Ever

Unlocking the value of the digital asset economy. Instant, asset-backed credit for the modern portfolio. Don't sell. Ever.

[Ever @ever_card]

time to talk about what we've actually been building.

Ever is a credit card backed by your crypto.

not a card that sells your coins every time you swipe. a real revolving line of credit, from a regulated lender, with your crypto pledged as collateral.

the idea is simple:

borrow against your crypto. don't sell it.

your stack stays intact. your conviction stays intact. and you still get to live your life in dollars.

if you've been following this account, you know the three questions that expose most of this category:

> am i borrowing, or am i selling?
> which bank said yes to this?
> if this company went bankrupt tomorrow, whose coins are they?

we started Ever because we couldn't find a card with the right answer to all three.

so we're building one. slowly, deliberately, through the full bank diligence gauntlet i wrote about. that's the only way a real one gets built.

the card ships when the machine underneath it is ready.

if you want to be early:

https://t.co/uJc7WYpYPr

the first names on the list get first access to our U.S. launch.

[Ever @ever_card]

bitcoin never closes.

liquidation engines never close.

your bank closes Friday at 5.

that gap is where careful borrowers get liquidated.

here's the thing nobody tells you about margin calls: they cluster at the worst possible hour.

> crypto is the only major market open on weekends, so it absorbs every shock alone
> weekend order books are thin, so the same selling moves price further
> Asia and Europe are awake while America sleeps

the pattern is old enough to have nicknames.

> December 2021, "Red Saturday": bitcoin fell more than 20% in about an hour on a Saturday morning. roughly $2.5 billion liquidated.
> August 2024: the crash landed Sunday night into Monday morning, while US banks were still dark.
> October 2025: a tariff headline hit after the Friday closing bell. stocks couldn't react. crypto could. $19 billion liquidated in 24 hours, the largest single day on record.

now put yourself in the loan. it's 2am Saturday, your LTV just crossed the line, and the lender gave you a cure window.

run every option through that hour:

> wire dollars: for most banks, the wire desk is closed until Monday
> ACH: no settlement on weekends. same-day ACH means same business day.
> sell stocks to raise cash: market's closed
> instant payments (RTP, FedNow): live 24/7 on paper. ask whether your lender accepts them for a cure at 2am.
> stablecoins: 24/7, if the lender takes them
> add bitcoin you already hold outside the loan: on-chain, any hour, usually credited within an hour or two. check.

notice which options survived. the ones that live on rails that never close.

which means every cure window has a hidden second number. "24 hours" that start Friday at 8pm contain zero business hours. the length of the window matters less than how many of those hours you can actually use.

so the borrower playbook writes itself:

> keep unpledged bitcoin ready to move, in a wallet you control, before you ever need it
> know your lender's crediting time for added collateral, in minutes, not business days
> have one 24/7 dollar path tested in advance
> decide your cure order now, in writing. at 2am you will not be thinking clearly.

and before you borrow at all, run the Saturday test:

if the call comes at 2am on a Saturday, which of my cure options work before Monday?

whatever passes is your real safety margin.

everything else is a weekday.

[Ever @ever_card]

in April 2024 a fintech lost track of $85 million of its customers' money.

nobody hacked it. and the trustee's reports never established where it went.

it just... couldn't say whose money was whose.

this is the Synapse story, and if you've ever seen "FDIC insured" on a fintech app, you need it.

first, how your dollars actually sit at most fintechs:

> the app doesn't hold your money. a partner bank does.
> the bank holds one big account "for benefit of" every user of the app. an FBO account.
> the app, or a middleware vendor behind it, keeps the ledger of who owns what inside that one account.
> your balance screen is that ledger. not the bank's books.

sound familiar? it's the pot model from my custody post, running on dollars instead of coins.

Synapse was the middleware in the middle. it ran the ledger connecting dozens of consumer apps to a handful of banks.

then it went bankrupt.

> the banks could see one big number
> the apps could see their users' balances
> the ledger that mapped one to the other lived inside the bankrupt company

more than 100,000 people were locked out of their savings. for months.

a former FDIC chair was appointed trustee and started reconciling. the ledger said customers were owed $265 million. the banks held $180 million.

the gap: somewhere between $65 and $96 million, depending on the report. and the source of it was never pinned down.

here's the part that surprised everyone: FDIC insurance did nothing.

because FDIC insurance covers a bank failing. the banks didn't fail. the ledger did.

