REX Shares and Tuttle Capital listed a fund on Cboe this week called the T-REX 2X Long ASST Daily Target ETF, ticker ASSX. The press release is dated 18 September 2026 and Cointelegraph reported it trading on Friday the 19th. It seeks “200% of ASST’s daily performance, before fees and expenses”. ASST is Strive, the bitcoin treasury company that holds 25,000 BTC, the fifth largest stack on any listed company’s balance sheet, most recently topped up by 469 coins funded through the sale of a perpetual preferred stock. Strive closed Friday at $30.09, up 6.4%.
So it is leverage on a stock that is itself leverage on bitcoin, and the obvious post writes itself. We wrote it in April, when GraniteShares listed a 3x XRP product and the argument was that Wall Street had built a casino chip with a prospectus. That post objected to what you cannot see: the swap counterparties, the hedging, the rebalancing. This one is the opposite. The most dangerous thing about ASSX is printed in the disclosure, in plain English, and it is entirely true.
TL;DR
- REX and Tuttle listed ASSX on Cboe, a 2x daily fund on Strive (ASST), which holds 25,000 BTC. The objective is 200% of one day’s move, not of anything longer.
- Run a stock up 10% and back down 9.09% and it is exactly where it started. The 2x fund is down 1.82%. Repeat that for 40 trading sessions and the fund is down 30.7% while the stock has not moved at all.
- The gap is not a fee, a tracking error or a failure. For a 2x fund the drag is exactly the variance of the underlying, so the better the fund does its stated daily job, the more of your money the path eats.
- Every “provably fair” claim in crypto gaming has the same shape: a guarantee whose scope is exactly one event, read by everybody as a guarantee about a sequence of them.
- A daily-reset leveraged fund and a gambler betting a fixed fraction of a bankroll are the same equation. Satoshie’s raffles are single-shot with no reset, but our proof is still one draw wide and we are not going to pretend otherwise.
The promise is kept every day and broken over a month
Take a stock at 100. On Monday it rises 10%, to 110. On Tuesday it falls 9.0909%, back to exactly 100. Two days, no movement, nothing to report.
The 2x fund did its job perfectly on both days. Monday it returned 20%, going from 100 to 120. Tuesday it returned twice minus 9.0909%, which is minus 18.18%, taking 120 to 98.18. The fund is down 1.82% and the thing it tracks is flat. Nobody made a mistake. Both daily results were precisely 200% of the underlying, which is the only thing the fund ever promised.
Now repeat the two-day cycle for two months, 40 trading sessions. The stock is still at 100, unchanged, twenty round trips later. The fund is at 69.28. It has lost 30.7% of its value tracking an asset that did not move.
Make it messier and it gets worse. Alternate +10% and -10% for the same 40 sessions. The stock ends down 18.2%, because up ten and down ten is not a round trip either. Twice that loss would be 36.4%, and most people holding a 2x product would accept 36.4% as the honest cost of being wrong. The fund is actually down 55.8%. The extra 19.4 percentage points were not charged by anybody. They are the path.
The drag is variance, not direction
This is not a quirk of the numbers I picked. For a fund resetting daily at k times exposure, the shortfall against k times the underlying’s compounded return works out to k(k-1) times half the variance, per unit of time. At k equals 2 that lands on something almost rude in its simplicity: the drag is the variance.
Put a number on it. A bitcoin treasury equity, geared by preferred issuance, running a 5% daily standard deviation is not an aggressive assumption. Five per cent daily squared is 0.25% a day, gone, regardless of direction. Annualised, that is roughly 79% volatility and about 63% of your compounded log return over a year, which means the fund finishes the year worth around 53p in the pound against twice what the stock actually compounded at. That is an idealised figure with zero fees, and fees are real and additional.
Read that sentence again with the product description next to it. The fund’s job is to deliver twice the daily move. It performs that job flawlessly. Performing it flawlessly is the mechanism by which the holder is separated from half of the return they thought they had bought. There is no villain in this story, no undisclosed parameter, no hidden clause. The loss is produced by the guarantee being kept.
When Solana shortened its slot time on Friday, a unit changed and everything measured in it was quietly repriced. Here nothing changes at all. The guarantee is stable, correct and continuously honoured, and it still does not mean what the holder believes it means, because it was never a claim about the quantity they care about.
The scope is in the disclosure and the disclosure moves the job to you
REX is not hiding any of this. The filing says the fund’s “performance for periods greater than a trading day will be the result of each day’s returns compounded over the period, which is very likely to differ from 200% of ASST’s performance”. That is the whole of the above, stated accurately, by the issuer, before anyone bought a share.
