Coinpric

Blog

  • The five ways people actually lose money in crypto

    The five ways people actually lose money in crypto

    The short version

    Being wrong about an asset is a small, survivable cost. Accounts are destroyed by five mechanisms instead: a position too large for the account, leverage that removes the room to be temporarily wrong, transaction costs compounding invisibly, an irreversible custody mistake, and theft. Every one is addressable without any ability to forecast, which makes them the highest-value thing to work on.

    There is a comforting story in which losing money in crypto is the result of a bad call — you thought it would go up and it went down. That story is comforting because it makes the problem sound like a skill you could acquire.

    It is mostly not what happens. Being wrong about direction, at a sensible size, costs a small percentage and teaches you something. What actually removes accounts is structural, and the five mechanisms below are all fixable by decisions made before any market opinion is formed.

    1. Size: the position was too large for the account

    The most common and the most avoidable. The asset does something ordinary — a 40% decline is an unremarkable event in crypto — and because the position was a large share of the account, an ordinary move becomes an unrecoverable one.

    The reason recovery is so hard is arithmetic, and it is worth seeing rather than being told:

    Loss taken Gain needed to get back to even
    20% 25%
    50% 100%
    75% 300%
    90% 900%

    At the top you are inconvenienced. At the bottom you are finished, because a 900% gain is not something anyone can plan around. This asymmetry is why avoiding large losses matters more than catching large gains, and why drawdown is the number experienced traders watch instead of returns.

    The fix, which requires no forecasting: decide what percentage of the account you can lose on one position before you open it, and let that plus your invalidation level determine the size. How to size a crypto position goes through it, and our position size calculator does the arithmetic.

    2. Leverage: no room to be temporarily wrong

    Leverage does not make you wrong. It removes the time to be right.

    At 10x, roughly a 9.5% adverse move liquidates the position. At 25x, about 3.5%. At 100x, about 0.5% — which is ordinary noise, a widened spread, a single wick. So a correct view held with high leverage is routinely liquidated before the view is vindicated, and the outcome is identical to having been wrong.

    There is a second, quieter drain. Funding on a perpetual is charged on notional rather than margin, so a typical 0.01% per eight hours is roughly 110% a year against your margin at 10x. The full arithmetic is in perpetual futures explained.

    The fix: know your liquidation price before opening, and treat it as a real number that ordinary volatility will reach. If your idea needs weeks to work and your leverage gives it hours, the leverage is the problem.

    3. Costs: the drain nobody totals up

    This one is invisible because it never arrives as a single event. Each trade costs a commission, half the spread and some slippage — realistically around 0.23% one way on a venue advertising 0.10%.

    Twenty round trips at 0.46% each leaves 0.9954 to the power of 20, which is 0.912. Nearly 9% of the account, gone, with no losing trade required. Someone trading several times a week can spend a quarter of their capital on costs in a year while believing they are roughly break-even.

    The fix: trade less, and measure what you actually pay against the mid price before each trade. The real cost of a trade has the method.

    4. Custody: irreversible, and usually not dramatic

    These losses rarely involve a market at all. A seed phrase lost or never written down. A transfer to a wrong address. Sending an asset over the wrong network. An exchange that fails while holding the balance, or freezes an account.

    What unites them is finality. There is no support desk for a blockchain transaction, and no reversal for a valid transfer. A market loss can be recovered from; a custody loss is permanent.

    The fix: test everything at trivial size first — a small withdrawal, a small transfer on each network, and a wipe-and-restore of any wallet before it holds anything meaningful. Then split holdings so no single venue’s failure and no single mistake can reach everything. What “not your keys, not your coins” really means covers both sides.

    5. Theft: engineered, not opportunistic

    Crypto theft is overwhelmingly about persuading you to authorise something, not about breaking cryptography.

    The recurring patterns: a “support representative” who contacts you first and needs your phrase; a wallet application downloaded from an advertisement rather than the official source; a transaction approval that grants unlimited access to your holdings rather than the single transfer you expected; a phone number transferred away from you so SMS codes arrive at someone else’s device; and an unsolicited message about an opportunity, which is never anything else.

    The fix: use an authenticator app rather than SMS. Never enter a seed phrase anywhere except restoring your own wallet. Read what a transaction approval actually grants before signing it. And treat every unsolicited approach as an attempt, because the base rate is close to one.

    What is not on this list

    Being wrong about an asset. It belongs in a different category entirely, because at a sensible size it costs a few percent and produces information. Every mechanism above turns a survivable error into a terminal one — which means the useful question is not “how do I predict better” but “what would make being wrong survivable”.

