$12.2B
thirteen-day volume — 3rd of nine networks measured
The tokenized-equity experiment
What 63 million trades reveal about Robinhood Chain
Thirteen days of Robinhood Chain's complete trade record — 63.5 million trades across 823,712 wallets — and what they show about how the chain is really used.
The findings in brief
Ten things thirteen days of complete trade data show.
- 93.5%
of all trading is memecoins. Tokenized equities are 6.5 percent. Trading one stock against another totalled $73,961 across thirteen days.
Section two → - 19.3%
of completed stock trades finish in profit, against 51.1 percent on memecoins. Every category of trader loses on equities — including high-frequency bots, at 33.3 percent.
Section three → - 361
separate contracts use the ticker GME. The 354 impostors out-traded the real GameStop token by two to one on $213.0 million of volume.
Section four → - −52.4%
is the median loss for holders of ticker-impersonating tokens — twice as bad as an ordinary memecoin. 64,394 wallets bought one.
Section four → - $59.17–$246.72
was the price range for the same tokenized GameStop share on the same day it woke up. It took 48 hours and eight more pools to settle near $21. Buyers at that day's median are down 86 percent.
The opening → - 84%
of the roughly 15,000 tokens launched every day never trade again after the first. Median lifetime: 13 trades, four wallets, under $1,200.
Section five → - 4.5%
of all trading gains went to the top 100 wallets. There is no cartel — profit is spread across 255,000 winners. The losses land on the holders: three in four open positions are underwater.
Section six → - $599K
is the total take of genuine arbitrage bots over thirteen days — one dollar in every twenty thousand traded. Whatever is costing traders money here, it is not front-running.
Section six → - 70.2%
of every trade on the chain is submitted by just five addresses; one alone submits 25.0 percent. For a venue listing tokenized securities, that is a chokepoint.
Section seven → - −27.7%
is the median loss on an ordinary memecoin position still open at the end of the window. Three quarters of all open positions are underwater, whatever they hold.
Section eight →
Robinhood Chain is real and large — $12.2 billion over thirteen days, third by volume of the nine networks measured, ahead of Ethereum. What it is used for is the story.
On 22 July, a tokenized share of GameStop sold for $246.72. That same day, on the same blockchain, the same token sold for $59.17.
Not a different GameStop. Not a knock-off. The identical contract, the identical asset, four times the price depending on which pool a buyer happened to land in.
Two days later it settled at $21.40 and has barely moved since, holding inside a three percent band for the rest of the window. GameStop's actual share price did nothing unusual during any of this. What changed was that eight more liquidity pools opened, and the token finally had a price instead of a rumour.
Anyone who bought at that day's median of $150.82 and held on is down 86 percent. Not because they were wrong about GameStop. Because for about 48 hours, the market they were trading in didn't work.
The week before, that token had barely traded at all — single figures a day. That pattern, of a token sitting dormant and then pricing itself in public, repeats across this chain.
Figure 1 — A price that didn't exist yet
Tokenized GameStop, 22–28 July, the week it woke up. The shaded band spans the 5th to 95th percentile of every trade that day — how far apart buyers and sellers were on what the thing was worth.
Median trade price5th–95th percentile of trades
Robinhood Chain is a real blockchain doing real volume. Over thirteen complete days in July it processed 63.5 million trades worth $12.2 billion, drawing 823,712 distinct wallets. That puts it third by volume among the nine networks in our dataset — ahead of Ethereum, ahead of Base, ahead of Arbitrum. By raw trade count it is second only to Solana.
It is also, by design, a place where tokenized US equities trade beside memecoins, around the clock, including on weekends when the New York Stock Exchange is shut. Apple, Nvidia, Tesla, GameStop and about ninety-five other tickers exist there as tradeable tokens.
We pulled the full trade record, scrubbed it, and went looking for what people actually do there, and how they fared doing it.
6.5%
of volume involves a stock. The other 93.5% is memecoins
19.3%
equity win rate, against 51.1% on memecoins
−52.4%
median fake-ticker holder loss, across 64,394 wallets
84%
of tokens die in a day; ~15,000 launched daily
Section oneThe chain is real. That part checks out.
Volume is easy to fake, so we didn't start there. We started with rhythm.
If a blockchain's activity is manufactured, it tends to hum along at a constant rate, because scripts don't sleep. Robinhood Chain does not hum. It breathes. Trading peaks between 13:00 and 18:00 UTC — which is to say, mid-morning to early afternoon in New York — and bottoms out at 4 a.m. UTC, when America is asleep. The peak hour carries 2.0 times the trades of the quietest one, and the shape repeats on all thirteen days.