Then comes the line that should be taught in schools: “The Fund is not intended to be used by, and is not appropriate for, investors who do not intend to actively monitor and manage their portfolios.” Scott Acheychek, COO of REX Shares, called the launch “the tool that was missing” and noted traders were already active in the name. Every word of that is defensible. Traders, plural, actively monitoring.
But notice what the disclosure does. It states the scope of the guarantee, one day, and then hands the entire remaining problem, every day after the first, to the holder as a homework assignment. The fund guarantees the step. You are on the hook for the walk. That transfer is legal, disclosed, and invisible to anyone clicking buy in a brokerage app that lists ASSX beside every other ticker with no hint that this one has a shelf life measured in sessions.
What this has to do with a coinflip
Every provably fair page in crypto gaming grades the same thing: whether the operator could have changed the outcome. Server seeds, client seeds, nonces, hash commitments, VRF proofs and verifier links are all integrity machinery and all of it answers one question about one event.
A Chainlink VRF proof is exactly one draw wide. It says this randomness was produced by a key committed in advance and could not have been chosen after the bet. It is arithmetic rather than inference, which is why we have argued that one proof beats a thousand results and that you cannot audit a casino by playing it. That still holds. This post is the other edge of the same fact. A guarantee narrow enough to be exact is also narrow enough to be silent about everything outside it, and what sits outside it is the only thing a player actually experiences: a sequence.
Here is the equivalence that ought to bother the industry more than it does. A daily-reset leveraged fund and a gambler staking a fixed fraction of their bankroll are the same machine. Both resize exposure to current equity at the start of every period. Both therefore compound geometrically rather than adding up. Both are hit by the same variance term, and in both the average outcome is pulled upward by rare large results while the median participant does considerably worse than the average implies. The 4pm rebalance and the autoplay button are the same button.
So an operator can hand you a valid proof for every spin you have ever made, hold nothing back, cheat at nothing, and the shape of your session remains unproved. Not badly proved. Unproved, because the proof was never about that. Meanwhile the quantities that decide how the session ends, the house edge on each bet, the gas on each entry, the platform fee, are individually trivial, fully disclosed, and applied multiplicatively to a shrinking balance a few hundred times an evening. They are the 0.25% a day. Nobody calls them out because per event, they are honest, and per event is the only tense in which the industry has learned to speak.
Where Satoshie actually stands
The narrow claim, and it is deliberately narrow. A Satoshie raffle entry is single-shot. Your stake is fixed when you enter and escrowed by the contract. The VRF coordinator is named in deployed code you can read before you commit anything. The returned randomness carries a proof the contract verifies on-chain before the payout callback is permitted to act. Resolution and payout happen in one transaction. Your odds are arithmetic over tickets minted, readable before purchase, not a number we are free to revise.
There is no leverage, so there is no maintenance margin and no liquidation engine. There is no overnight position, so there is nothing to reset at 4pm. There is no bonus balance carrying a wagering requirement that obliges you to keep betting in order to leave with your own money. Nothing inside the product resizes itself to your balance, which means the product itself does not compound against you.
And that is where the claim stops. Our guarantee is one draw wide, exactly like everybody else’s, and a sequence of individually verified draws is still a sequence. We have said before that provable fairness protects you from the house and not from yourself, and the arithmetic above is the sharpest statement of why. If we ever start implying that a verifiable outcome is a safe activity, hold this post up to us.
Three honest limits
First, the criticism lands on us. Gas and platform fees are per-event costs paid on every entry and they compound in precisely the shape described above. Being on-chain does not exempt us from the maths, it just makes the costs enumerable by a stranger with a block explorer.
Second, a VRF proof settles randomness and nothing else. It is silent on whether a contract can be drained, whether a prize is funded, and whether our front end, an ordinary web app and the least trustworthy thing we ship, is serving you the interface we think it is.
Third, this is not an argument against ASSX, against leveraged ETFs, or against REX. A disclosed daily objective with the compounding consequence spelled out in the filing is better documentation than most of crypto gaming has ever produced. The objection is to the reading, not to the product, and the reading is the industry’s fault rather than the issuer’s.
Three questions worth asking anywhere
Is the guarantee on this page about one event, or about a sequence of them? If it is about one event, who told you which one you were buying?
What is charged per event here, and what does that come to over the number of events I actually intend to play, rather than over one?
If I verified every result I have ever had on this platform, and all of them checked out, what exactly would I have proved about how the evening ends?
Look at a mountain switchback from above. Every hairpin is correctly surveyed, correctly banked, correctly signposted. Drive it and you will cover eight kilometres to advance two. Nobody lied to you about a single corner. The corners were never the thing you were trying to measure.
📷 Photo by Jerry Kavan on Unsplash