"FDIC insured" on an app is a promise about the bank's survival, not the app's. Voyager taught crypto that lesson in 2022. Synapse taught fintech the same one in 2024.

(this is also the real answer to "why does a card take a year to launch." after Synapse, banks stopped taking a fintech's ledger on faith.)

so the test from my custody post has a twin.

for coins: point to my coins.

for dollars: point to my dollars.

any company touching your money should answer three things:

> where does the money physically sit?
> who keeps the ledger?
> who reconciles that ledger against the bank, and how often?

if the answer to the third is "the same company that keeps the ledger," you've found the gap before it finds you.

[Vraj Patel @vrajhq]

@econoar The fact that we’re still holding steady is crazy

[Ever @ever_card]

in 2021 you could deposit bitcoin on a platform and earn 6%.

bitcoin doesn't pay interest. no coupon, no dividend, no rent.

so who was paying, and why?

that question was the entire due diligence. almost nobody asked it.

start from first principles: yield on an asset that produces nothing has to be manufactured somewhere else. there are only a few factories:

> lend it to a trader who pays to borrow it (you now hold their credit risk)
> the basis trade: long spot, short futures, pocket the gap (real, but only in bull markets)
> the GBTC premium: buy bitcoin, wrap it, sell the wrapper at a markup (real, until it wasn't)
> DeFi farming: park it in code and collect tokens (code risk plus token risk)
> new deposits pay old depositors (this one has a name)

every one of those is a trade. every trade has a season.

here's what happened to the season.

in the bull market, traders paid up to borrow bitcoin. annualized funding on futures ran into the double digits. the GBTC wrapper sold at a premium that sometimes topped 30%. lending bitcoin genuinely earned real money.

then in February 2021 the GBTC premium flipped to a discount. the basis trade compressed. the honest yield on bitcoin drifted toward zero.

the promised yield didn't.

that's the whole autopsy in one sentence: the trades ended, the promises didn't, and the gap had to be filled with something.

the something was risk. specifically, unsecured loans to the same handful of hedge funds still willing to pay up, because they were the ones in trouble.

receipts:

> Voyager lent more than $650 million to a single hedge fund, Three Arrows Capital, unsecured. more than half its entire loan book, one borrower. the fund defaulted in June 2022. Voyager was bankrupt within two weeks.

> BlockFi lent the same fund about a billion. in February 2022 the SEC had already fined BlockFi $100 million for selling its yield product as an unregistered security. after the fund blew up, BlockFi took a $400 million lifeline from FTX. it went down with FTX in November.

> Celsius paid up to 17% APY. the court-appointed examiner later found that new customer deposits were, in part, funding the rewards and propping up the company's own token. the CEO pleaded guilty to fraud and was sentenced to 12 years.

and the tell that should have ended every one of those conversations early: Voyager told customers their deposits were FDIC insured. the FDIC and the Fed publicly ordered them to stop. the insurance covered the bank failing. not Voyager failing.

the physics underneath all of it:

yield above the riskless rate is a risk premium. it exists because someone is holding a risk. if you can't say who, it's you.

so before you accept yield on your bitcoin, ask the three-word question:

name the trade.

then ask what happens to your coins when that trade stops working. because every trade stops working.

the honest structure is boring. your bitcoin sits, earns nothing, and stays yours.

the yield you didn't earn is the risk you didn't take.

[Ever @ever_card]

@coinbase Here we go

[Ever @ever_card]

the largest collateralized lending market in America isn't on Wall Street.

it's the house next door.

> American homeowners hold roughly $35 trillion in home equity
> they've borrowed nearly $13 trillion against their homes
> HELOC balances alone have climbed past $380 billion

and nobody finds any of this strange.

your neighbor opens a credit line against the house to redo a kitchen, cover tuition, float the business through a slow quarter. at dinner they don't say "i levered my largest position." they say "we tapped some equity."

tapped. the most domesticated word in finance.

now run the mirror.

the same neighbor holds bitcoin that's up 5x. they need cash. what does everyone tell them?

"just sell some."

and if they mention borrowing against it instead: "too risky."

read that back. borrowing against the illiquid asset is prudence. borrowing against the liquid one is gambling.

hold that thought and grade each as collateral:

a house:
> one buyer at a time, months to close
> priced by an appraiser's opinion, and only when someone asks
> you cannot sell 3% of it

bitcoin:
> divisible to eight decimal places
> a live global price every second of every day
> settles in minutes, around the clock

as pure collateral, bitcoin is arguably the cleaner asset. what it costs you is volatility. bitcoin moves in a week what housing moves in a bad year.

but volatility is a parameter, not a principle. it's why a house can carry an 80% loan and bitcoin belongs far lower, behind a wide buffer. the number changes. the logic doesn't:

you don't sell a long-term asset to solve a short-term problem. you borrow against it, with room to survive.