That is the fingerprint of people. Specifically, American people, checking their phones during the workday.
Figure 2 — The chain keeps American hours
All 63.5 million trades, pooled by hour of the day. The shaded block is the New York Stock Exchange cash session.
Trades per hourUnique wallets per hourNYSE cash session, 13:30–20:00 UTC
Weekends slow but never stop. The quietest Saturday is $650 million against a weekday average of $1.04 billion — but weekend trade count falls by less than volume does, which means weekend trades are smaller, not merely fewer. And there is a genuine long tail of participants: 61 percent of wallets show up on exactly one day and never return, which is exactly what a consumer app looks like.
So the chain is real. The next part is where it gets slippery.
The 9,003 wallets active on all thirteen days — just 1.1 percent of the population — account for 37.3 percent of all volume. Read quickly, that is a small committed core carrying a transient crowd: healthy, normal, the shape of any consumer platform.
Then look at what the core is made of. Trading bots — wallets averaging more than a hundred trades a day, which no human places by hand — are 1.7 percent of the population and 51.3 percent of the volume. The committed core is mostly automated.
The crowd is human and shallow. The depth is automated.
Half the money on this chain moves through wallets that are 1.7 percent of the population and almost certainly not people.
Bot share of thirteen-day volume: 51.3%
Section twoWhat people actually trade
Start with what actually changes hands. Of $12.4 billion in trading, 93.5 percent is memecoins traded against crypto. Tokenized equities account for 6.5 percent.
One thing barely happens at all. Trading one stock against another — Apple against Microsoft, the bread and butter of any equity market — totalled $73,961 across thirteen days. That is not a market. That is a rounding error.
What people do instead has no precedent anywhere else in our data. On most chains a memecoin is priced against ether or a stablecoin — that is the other half of the pair, the thing you pay in. Here, $347 million of memecoin volume is priced against tokenized stocks: AAPL, NVDA, INTC and GME. The dog coin's price is quoted in Apple shares rather than dollars. Nearly two hundred thousand times a day, someone buys a joke token with a piece of a company.
Figure 3 — What actually trades
Every trade classified by what sat on both sides of it. Thirteen days, 16–28 July.
Weekday volumeWeekend volume
Section threeThe safest-looking trade is the worst one
Across every category of trader we measured — retail, whales, algorithms, high-frequency bots — every single one loses money trading tokenized stocks. Every single one makes money trading memecoins.
Just 19.3 percent of completed equity round-trips — a buy, then a matching sell — finish in profit. On memecoins, it is 51.1 percent.
Sophistication does not save you. High-frequency bots, the most technically capable participants on the chain, manage a 33.3 percent win rate on stocks. Whales manage 22.5 percent. These are catastrophic numbers for anyone who thinks of blue-chip equities as the conservative end of the risk spectrum.
The mechanism is unglamorous. A tokenized stock sits in a pool with very little money in it, and every round trip pays the spread twice — the spread being the gap between what you pay to buy and what you get back to sell, charged once going in and once coming out. A memecoin can move 40 percent in an afternoon and cover that cost many times over. Apple cannot. Apple moves half a percent and the spread eats you.
We tested this a second way, because it is the sort of result that ought to be checked. Sorting every position by whether its token rose, fell, or went nowhere produces the confirmation: tokens that went nowhere had the worst win rate of all, 37.6 percent — against 48.2 percent for tokens that fell by half and 44.8 percent for tokens that doubled. No movement to capture, only costs to pay.
Figure 4 — Everyone loses on stocks. Everyone wins on memecoins.
Win rate on completed round-trips, by trader class. 108,040 equity positions and 4,329,620 memecoin positions.
Tokenized equitiesMemecoins
Section fourThree hundred and ten tokens called GME
On Robinhood Chain, a ticker symbol is not an identifier. It is a claim, and usually a false one.
Search for GME and you will find 361 distinct contract addresses using that symbol. Exactly one is Robinhood's tokenized GameStop share. Of the remaining 360, some 354 do not even bother copying the issuer name — they are ordinary memecoins wearing a famous ticker like a costume. The half-dozen that do copy the name are dealt with below.