    That question has answers, all of them available to anyone, none requiring any view about where prices go. It is also, unhelpfully for anyone selling anything, extremely boring: size sensibly, avoid leverage, trade rarely, test transfers small, and never type your seed phrase. That is most of it.

    Nothing here is financial advice, and none of it is a promise that following it produces gains — it will not. Crypto assets are volatile enough to lose your entire position regardless of process. See our disclaimer.

    Key takeaways

    • Accounts are destroyed by size, leverage, costs, custody mistakes and theft — rarely by being wrong about direction.
    • A 50% loss needs a 100% gain to recover, and a 90% loss needs 900%. Avoiding large losses beats catching large gains.
    • Leverage does not make you wrong; it removes the time to be right. At 100x, ordinary noise liquidates you.
    • Twenty round trips at realistic costs removes nearly 9% of an account with no losing trade at all.
    • Custody losses are permanent. Test every transfer and every wallet restore at trivial size first.
    • Theft works by persuading you to authorise something. Every unsolicited approach is an attempt.
    • The fixes need no forecasting ability, which is what makes them the highest-value work available.
  • Custody: what “not your keys, not your coins” really means

    Custody: what “not your keys, not your coins” really means

    The short version

    An exchange balance is a claim on a company. A private key is direct control. The slogan is right that these are different things, and it usually stops before the important part: self-custody does not eliminate risk, it transfers it to you, and your mistakes are irreversible in a way an exchange’s usually are not. There is no safe option — only two different failure modes, and a sensible split between them.

    “Not your keys, not your coins” is one of the few crypto slogans that is technically accurate. It is also usually deployed as though it settled the question, and it does not. It describes what you gain by holding your own keys and says nothing about what you take on.

    What an exchange balance actually is

    When you hold crypto at an exchange, the coins are not yours in the sense a slogan implies. The exchange controls keys to addresses holding assets in aggregate; your balance is an entry in its internal database representing a claim against the company.

    In ordinary operation this is invisible and irrelevant — you can trade and withdraw, so it feels like ownership. It becomes very relevant in exactly two situations: if the company becomes insolvent, where you are a creditor and your recovery depends on the insolvency process and on whether assets were segregated; and if the company restricts your account, for compliance, sanctions, suspected fraud or error, where the balance is intact and you cannot reach it.

    That is the real content of the slogan. Not that exchanges steal, but that an exchange balance is a claim, and claims can fail while the assets still exist.

    What self-custody genuinely gives you

    Holding the keys means controlling the addresses. No company can freeze it, no insolvency can capture it, no compliance decision can suspend it, and no counterparty can lend it out. For an asset whose entire proposition is that it does not require permission, this is the version that delivers on it.

    Those are real, valuable properties. They are also the complete list — and each one has a mirror image.

    What self-custody takes on

    Every protection you removed was also a safety net.

    No password reset. Lose the seed phrase and the assets are permanently unreachable. Not frozen, not disputed — gone, provably still sitting at an address nobody can open. A meaningful share of all bitcoin is estimated to be in exactly this state.

    No reversals. Send to the wrong address and there is no support desk. The transaction was valid and executed as instructed.

    No fraud protection. If you are deceived into authorising a transfer, that transfer is final. Card networks reverse fraudulent payments; blockchains do not.

    You are now the target. Attacks shift from the exchange’s security team to you: fake wallet applications, malicious transaction approvals, clipboard malware that swaps addresses, and social engineering aimed at the phrase.

    Inheritance becomes your problem. If you die without your family being able to reach the assets, they are lost. An exchange has a probate process, however tedious. A seed phrase only you know does not.

    The comparison, without a winner

    Risk At an exchange In self-custody
    Company failure Real, and outside your control None
    Account frozen Possible at any time Impossible
    You lose access Recoverable via support Permanent and total
    Mistaken transfer Sometimes recoverable internally Never recoverable
    Being personally targeted Lower; the venue is the target Higher; you are the target
    Inheritance A defined legal process Entirely your arrangement

    Read down the two columns. Neither is safe. Exchange risk is someone else’s failure, generally with some recourse. Self-custody risk is your own failure, with none. Which you should prefer depends far more on your circumstances and habits than on any principle.

    The custody options, briefly

    Exchange account. Convenient, necessary for trading, and full exposure to one company.