"but 2008." right. home lending did blow up once. look at how:

> combined LTVs at 100% and beyond
> stated income, nobody checking
> underwriting that assumed prices only rise

2008 wasn't proof that borrowing against homes is wrong. it was proof of what happens when the buffer goes to zero. kill the buffer and any collateral kills the system.

one last thing about why the double standard exists at all.

you can't sell your house six dollars at a time. the friction forces you to think in decades.

bitcoin removed the friction. and everyone quietly mistook "easy to sell" for "right to sell."

so here's the test. next time selling feels like the responsible move, swap the asset and say it out loud:

"i sold a piece of my house to pay for the trip."

if it sounds absurd with a house, ask why it sounds prudent with bitcoin.

same logic. same wealth. the only difference is which asset your bank already understands.

[Vraj Patel @vrajhq]

we’re running a closed pilot for Ever's crypto-backed loan.

small group, real feedback, direct line to me.

this is the loan product, not the card.
if you've had a moment where you needed liquidity and sold anyway, you're exactly who i want to talk to. DM me.

U.S. only

[Ever @ever_card]

the most dangerous number in crypto lending is the one that sounds most generous.

"borrow up to 90% of your bitcoin's value."

that 90% isn't a gift. it's the distance to your own liquidation. and it's shorter than you think.

every loan against a volatile asset lives between two numbers:

> the LTV you borrow at
> the LTV where liquidation triggers

everyone reads the first number. almost nobody reads the second. the gap between them is the actual product.

here's the math. the price drop that liquidates you:

1 − (borrow LTV ÷ liquidation LTV)

run it on three loans:

> borrow at 90%, liquidate at 95%: a 5% dip ends you
> borrow at 75%, liquidate at 90%: a 17% dip ends you
> borrow at 50%, liquidate at 85%: you survive a 41% crash

now put those buffers next to bitcoin's actual behavior:

> 5% days: routine. dozens per year.
> 20% inside a single week: dozens of times on record
> 30%+ drawdowns: most years of bitcoin's existence
> nearly 40% in one day: March 2020
> roughly half in weeks: May 2021
> about 75% peak to trough: the 2022 bear

so the 90% loan isn't really a loan. it's a countdown. the market doesn't have to do anything unusual to liquidate you.

it just has to be bitcoin.

why do lenders lead with the big number anyway?

because "borrow more" converts better than "survive longer." max LTV is marketing. the buffer is the product.

and here's the tell worth checking: some platforms charge a fee when they liquidate you. read that twice. if a lender earns money on your liquidation, your liquidation is not their failure mode. it's a revenue line.

the reframe:

the lender offering you less against the same coins isn't being stingy. they're pricing bitcoin's actual history. they're planning for you to still be a customer after the next crash, because they expect a next crash.

a loan against bitcoin should be built to survive bitcoin.

so before you pledge a single sat, run the only calculation that matters:

1 − (your LTV ÷ their liquidation trigger) = the crash you can survive

if that number is smaller than a normal bitcoin week, you're not borrowing.

you're waiting.

[Ever @ever_card]

FICO doesn't predict whether you can repay a loan.

it was never designed to.

look at what the score is actually made of:

> 35% payment history: did you obey past credit agreements?
> 30% utilization: how much of your limits are you using?
> 15% length of history: how long have you been in the system?
> 10% new credit: are you asking for more right now?
> 10% mix: how many kinds of credit have you handled?

now look at what's missing:

your income. your savings. your assets. your net worth.

