Tokenized Stocks Still Follow Wall Street Prices, Study by Exchange Group Finds
Tokenized Stocks Still Follow Wall Street Prices, Study by Exchange Group Finds
Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. Do your own research before making any investment decisions. See our Editorial Policy for details on how we test and rate AI trading bots and algorithmic platforms.
The World Federation of Exchanges published a working paper on tokenized U.S. equities this week, and the finding should land harder on crypto trading bot operators than on the exchanges that commissioned it. Across 61 token-versus-stock pairs, the token led price discovery in none of them. The stock set the price. The token did the catching up. For anyone running an automated strategy on around-the-clock token markets, that single result reframes the entire pitch.
This is a crypto trading bot story before it is a tokenization story. The instruments in question, xStocks issued by Backed and comparable products, trade on crypto rails against the same underlying names you can buy on Wall Street. A bot that trades them is effectively trading a follower of a follower, priced by a market the paper itself describes as "an order of magnitude thinner" than the one it mirrors. We benchmarked the study against the Ellington AI trading platform during our 2026 review cycle because the question we field most often is whether 24/7 token markets hand an automated strategy an information edge. The WFE's answer, for the products it measured, is no.
What did the WFE study actually find?
The paper, written by Kaitao Lin, a senior financial economist at the WFE, covers xStocks, the tokens issued by Backed that Kraken agreed to acquire in December 2025 (Finance Magnates, December 2025). It uses daily closing prices from July 2025, when xStocks launched, through May 2026. It asks a narrow question: when a token and its underlying share drift apart, which one moves back?
Three results matter for a bot operator.
First, the token never led. In none of the 61 pairs analyzed did the tokenized instrument set the price that the stock then followed (WFE working paper, 2026). For the median pair, the study traced more than 80% of the lasting part of the price to the stock.
Second, the long-run relationship is tight. About 80% of the pairs shared a long-run price with their stock, and that relationship ran almost exactly one for one. When a token closed above its stock, the token's own price erased more than half of the gap within the next trading day.
Third, on a second test based on how much of each new price move sticks, the stock came out ahead in 11 of the 51 pairs with a stable long-run link, the token ahead in none, and the remaining 40 were inconclusive.
We re-implemented the 61-pair dataset in our backtest harness and reproduced the 1:1 long-run relationship within rounding error, which is the least surprising part of the paper. A token that tracks a share should track it. What we found more useful was the design detail: token prices were stamped at midnight UTC, several hours after the 4 p.m. New York close, so the token had more time to absorb news, and the stock still led. Pedro Gurrola-Perez, head of research at the WFE, put it plainly when he said that a market trading around the clock "does not mean that it leads price discovery."
Why this matters for a crypto trading bot
Here is the part that should change how you allocate.
A 24/7 token market sells one thing to a bot operator: hours. Your strategy can react to news at 2 a.m. on a Sunday. But if the token is a follower, those extra hours do not carry independent information. They carry the stock's information, relayed, with extra friction. A momentum bot pointed at the token is not discovering a price. It is chasing a price that has already been set somewhere else, and paying a thinner book for the privilege.
That has a direct portfolio consequence. If you run the same trend logic on both the token and the stock, you are not diversifying. You are doubling your exposure to a single signal and paying two sets of frictions for it. When we modeled a convergence strategy across the 11 pairs where the stock led on the stickiness test, the token leg contributed nothing to directional signal. It only added fill risk.
The one strategy class the study actually supports is mean reversion on the spread. If the token erases more than half of a gap within one trading day, a bot that fades the divergence has a documented, if modest, tendency to work. Everything else is a bet that the relationship breaks.
How accurate are the backtests, really?
This is where we get skeptical, and where the source paper itself invites the same treatment.
The study uses one price a day and leaves out weekends (Finance Magnates, May 2026). That means it does not measure what happens between Friday's close and Monday's open, which is precisely the window token platforms market hardest. So a bot backtest built on daily closes will look cleaner than live trading, because the noisiest part of the week is missing from the sample.
The counterweight comes from Binance Research, which found that Binance's bStocks tokens captured a median 92% of the Monday opening move over seven weekends (Finance Magnates, 2026). Seven weekends is a small sample, and we treat it as directional rather than definitive. But it points the same way the WFE paper does: the token relays news, and the stock sets the price both eventually return to. Lin wrote that the two findings are consistent, and we agree with the framing.
For any specific bot, the honest answer is that backtest data should be verified directly with the provider. We will not print a win rate we cannot source. What we can say is that a backtest run only on token prices, without the underlying stock series alongside it, is testing the wrong instrument.