    Software wallet. Keys on a phone or computer that you control. Genuine self-custody, but the keys live on an internet-connected device that may already be compromised.

    Hardware wallet. Keys generated and held on a dedicated device that signs transactions without exposing them. The practical standard for meaningful amounts. It protects the key from a compromised computer; it does not protect you from approving a bad transaction, and it does not protect the seed phrase you wrote down.

    Multi-signature. Requires several keys to move funds, so no single loss or compromise is fatal. Genuinely more robust, meaningfully more complex, and the complexity is itself a risk if you do not understand it.

    See cold storage and self-custody for the terms.

    What the seed phrase is, and the mistakes that lose it

    The phrase is not a password to a wallet — it is the mathematical source from which all your keys are derived. Anyone with it has your assets, from anywhere, forever. It is the only thing that matters.

    Which means the two failure modes are opposites, and defending against one worsens the other. Store it in too few places and you risk losing it to a fire or a flood. Store it in too many and you multiply the chances someone finds it.

    The mistakes that recur: photographing it, which puts it in cloud backups; typing it into anything, ever, other than restoring your own wallet; storing it in a password manager, which makes it as strong as that account; keeping the only copy where the device is, so one event takes both; and telling nobody it exists, which loses it on your death.

    The single most important rule is short: no legitimate person or service will ever ask for your seed phrase. Not support, not a wallet developer, not a migration tool. Every request is theft.

    What we would actually suggest

    Not “self-custody everything”, and not “leave it on the exchange”. Split it by function.

    Keep at an exchange what you are actively using — the amount you would trade or spend soon, sized so the venue’s failure would be a bad week rather than a catastrophe. Move the long-term holding to custody you control, and only after you have tested the recovery process.

    That last part is the step people skip. Set up the wallet, write down the phrase, then wipe the device and restore from the phrase before sending anything meaningful to it. An untested backup is not a backup. Restoring successfully is the only evidence that the words you wrote are the words you need.

    And write down, somewhere your family can find, that these assets exist and how someone competent could reach them. It is unpleasant to think about and it is the difference between an inheritance and a permanent loss.

    None of this is advice about how much to hold or where. It is a description of two sets of trade-offs. See our disclaimer.

    Key takeaways

    • An exchange balance is a claim on a company, and claims can fail while the assets still exist.
    • Self-custody removes company risk entirely and replaces it with your own error, which is permanent and unrecoverable.
    • Lost seed phrases are not disputes — the assets sit at an address nobody can ever open.
    • A hardware wallet protects the key from a compromised computer. It does not protect you from approving a bad transaction.
    • Storing a seed phrase in too few places risks loss; too many risks theft. Both are real failure modes.
    • No legitimate person or service ever asks for a seed phrase. Every such request is theft.
    • Wipe and restore from your written phrase before trusting it with anything. An untested backup is not a backup.
  • Your first crypto purchase, start to finish

    Your first crypto purchase, start to finish

    The short version

    Decide the amount before you decide the asset. Verify the venue’s licence on the regulator’s own register rather than trusting a logo. Make the first purchase small enough to be an experiment. Then, before you add anything, test a withdrawal end to end — because that is the step that reveals whether the account actually works, and the one almost nobody does first.

    Most first-purchase guides are structured around the exchange sign-up flow, which means they are structured around what the exchange wants you to do next. This one is ordered by what protects you, which puts some steps much earlier than usual.

    Step 1: decide the amount, before anything else

    The first decision is not which asset. It is how much, and the only sound answer is an amount whose complete loss would change nothing important in your life.

    That is not a rhetorical flourish. Crypto assets have repeatedly fallen 70% or more from a high, and individual tokens have gone to nothing permanently. Any amount you commit should be one you can watch fall by that much without it affecting rent, debt or sleep.

    Write the number down before you look at a single chart. Deciding the amount after forming a view about an asset means the view sets the size, and that is the mechanism behind most of the losses described in the five ways people actually lose money.

    Step 2: choose a venue, and verify it yourself

    You need somewhere that accepts your currency, operates in your jurisdiction, and is licensed there. That last point requires actual checking:

    • Find the legal entity name — usually in the footer or terms, not the marketing.
    • Search your regulator’s public register for that exact name.
    • Confirm the licensed entity is the one whose terms you will accept. Group structures sometimes route customers to a different company from the licensed one.

    A licence is not a guarantee, and licensed firms have failed. It does mean there is a supervisor, a complaints route and a defined insolvency process. An unlicensed venue offers none of those, whatever its interface looks like. What a review can and cannot establish about a venue is covered in how to judge an exchange.