FICO has no idea what you own or what you earn. by design.

so the score doesn't answer "can this person repay?"

it answers "does this person comply with credit agreements?"

and that's not a flaw. it's the entire genius of the system.

credit used to be a judgment call. a loan officer sized you up, and "character" meant whatever he wanted it to mean.

then in 1989 Fair Isaac shipped one general-purpose number. in 1995 Fannie Mae and Freddie Mac made it the standard for American mortgages. same math for everyone, no opinion about your zip code or your last name. credit access exploded. today FICO's own pitch is that 90% of top US lenders run on it.

and the number works because it's an incentive machine:

your score is collateral you've posted with every lender at once. default on one, lose access to all. that standing threat is what makes unsecured lending possible at scale.

but a system that measures obedience has a blind spot the size of a balance sheet:

> roughly 45 million American adults are credit invisible or unscorable
> a retiree with a paid-off house and seven figures saved: thin file
> an immigrant with a decade of flawless payments abroad: no file
> someone holding a fortune in BTC and no card history: invisible

none of them lack the ability to repay. they lack a record of obedience.

and obedience recorded in calm times is exactly the data that goes quiet in a storm. every lender eventually learns that the same expensive way.

here's the frame almost nobody uses:

lending has always had two ways to answer "will i get my money back?"

> unsecured lending bets on your record. FICO is how that bet got scalable.
> secured lending holds an asset. the pawnshop, the margin loan, the mortgage. protection doesn't require prediction.

two different questions. "will they obey?" versus "am i covered if they don't?"

the best-designed credit products use both: the asset answers this lender's question, and the on-time record you build answers the next lender's.

that's what a secured credit card has quietly done for decades. collateral opens the door. reporting builds the file.

so when you size up any credit product, ask what's actually being underwritten:

my history, or my assets?

if the lender can't answer that in one sentence, they're underwriting hope.

[Ever @ever_card]

your bitcoin on a platform lives in one of two places:

a vault with your name on it, or a row in someone's spreadsheet.

most people never ask which. here's how to tell them apart.

model 1: omnibus.

> every customer's coins go into one big pool of wallets
> the company's internal database tracks who owns what
> your balance is an entry in that database
> on-chain, "your" coins don't exist. only the pot does.

to be fair, there are real reasons exchanges run this way. pooled wallets make trading instant and withdrawals cheap. an internal transfer is just a ledger edit.

but be clear about what you're trusting: not the blockchain. the spreadsheet.

model 2: segregated.

> your coins sit in an account established for you specifically
> identifiable. auditable. not a claim on a pool.
> if the company's database burned down tomorrow, your coins would still be findable

the uncomfortable part: from inside the app, both models look identical. same balance screen. same green number.

the difference only shows up in three moments: an audit, a bank run, a bankruptcy.

receipts:

> Mt. Gox ran the pot model. coins leaked for years while the database kept showing full balances. final damage: 850,000 BTC. about 7% of all bitcoin that existed at the time.

> QuadrigaCX, once Canada's biggest exchange. the founder dies suddenly. court-appointed auditors finally open the cold wallets: empty. and not freshly emptied. drained eight months before he died. the ledger said ~$150 million in crypto. the wallets said zero.

> FTX. one pot, one very wrong database, an $8 billion hole. the restructuring chief called it "a complete failure of corporate controls."

three companies. three eras. one design flaw: the ledger and the coins were allowed to drift apart, and only the company could see the gap.

yes, holding your own keys sidesteps all of this. but the moment you use any platform, an exchange, a lender, anything, a custodian enters the picture. so learn to grade one.

the test is four words: point to my coins.

> segregated custody answers with an account. yours.
> omnibus answers with a pot and a promise that the math works out.

and if your coins are collateral for a loan, the question gets sharper. collateral that can be pointed to is hard to quietly reuse. collateral in a pot is one ledger edit away from being someone else's trade.

so before you deposit anywhere, ask it:

can you point to my coins?

if the answer is a pause, that's the answer.

[Ever @ever_card]

credit cards have one genuinely great feature, and most people who use one every day can't explain it.

the grace period.

here's the clock, start to finish:

> days 1 to 30: your billing cycle. purchases stack up.
> the statement closes and the bill goes out.
> by law, you get at least 21 days from that statement to the due date.
> pay the full statement balance by the due date and every purchase on it cost you zero interest.

buy something on day 1 of the cycle and that's up to ~50 days of free float. the bank fronted the cash for almost two months and charged nothing for it.

now the part nobody explains at signup:

the grace period isn't a feature of the card. it's a feature of your behavior.

carry even $1 past the due date and on most cards it's gone:

> interest hits the whole balance, calculated daily, not monthly
> new purchases start accruing from the day you swipe, not the due date
> it usually stays gone until you've paid in full again. on some cards, two statements in a row.
> pay it all off and you can still get one more charge next month for the days in between. that one's called trailing interest.

and the sticker apr hides the mechanism. a 22% apr is really 0.06% charged every single day, on a balance that includes yesterday's interest.

cash advances? no grace period, ever. the clock starts the second the cash leaves the machine, plus a fee.

the cost of misreading this one clock:

> roughly half of active us card accounts carry a balance month to month
> americans pay north of $100 billion a year in card interest

here's why this matters beyond your everyday card.

almost every loan product on earth starts the interest clock the day you draw. including most loans against crypto.

a card with a grace period doesn't start the clock at all, as long as you pay by the due date.

same dollar borrowed. two completely different clocks.

so before you take on any form of credit, ask one question:

when does the interest clock actually start?

it's always in the agreement. most people learn it from the first interest charge instead.