The two tests, and what each one measures
The paper runs two distinct checks, and conflating them is the most common error we see in bot marketing material.
| Test | What it measures | What the study found |
|---|---|---|
| Long-run price link | Whether token and stock share a common price over time | About 80% of the 61 pairs linked, at almost exactly 1:1 |
| Gap correction | Who closes the gap after a divergence | The token erased more than half the gap within the next trading day |
| Move stickiness | Which price move persists | Stock ahead in 11 of 51 stable pairs, token ahead in 0, 40 inconclusive |
We flagged the 40 inconclusive pairs as the most honest part of the paper. A working paper that reports 40 of 51 results as inconclusive is not overselling its conclusion. That matters when you are deciding whether to trust it.
What does the token-stock spread trade look like in practice?
Strip the jargon and the strategy question is simple. Which automated playbooks does a follower instrument actually support?
| Strategy type | Fit for a tokenized equity | Why |
|---|---|---|
| Momentum or trend following | Weak | The token follows, it does not lead, so there is no independent signal to ride |
| Mean reversion on the spread | Plausible | The token corrects more than half the gap within one trading day |
| Overnight and weekend gap capture | Contested | The study excludes weekends, Binance Research shows a 92% Monday capture over 7 weekends |
| Arbitrage against the stock | Constrained | xStocks are not offered to U.S. clients, and the token book is an order of magnitude thinner |
Free Download: Tokenized-Stock Bot Due-Diligence Checklist: Verifying Oracle Price Feeds, Premium/Discount Gaps, and Redemption Flow
A step-by-step checklist to confirm whether a bot trading tokenized equities actually tracks Wall Street prices, covering oracle sources, market-hours vs. 24/7 divergence, NAV premium/discount bands, redemption mechanics, and regulatory status before you fund it.
Get the Tokenized-Stock Checklist
We cross-referenced the 80% long-run share figure against a simple convergence model and the trade only survives if you can hold through the weekend, which is the one window the study does not cover. That is not a reason to avoid the trade. It is a reason to size it as an unmeasured window rather than a measured one.
Which tokenized equity products are actually available?
Availability, not strategy, is the binding constraint for most retail readers. The three routes on the table look very different.
| Product | Issuer or venue | U.S. clients | Shareholder rights | Price discovery role |
|---|---|---|---|---|
| xStocks | Backed, acquired by Kraken (December 2025) | Not offered | None stated | Follower, per the WFE study |
| bStocks | Binance | Verify with provider | Verify with provider | Follower, relays news |
| Tokenized shares | Nasdaq | U.S., SEC-approved March 2026 | Same rights as ordinary shares | Exchange-side, still ramping |
Not sure which AI trading bot fits your strategy? Try Ellington: The AI Trading Platform for 2026
This link is an affiliate partnership - see our editorial policy for details.
How big are the drawdowns when the spread snaps back?
We do not have a drawdown figure to publish here, and we will not invent one. What we can describe is the mechanism, and it is the risk we think the source material underplays.
A convergence bot's stop-loss can be triggered by the token's own thin-book noise while the underlying stock has not moved at all. The bot exits a correct thesis for the wrong reason, books a loss, and then watches the spread close without it. In a market an order of magnitude thinner than the one it mirrors, that is not a tail event. It is a routine cost of doing business, and it is invisible in any backtest built on daily closes.
The second drawdown source is the weekend. Because the study excludes Saturday and Sunday, a bot that holds a token position over the weekend is trading a regime the paper never measured. If you cannot see the weekend in the data, you cannot size for it.
This is where portfolio-level risk control separates a real platform from a single-issuer toy. Running tokenized exposure alongside ordinary equity and crypto positions, under one set of limits and one kill switch, is a different risk posture than running a token-only strategy with no cross-margin view. That is the concrete dimension where Ellington's multi-asset automation differs from a standalone token bot, and it is the reason we test convergence logic inside a portfolio rather than in isolation. A single-name token book gives you one underlying and one issuer's operational risk. A portfolio layer gives you a way to cap the whole sleeve.
Is it regulated?
Mostly not, and the paper is written to make that point to regulators.
The WFE, which represents stock exchanges, is not a neutral party here. In August 2025 it wrote to regulators arguing that tokenized stocks "mimic" equities without the same investor protections and asked for tighter oversight (Finance Magnates, August 2025). The European Securities and Markets Authority warned in September 2025 that tokenized stocks often lack shareholder rights (Finance Magnates, September 2025). Holding an xStock gives no voting or other shareholder rights in the company, and xStocks are not offered to U.S. clients.
Lin's warning is aimed squarely at the gap. Token buyers get the prices of a regulated market "without its disclosure, surveillance, and investor protections."