    Step 3: expect identity verification, and prepare for it

    Any regulated venue will require identity documents, proof of address and often a source-of-funds declaration. This is a legal obligation on them, not an imposition on you.

    Two practical notes. Names must match your documents exactly — a mismatch between your bank account name and your exchange account name is the most common cause of a stuck deposit. And verification can take anywhere from minutes to days, so complete it before you have money waiting.

    Step 4: secure the account before it holds anything

    Do this while the account is empty, because it is the point of maximum motivation and zero urgency.

    Use a unique password stored in a password manager. Enable two-factor authentication using an authenticator app, not SMS — phone numbers can be transferred away from you by someone impersonating you to a mobile operator, which is a well-documented attack. Store the recovery codes somewhere offline. Enable withdrawal address allowlisting if offered, and any withdrawal delay the venue provides: a delay is only an inconvenience to you and a serious obstacle to someone else.

    Step 5: choose the asset, with the size already fixed

    Only now does the asset matter, and the size is already decided so the decision cannot inflate it.

    For a first purchase, prefer assets with long histories, deep liquidity and published coin pages you can actually read. Not because they will perform better — nobody knows that — but because they are the ones where information is verifiable and exit is possible at a price close to the one you see. Before buying anything less established, work through how to judge an altcoin, which is mostly a list of ways to find what would embarrass you later.

    Step 6: make the purchase small and deliberate

    Use a limit order rather than a market order if the interface offers one. It costs less, as the real cost of a trade works through, and more importantly it forces you to state a price rather than accepting whatever appears.

    Then check the confirmation against what you expected: units received, price paid, fee charged. This is where you discover the difference between the advertised fee and the amount actually deducted, and it is much better to discover it on a small transaction.

    Consider making the first purchase a fraction of your decided amount. The rest can follow once you know the mechanics work — and if you intend to spread purchases anyway, the arithmetic is in buying in instalments or all at once.

    Step 7: test a withdrawal, now, while it is small

    This is the step almost every guide omits and the one we would keep if we could keep only one.

    Send a small amount off the exchange — to a wallet you control, or back to your bank. You are checking that withdrawals are enabled on your account, that the limits are workable, that the fee is what was advertised, and that the funds actually arrive.

    Do it while the sum is trivial. Discovering that withdrawals require additional verification, or are capped far below your deposit limit, is a minor annoyance at a small size and a serious problem at a large one. An account you have never withdrawn from is an account you have not finished testing.

    Step 8: decide where it lives

    Assets left on an exchange are held by that company, and its failure is your loss. Assets in a wallet you control cannot be frozen by anyone — and cannot be recovered by anyone either, including you. Neither is safe; they fail differently. What “not your keys, not your coins” really means covers both sides honestly, and it is worth reading before you move anything, because the mistakes in self-custody are permanent in a way exchange mistakes usually are not.

    Step 9: start the tax record today

    From the first transaction, record the date and time, what you bought, the amount in your own currency, the fee and the venue. In most jurisdictions the purchase itself is not a taxable event, but it establishes the cost that every future calculation depends on.

    Reconstructing this later, across venues that have changed their export formats or closed, is the single largest avoidable cost in crypto — see crypto tax: the four questions every jurisdiction asks. A spreadsheet started on day one takes seconds per transaction.

    What not to do in the first month

    Do not use leverage. The arithmetic in perpetual futures explained shows that ordinary volatility liquidates a high-leverage position before any view has time to be right.

    Do not act on unsolicited contact. Nobody legitimate messages a stranger about an investment opportunity, and no support representative ever needs your recovery phrase or password.

    Do not add to a losing position to lower your average, having not planned to. That is a decision made by discomfort, and it is how a small mistake becomes the whole account.

    Nothing in this article is financial advice, and none of it is a recommendation to buy any asset. It is a description of a careful process. Read our disclaimer for the full position.