They are not a fringe phenomenon. The fakes did $213.0 million of volume across 1.34 million trades, out-trading the genuine token by two to one. Every major ticker carries the same infestation: 115 addresses claiming SPCX, 114 claiming COIN, 71 claiming TSLA, 60 claiming NVDA. The most impersonated ticker of all is ROBIN, with 416 addresses.
Volume, though, is not harm. So we measured harm.
Of the 64,394 wallets that bought a ticker-impersonating token, those who sold in time did well — $22.6 million in realized gains, a 52 percent win rate. Those who didn't are still holding, having paid $40.1 million for what they bought, and 79 percent of those positions are now worth less than that.
The median one has lost 52.4 percent. Twice as bad as an ordinary memecoin, and eleven times worse than a genuine tokenized equity.
Table 1 — What each kind of token did to the people holding it
Open positions are valued at each token's last observed price. Median outcome, not average — see the method note. Thirteen days, 16–28 July.
| What they bought | Wallets | Realized | Still open | Amount paid | Underwater | Median return |
|---|---|---|---|---|---|---|
| A genuine tokenized stock | 52,451 | −$1.76M | 16,536 | $10.2M | 78.2% | -4.7% |
| An ordinary memecoin | 810,801 | +$191.02M | 2,001,234 | $863.7M | 75.1% | -27.7% |
| Something wearing a stock's ticker | 64,394 | +$22.65M | 29,038 | $40.1M | 79.4% | -52.4% |
How to actually tell them apart
The obvious defence is to check the token's name rather than its symbol, since the genuine ones carry an issuer suffix. That defence fails too. Seven separate contracts carry the exact string GameStop - Robinhood Token. Intel, Microsoft and Nvidia each have two.
The name-copying fakes are small against $213.0 million for the symbol-copying ones — but they are enough to defeat name-based filtering entirely.
The only reliable check is the contract address. Verified GameStop is 0x1b0e319c6a659f002271b69db8a7df2f911c153e. Verified Nvidia is 0xd0601ce157db5bdc3162bbac2a2c8af5320d9eec. You can check any token's real address and its live trading history on DexRabbit's Robinhood token list before you touch it.
Section fiveA factory that makes 13,000 tokens a day
Robinhood Chain reports tens of thousands of tradeable tokens. It is worth understanding what that number counts.
Between 10,000 and 26,000 tokens appear for the first time every single day. About 84 percent of them never trade again after that first day. The median new token sees between 8 and 20 trades, attracts a handful of wallets, moves somewhere between $430 and $1,160, and goes permanently silent.
Outside the first cohort, only 4 to 13 percent are still trading 48 hours later.
We checked whether this was an artefact of ignoring tiny trades. It isn't — the death rate is stable in the low eighties across every cohort in the window. The token count is mostly a body count.
Figure 5 — Born Monday, gone Tuesday
New tokens by first day seen, with the share that never traded again filled in.
Dead after one dayStill trading next day
Section sixSo who is taking the money?
The obvious answer is insiders. It is also wrong, and worth saying plainly because it contradicts the standard story about chains like this.
Of $330.3 million in gross trading gains, the top ten wallets captured 0.9 percent. The top hundred captured 4.5 percent. Even the top thousand — well under one percent of all winners — took only 18.3 percent. Profit here is spread across roughly 255,000 winning wallets, most of them making a few hundred dollars.
There is no cartel. Which makes the question sharper, not softer: if the winners are ordinary and numerous, who lost?
The holders. More than two million positions were bought and never sold, and three quarters of them are underwater. The median one has lost 27 percent of its value. That is where the realized profit came from.
This is not extraction by a syndicate. It is a very large number of people selling to a slightly later very large number of people, with the losses landing on whoever stopped selling.
Nor, it turns out, is it the bots. We tested for sandwich attacks and found that 5 percent of trades looked like classic front-running — same wallet, same pool, same second, both directions. Almost all of it was an illusion created by routers splitting orders. Filter to events where the quantities actually match and genuine arbitrage amounts to 4,403 wallets taking $599,000 over thirteen days — about one dollar in every twenty thousand traded.
If you are losing money on Robinhood Chain, it is not because a bot got in front of you.
0.9%
of $330.3M in gains taken by the top 10 wallets
4.5%
taken by the top 100 — the winnings are spread wide, not captured
−27.7%
median loss on an open memecoin position; 75% underwater
$599K
total real arbitrage — $1 in every $20,000 traded
Section sevenFive addresses run the whole thing
One structural detail deserves more attention than it gets.
