, sell, or hold any token or asset.
The findings describe trades observed on public blockchains between 10 July and 2 November 2026. They may be incomplete or incorrect and may be revised as more data becomes available.
References to Robinhood, Binance, Coinbase, xStocks, Ondo, Backpack, Yahoo Finance, Nasdaq or any other named company describe what the public record shows about tokens, trades and prices. They are not statements about any party's compliance, solvency or conduct.
Nothing herein should be relied upon as a definitive determination of fact. Readers should conduct their own independent verification before taking any action. The authors and publisher accept no liability for any loss or damage arising from reliance on this material. All trademarks and company names are the property of their respective owners.
Reported by Gaurav Agarwal for Bitquery Research, with AI tools; every figure was checked against the raw data.
Ask these questions in plain English
The trades counted here come from Bitquery's DEX data across five blockchains. The Bitquery MCP server puts that data behind an AI assistant, so you can ask what a token traded at in any hour of a weekend without writing the query yourself.
Run the next investigation in minutes
Ask any question of the blockchain. Bitquery's AI Investigations agent traces funds, attributes wallets, and writes auditable reports across every major chain. Prefer to do it yourself? Connect the Bitquery MCP server to your AI assistant, pull the raw datasets, or query the chain directly with the Bitquery API.