Who is pricing tokenized stocks? The answer is not Kraken.
Written by: stein
Translated by: Chopper, Foresight News
On Wednesday, October 28, buyers on Kraken were a bit anxious: all day long, they kept chasing the tokenized Nvidia stock NVIDIA higher, and by the close, aggressive buying exceeded aggressive selling by $224,000. As a result, the closing price that day fell 1.85%. (Note: all times mentioned in this article are U.S. local time)
So the question arises: why would a group of chase buyers lose money?
Stock prices are determined where real shares are traded, such as exchanges like Nasdaq and the New York Stock Exchange. They provide a price reference for on-chain stock tokens. But once the stock market closes, these tokens must keep trading without a reference, and prices can then deviate wildly.
With that question in mind, I went through 60 trading days of trades in 6 tokenized stocks on Kraken to see when buying pressure really moved prices and when it was just noise.
What I looked at
I examined every trade in 6 tokenized stocks on Kraken between September 14 and November 13, 2026: 159,002 trades totaling $102 million.
Each trade has a label showing who was in a hurry to trade. “Buy” means the buyer couldn’t wait and immediately took the seller’s ask; “Sell” means the seller couldn’t wait and accepted the buyer’s bid. This kind of “urgency” is what traders call an aggressive trade.
First, a few definitions:
Net pressure: the dollar amount of aggressive buying minus the dollar amount of aggressive selling.
CVD (cumulative volume delta): Think of it this way. There are buyers and sellers in the market. Every time someone clicks “Buy now,” buyers get 1 point; every time someone clicks “Sell now,” sellers get 1 point. CVD is the difference between the two sides’ clicks. If buyers click 10 times and sellers click 4 times, CVD is +6.
Price response: how much the price moved, measured in basis points. 1 basis point = 0.01%; 100 basis points = 1%.
Time of day: U.S. trading hours (9:30–16:00 New York time), premarket and after-hours, overnight, and weekends.
Also, activity levels vary across tokens, so $3,000 of buying means different things for each one. To make a fair comparison, I set a separate “normal level” for each token. Specifically, I divided the 60 days into one-minute intervals (10:00–10:01, 10:01–10:02, and so on), measured the amount of “urgent buying minus urgent selling” in each interval, and then found the typical value for each token. For AAPLx, it was about $190; for METAx, about $360. In other words, $1,900 of net pressure is 10 times the normal level for AAPLx, but only about 5 times the normal level for METAx. So when I say “10 times the normal level” below, I mean relative to what’s typical for that token in a one-minute window.
One surprising finding was that in every trade, the buyer and seller agree on the same price; what matters is who initiated the trade. An impatient buyer takes the seller’s higher ask, while an impatient seller accepts the buyer’s lower bid. So on most exchanges, a buyer-initiated trade is usually slightly higher than the immediately preceding seller-initiated trade. But on Kraken, it’s often the other way around, because the price moves on its own between trades. Kraken’s price keeps drifting even when no one is pushing it. This is the first clue that something outside Kraken—such as Nasdaq—is pushing prices.
The scorecard
Here’s a comparison of October 28 with another Wednesday, October 14. NVDAx fell about 1.8% on both days. On October 14, the scoreboard matched the price: sellers’ aggressive trades exceeded buyers’ by $183,000, and the price fell. On October 28, buyers’ aggressive trades were higher, and the price still fell.
October 28 wasn’t an isolated case. I checked all 45 U.S. trading days for 6 tokens (265 token-days in total): on 128 of those trading days, or about 48%, CVD and the closing price moved in opposite directions.
But weekends are different. When Nasdaq is closed, CVD and prices move in the same direction more often. Measured as a correlation from 0 to 1, it was 0.24 on weekends and 0.04 during U.S. trading hours.
Echo or amplifier?
This is the core test. In every minute when buyers outweighed sellers (or vice versa), did the price move in that direction?
During U.S. trading hours, with moderate pressure (1–3 times the normal level), the price moved in the direction of the pressure only 51.2% of the time—about the same as a coin flip. Overnight, with stronger pressure, the price moved in the direction of the pressure 76% of the time.
Buying pressure is only a signal within the minute it occurs
Buying pressure is like footprints in the sand: it clearly tells you someone just walked here, but not where they’re going next.
The strongest pressure (more than 10 times the normal level) moved the price 5 basis points in its direction during the minute it occurred. After that, the buying pressure disappeared. Over the next 1, 5, and 15 minutes, the price moved only 0.6 to 1.3 basis points.
During U.S. trading hours, looking at longer time periods doesn’t help. Over 1 minute, the price still follows buying pressure a little; over 1 hour, that connection nearly disappears.
If you pick out the most extreme minutes of one-sided buying or selling (the top 2%), then track the price for the next 15 minutes:
U.S. trading hours: the 1-minute move jumps 7.8 basis points, then continues in the same direction to 14.1 basis points. That’s because Kraken’s flow is following a real move happening on Nasdaq.
Nighttime / weekends: when the 1-minute move jumps 6.8 basis points, it later falls back to 2.4 basis points. With Nasdaq unavailable to follow, a single large order can move a thin market; afterward, prices gradually retreat.
This shows that large spikes during the day are a signal worth watching, while large spikes at night mostly fade over time.
The same buying pressure—why does the price sometimes rise and sometimes fall?
Two traders apply the same buying and selling pressure, but get completely different results. It’s like pushing two doors with the same force: one opens, while the other is locked, and you can’t tell from the outside.
I filtered for all the minutes when one side’s pressure was roughly the same—about $1,720 more in buying. In those 100 minutes, the price moved in the direction of the pressure about 64 times, and against it about 36 times. And in a typical 1-minute window, the price moved very little—about 0.01%. So the size of the pressure alone tells you very little.
Three things can tell you which door you’re pushing:
Is Nasdaq open? During U.S. market hours, 41% of buying pressure was against the price direction; overnight, only 27% was.
How busy is this 1-minute window? After a quiet hour, 31% of buying pressure was against the price direction; after a busy hour, it was 38%, because price levels change most often when things are busy.
One trade, or several? A single trade moved the median price by 0.5 basis points; a sequence of 4 to 10 trades sweeping across multiple price levels moved it by 2.1 basis points.
These are clues about what’s behind the door, not a way to see through it. To see directly what’s behind the door, you need the order book—and Kraken doesn’t keep order book history.
Buying pressure is worth treating as a signal in these situations:
When the U.S. market is closed, especially overnight and on weekends, Kraken’s own traders have a greater influence on pricing.
One-sided pressure more than 10 times the normal level moved the price in its direction 66% of the time; moderate pressure did so only 55% of the time.
Several trades in a row, sweeping through multiple resting order prices
A sudden large order
The takeaway: during U.S. trading hours, Kraken order flow is an echo of a move happening on Nasdaq; overnight and on weekends, it’s an amplifier—but some of the sound it produces fades away.