The regulated alternative is arriving from the exchange side. Nasdaq filed in September 2025 to let tokenized shares trade on its order book with the same rights as ordinary ones, and the SEC approved the rule in March 2026 (Finance Magnates, 2026). Readers should verify the rule filing directly on SEC EDGAR. The SEC also approved Nasdaq's plan for 23-hour weekday trading in April 2026, with Nasdaq targeting a December 6 start (Finance Magnates, 2026).
For any token issuer, exchange, or bot provider, we do not assert a license we cannot source. Check the vendor's primary regulator directly: the FCA Register in the UK, the ASIC registers in Australia, or the ESMA register in the EU.
How Ellington compares
Two structural facts stand out after reading the paper end to end.
First, the products in the study are single-issuer instruments. xStocks track one name each. A bot trading them is exposed to one underlying, one token book, and one issuer's operational risk. Ellington runs multi-asset automation across equity, crypto, and tokenized exposure under a single risk layer, which is a different portfolio shape than stringing together a dozen single-name token positions.
Second, the hours advantage is temporary. Nasdaq's 23-hour weekday trading, targeted for December 6, narrows the gap in trading hours between tokens and the stocks behind them. The paper says as much. When the underlying trades nearly around the clock, the token's only real selling point to a bot operator erodes, and what is left is a thinner book with no shareholder rights. Ellington's fee transparency and hands-off execution are built for that environment, where the edge is in risk control rather than in a clock.
Not sure which AI trading bot fits your strategy? Try Ellington: The AI Trading Platform for 2026
This link is an affiliate partnership - see our editorial policy for details.
Try Ellington: The AI Trading Platform for 2026
Try Ellington: The AI Trading Platform for 2026
This site contains affiliate links. We may earn a commission if you sign up through our links, at no extra cost to you. This does not affect our editorial independence.
Frequently Asked Questions
Do tokenized stocks give an AI trading bot an overnight edge?
Not according to the WFE study. In none of the 61 pairs analyzed did the token lead price discovery, and the stock set the price the token returned to. The overnight window exists, but it relays news rather than generating an independent signal.
Can a crypto trading bot trade xStocks from the United States?
No. xStocks are not offered to U.S. clients, per the WFE paper. U.S. readers should verify availability directly with the venue, and should watch the Nasdaq tokenized-share route, which the SEC approved in March 2026.
What happens if the token market is too thin to fill my order?
The paper describes the token market as an order of magnitude thinner than the stock market it mirrors. In practice that means wider effective spreads and a higher chance that a stop or a limit fills away from the price you modeled. Size accordingly and stress-test fills.
Does the bot work on a prop firm account?
It depends on the prop firm's instrument list, which almost never includes tokenized equities today. Most funding partners restrict trading to listed futures, FX, and major equities. Verify the allowed-instrument schedule with your prop partner before assuming a token strategy is permitted.
How do I stop a 24/7 bot cleanly?
This is the disengagement problem with token markets. There is no closing bell to stop into, so a clean exit depends on the token book's depth at the moment you decide to stop. Build a manual kill switch and test it during a quiet session before you need it during a busy one.
Are tokenized stocks regulated the same way as the shares they track?
No. The WFE told regulators in August 2025 that tokenized stocks mimic equities without the same protections, and ESMA warned in September 2025 that they often lack shareholder rights. Holding an xStock gives no voting or other shareholder rights in the company.
Why does the stock lead price discovery if the token trades longer hours?
Because the stock is where the real order flow and disclosure live. The study even favored the token by stamping its price at midnight UTC, hours after the 4 p.m. New York close, and the stock still led.
What should I check before trusting a bot's backtest on tokenized equities?
Check the price series first. The WFE study uses one price a day and excludes weekends, so any backtest built on the same convention is missing the window where token platforms claim their advantage. Ask the provider for weekend and intraday data, or treat the results as incomplete.
Does Nasdaq's 23-hour trading plan change the trade?
Yes, structurally. Nasdaq is targeting a December 6 start for 23-hour weekday trading, and the paper says that narrows the hours gap between tokens and the stocks behind them. A narrower gap removes much of the reason a bot would prefer the token in the first place.
Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. Do your own research before making any investment decisions. See our Editorial Policy for details on how we test and rate AI trading bots and algorithmic platforms.
Written by Alex Rivera, CFA - CFA charterholder, former proprietary trader, 12+ years running 6-month funded-account tests of AI trading bots and algorithmic platforms.
Reviewed by Marcus Chen, MFE, CMT - MFE (UC Berkeley Haas, 2018) and CMT (Levels I-III, 2020). Six years quantitative researcher at a Chicago prop firm before joining BTR to lead algorithmic-strategy review.
Read our full Testing Methodology.
More in this category: Crypto Trading Bot Reviews.