Robinhood Chain uses relayed transactions — users don't pay gas or submit their own trades; infrastructure does it for them. Only a few thousand addresses do that submitting. The distribution is not remotely even.
A single address submits 25.0 percent of every trade on the chain. The top five together submit 70.2 percent. The largest, 0xcaf681a6…5cb2, fronted 18.4 million trades on behalf of 469,815 different traders.
For an ordinary memecoin venue that is a footnote. For a venue hosting tokenized securities, it means five operators are a chokepoint: if they stop, throttle, or filter, most of the market stops with them. That is a governance question, and nobody appears to be asking it.
Section eightThe weekend problem
The most interesting thing tokenized stocks can do is trade when the stock market is closed. It is also the thing most often held up as their central risk. Thirteen days gives us two independent weekends to test that, and the test does not support it.
Across the two weekends in this window, the same tickers behave differently each time. AMD rose 2.3 percent over the first weekend and fell 5.8 percent over the second. Nvidia was flat, then down 5.1 percent. The megacaps — Apple, Microsoft, Google, Meta — moved less than two percent both times. There is no repeatable weekend effect here; the second Monday simply happened to be a day semiconductors fell, and the tokens tracked it, which is what they are supposed to do.
The genuine risk is liquidity, and it has nothing to do with the calendar. MicroStrategy's token is the clearest case: on Tuesday 21 July its median trade printed at $1,296, and the next day at $86. That is a fifteen-fold swing on a weekday, on a few dozen trades. The same token was untradeable for most of the preceding week — six trades on the Sunday, none at all on several days either side.
Which is why the headline cases cannot be tested twice at all. MicroStrategy and Reddit only became liquid on 22 and 24 July; before that there was nothing to measure. The one thin name with a record across both weekends is the oil fund USO, and it gives −0.3 percent over the first and −7.6 percent over the second. Inconsistent, on the only sample that permits the comparison.
The rule worth carrying away is not about weekends. It is that a tokenized equity with only a few thousand dollars a day of trading will print prices that mean nothing, on any day of the week, and those prints are what your position is then valued at.
Figure 6 — What Monday did
Friday close to Monday close, for every tokenized equity with a record across both weekends. Teal is 18–19 July, pink is 25–26 July.
FellRoseSemiconductors
Section nineHow it stacks up
Against the eight other networks we track over the same thirteen days, Robinhood Chain's scale is genuine. Its character is the outlier.
Its users trade far more often for far less money than anyone else's. Seventy-nine trades per wallet — second only to Solana, and nearly five times Ethereum's rate — but $15,044 of volume per wallet, about a third of Ethereum's. That is a retail app, not a settlement layer.
Table 2 — Nine networks, same thirteen days
Wallet counts use an identical definition and the same approximate estimator on every chain — consistent for comparison, and about half a percent above an exact count. Robinhood's exact figure is 823,712. Volume in USD, before the per-token price filter applied elsewhere in this article.
| Network | Trades | Wallets | Volume | Median trade | Vol / wallet | Trades / wallet |
|---|---|---|---|---|---|---|
| Solana | 300,682,046 | 3,040,033 | $99.48B | $96.65 | $32,724 | 98.9 |
| Binance Smart Chain | 60,220,271 | 1,601,382 | $15.55B | $53.47 | $9,710 | 37.6 |
| Robinhood | 64,605,402 | 823,043 | $12.38B | $48.42 | $15,044 | 78.5 |
| Ethereum | 3,538,905 | 211,985 | $8.75B | $98.78 | $41,264 | 16.7 |
| Base | 7,648,931 | 207,905 | $5.40B | $38.36 | $25,995 | 36.8 |
| Arbitrum | 2,426,652 | 57,356 | $1.44B | $129.98 | $25,064 | 42.3 |
| Polygon | 11,822,194 | 671,411 | $1.32B | $6.12 | $1,963 | 17.6 |
| Tron | 93,668 | 8,262 | $514M | $37.64 | $62,223 | 11.3 |
| Optimism | 441,036 | 6,026 | $52M | $24.73 | $8,547 | 73.2 |
Section tenIf you are going to trade there anyway
Six things the data says, in the order they will cost you money.
Check the contract address, not the ticker, and not the name.
Three hundred and sixty-one addresses claim GME, and ROBIN has 416. The median ticker-impostor holder is down 52.4 percent. Look the token up first.
Don't touch a tokenized stock in its first 48 hours.