[Ever @ever_card]

wall street's most profitable habit is also its most boring product:

wealthy people don't sell their assets. they borrow against them.

it's called securities-based lending. here's the whole mechanism:

> you hold a $1m stock portfolio
> the bank opens a credit line against it, usually 50 to 70 cents on the dollar
> the shares never leave your account. still yours, still compounding.
> you draw when you need cash and repay on your schedule

private banks have run this play so long the european version is named after medieval italian moneylenders: lombard loans.

and the scale is absurd:

> morgan stanley's book alone has topped $75 billion
> bank of america's: north of $65 billion
> total margin balances at us brokerages have run past $800 billion

morgan stanley's own ceo described the product in one line: "you lend wealthy clients their money back."

why borrow instead of sell?

> the asset keeps compounding while you spend
> no forced exit from your highest-conviction position
> selling creates a second decision: when to buy back in. most people never do at a better price.

now look at bitcoin holders.

many are up 5x, 10x, 20x. net worth transformed. spending power: unchanged.

and the asset they have the most conviction in is the one asset their bank won't lend against.

so they do the one thing private banking exists to prevent:

they sell their best asset to cover ordinary life.

the hodler's dilemma was never really about volatility. it's about access.

the playbook is a century old. it just wasn't built for you.

so before you sell something you plan to hold for a decade, ask the private-banker question:

why sell what you can borrow against?

[Ever @ever_card]

in crypto lending, a margin call usually isn't a call.

it's a receipt.

your collateral was sold while you slept. the email is just the paperwork.

here's how it actually works:

when you borrow against bitcoin, the loan lives or dies by one number:

LTV. loan-to-value. what you owe divided by what your collateral is worth.

> borrow $30k against $100k of btc = 30% ltv
> btc drops 40%: same loan, $60k of collateral = 50% ltv
> every lender has a line. cross it, and they can sell your coins to protect the loan.

nothing wrong with that. it's how secured lending has worked for a century.

the difference is what happens at the line.

in traditional finance, a margin call is a process:

> your broker notifies you
> you typically get days to add collateral or pay down
> you have a say in what gets sold if it comes to that

in most of crypto, a margin call is an event:

> an engine detects the breach
> it sells. instantly. at 3am. into whatever liquidity exists.
> often all of it, not just enough to fix the ratio

now watch what that design does at scale.

may 2021:

> btc roughly cut in half in a matter of weeks
> each leg down pushed thousands of loans across their line at once
> engines dumped collateral into a falling market
> that selling pushed prices lower, which breached more loans, which forced more selling
> on the worst day, over $8 billion in positions were force-liquidated in 24 hours

that's a liquidation cascade. the safety mechanism became the crash.

volatility didn't do that. design did.

the same drawdown treats two borrowers completely differently depending on three choices their lender made:

> warned before, or after?
> time to cure, or milliseconds?
> sell everything, or only enough?

so before you pledge a single sat, ask your lender one question:

if my LTV breaches at 3am, what happens in the next hour, and what am i allowed to do about it?

the answer tells you whether you're a client or exit liquidity.

[Ever @ever_card]

@brian_armstrong Buying a house without selling is the big version.
Buying anything without selling is the frequent one.

[Ever @ever_card]

@Kalshi what's new

[Ever @ever_card]

@cryptorover Big deal if you’re building on a sponsor bank.
It doesn’t lower the bar. It means you get judged on your controls, your capital, and your compliance package rather than on your category. Better world for anyone who did the work.

[Ever @ever_card]

@AshCrypto All I see is better buy-ins

[Ever @ever_card]

@APompliano That lag has been 70 days, 90 days, and 108 days depending on when the chart was screenshotted.
Refitting after every miss isn’t a leading indicator. It’s a rolling apology.

[Ever @ever_card]

@cryptothedoggy Need to hodl for Uptober

[Ever @ever_card]

@cryptorover After hodling for this long… might as well keep going.