GameStop traded between $59.17 and $246.72 on the day it activated, after a week of near-zero volume. Wait for the pool count to stabilise before believing the quote.
Don't round-trip blue chips here.
A 19.3 percent win rate is not variance, it is structure — you are paying the spread twice on an asset that moves less than the spread.
Thin equity tokens print meaningless prices on any day.
MicroStrategy printed $1,296 on a Tuesday and $86 the next day. Across two weekends no gap pattern repeats — the risk is how little the token trades, not the calendar.
Ignore the token count.
Around 84 percent of the 15,000 daily launches never trade a second day. Screens ranked by "new tokens" are ranked by noise.
Your losses are not the bots' fault.
Real arbitrage on this chain totals $599,000 over thirteen days. The money is lost to spread, launch chaos, and buying what other retail wallets are already selling.
Method
How we did this
Every figure comes from the complete Robinhood Chain trade record for 16–28 July 2026, held in Bitquery's trading_rt dataset. Robinhood indexing begins on 15 July and that day is partial, as is 29 July, so both are excluded.
Two volume figures appear in this article and they are not a contradiction. $12.2 billion is the total after every filter below, including a per-token price check; it is the number we use for headline claims. $12.4 billion is the total before that last filter, and it is what appears in the cross-network table, because the price check has to be computed per token per chain and we could only apply it consistently to Robinhood. Rankings are unaffected either way.
Raw data is not clean data. We started with 81,928,427 trades and finished with 63,456,466. We removed 71 records with impossible dollar values, and 8.4 million where the two sides of the trade disagreed by more than 5 percent about what it was worth. We quarantined 8.9 million trades under a dollar from the profit analysis, though we kept them in activity counts.
One correction mattered more than the rest. Some records carry wildly wrong prices on both sides, so cross-checking the two doesn't catch them: Nvidia's average traded price computes to $590,048 against a median of $207.70, because a handful of records price it in the billions. We added a rule discarding any trade priced outside a band of one-fifth to five times its token's daily median, which removed a further 1,148,936 records.
| Stage | Rule | Removed | Remaining |
|---|---|---|---|
| Raw | Network and date filter | — | 81,928,427 |
| Rule 1 | USD value above $10M (corrupt) | 71 | 81,928,356 |
| Rule 2 | The two USD sides disagree by over 5% | 8,380,088 | 73,548,268 |
| Rule 3 | Zero price or zero amount | 2,985 | 73,545,283 |
| Quarantine | Trades under $1 (kept in activity counts) | 8,939,881 | 64,605,402 |
| Rule 6 | Price outside 0.2×–5× the token's daily median | 1,148,936 | 63,456,466 |
We ran the same adversarial checks against this rebuild, and two changed it. The first is that aggregate holder returns are not a usable statistic: barely one position in a thousand hits the cap we place on price moves, yet those alone swing the total from minus 18 percent to plus 74 percent. Every holder figure in this article is therefore a median, not an average — which is also a correction to an earlier six-day version of this analysis, where the averages were reported.
The second is the weekend claim in section eight, which an earlier version stated as a rule about trading-volume thresholds. With a second weekend available it does not reproduce, and the section now says so.
The checks that held: the profit-and-loss direction is not inverted — positions in tokens that went nowhere show the worst win rate of all, which is only possible if traders pay the spread twice. Token mortality is not an artefact of ignoring sub-dollar trades. And the ordering of holder outcomes — impostor worse than memecoin, memecoin worse than genuine equity — is stable however the tails are treated.
Three limits are worth stating. Wallets are addresses, not people, and one person can run hundreds. We have no off-chain equity price feed, so every price here is the on-chain one; we make no claim about deviation from the New York close. And the source tables carry a thirty-day retention window, so this window stays queryable only until roughly the middle of August, after which the records behind this article are deleted.
Resources
Explore the data yourself
Every token, pool and trade discussed here is publicly inspectable.
DexRabbit — Robinhood token list
Live rankings for every token on the chain, with volume, traders and price history. The fastest way to check whether the ticker you are looking at belongs to the contract you think it does.
dexrabbit.bitquery.io/robinhood/tokens
Bitquery Explorer — Robinhood Chain
Block-level exploration of transactions, addresses, contracts and transfers, including the relayer addresses and router contracts referenced above.
explorer.bitquery.io/robinhood
Query this data yourself
Every figure above came from Bitquery's Robinhood Chain trade data. Point an AI agent straight at it through our MCP server, or start from the Robinhood trades API docs.