    Key takeaways

    • Decide the amount before the asset, and write it down before looking at a chart.
    • Verify the venue’s licence on the regulator’s own register, and check the licensed entity is the one you contract with.
    • Secure the account while it is empty: authenticator app rather than SMS, offline recovery codes, withdrawal allowlisting.
    • Use a limit order — it costs less and forces you to state a price instead of accepting one.
    • Test a small withdrawal end to end before adding funds. An account you have never withdrawn from is untested.
    • Start the tax record on day one. Reconstruction later is the biggest avoidable cost in crypto.
    • In the first month: no leverage, no unsolicited contact, and no unplanned averaging down.
  • Buying in instalments or all at once: what the arithmetic actually says

    Buying in instalments or all at once: what the arithmetic actually says

    The short version

    Buying fixed amounts at intervals gives you an average cost below the average of the prices you paid. That is not a claim, it is arithmetic, and it always holds. It does not follow that averaging in beats buying at once — in a rising market, buying at once wins clearly, because more of your capital was exposed for longer. Two worked examples below, one in each direction, with the same numbers.

    The argument about averaging in versus buying at once is usually conducted with confident claims and no numbers. It is a question with an arithmetic core, and once you do the arithmetic the disagreement mostly dissolves — because the two sides are each right about a different thing.

    The provable part

    Spending a fixed amount at each interval buys more units when the price is low and fewer when it is high, automatically. Your average cost per unit is therefore not the average of the prices — it is the harmonic mean of them, which is always lower whenever prices vary at all.

    Four purchases of 100 each, at prices of 100, 50, 25 and 50:

    Purchase Price Amount spent Units bought
    1 100 100 1.00
    2 50 100 2.00
    3 25 100 4.00
    4 50 100 2.00
    Total 400 9.00

    Average cost is 400 divided by 9, which is 44.44. The simple average of the four prices is 56.25. You paid 21% less per unit than the average price, without predicting anything.

    This is a mathematical identity, not a market observation. The harmonic mean of a set of positive numbers is always less than or equal to their arithmetic mean, with equality only when every number is identical. So the effect never fails — it just gets smaller as prices get steadier.

    Why that does not settle the question

    The comparison people actually care about is not “average cost versus average price”. It is “instalments versus buying the whole amount at the start”. Those are different questions, and the second has a different answer.

    Same 400, same four periods. Buying at once means 4 units at 100. Ending price 50:

    Approach Units held Value at 50
    Instalments 9.00 450.00
    All at once 4.00 200.00

    Instalments win decisively — a falling-then-recovering path is the best case for them. Now the same method on a rising path, prices 100, 125, 150, 175:

    Approach Units held Average cost Value at 175
    Instalments 3.04 131.66 531.67
    All at once 4.00 100.00 700.00

    Buying at once wins by a wide margin. Note that instalments still delivered on their promise — the average cost of 131.66 is below the 137.50 average of the prices paid. The mechanism worked perfectly and the outcome was still worse, because three quarters of the capital arrived after the asset had already risen.

    That is the whole resolution. Averaging in reliably improves your entry price relative to the prices available. Buying at once maximises time exposed. When an asset rises over the period, exposure dominates. When it falls and recovers, entry price dominates. Neither approach is better in general, because the question is really about which of those two you are trying to optimise.

    What actually decides it for a real person

    Since the arithmetic does not produce a winner, the decision rests on circumstances — and here the considerations are more practical than mathematical.

    Do you have a lump sum at all? Most people investing from income never face this choice. Money arrives monthly, so it is invested monthly. The comparison is academic, and the honest framing is that instalments are not a strategy but a description of the cashflow.

    What happens if you are immediately down 40%? This is the question the arithmetic cannot reach, and it decides more outcomes than any calculation. A plan abandoned at the worst moment performs far worse than either approach followed consistently. If buying at once would mean watching a large loss on a single decision you made on one day, and that would make you sell, then instalments are better for you — not because the expected value is higher, but because you will still be there.

    How volatile is the asset? The benefit of averaging scales with dispersion. In a market that routinely halves and doubles, the gap between harmonic and arithmetic mean is large. In a stable one it is negligible, and the fixed costs of many small purchases can exceed the benefit.

    What do the transactions cost? Twelve purchases incur twelve sets of fees and twelve crossings of the spread. On the numbers in the real cost of a trade, at roughly 0.23% per purchase that is about 2.8% of the total — which can wipe out the averaging benefit entirely on a low-volatility asset. Fewer, larger instalments are usually better than many tiny ones.

    Two things this does not do

    It is not risk reduction. Once fully invested you hold exactly the same position either way, with exactly the same exposure. Averaging in changes the path to the position, not the position. Describing it as “reducing risk” conflates the transition with the destination.

    It does not protect against a permanent decline. If an asset falls and never recovers, averaging in means buying more of it on the way down. The lower average cost is worthless if the price never returns. Averaging is a method for handling volatility, not for handling being wrong about the asset.

    Our instalment calculator will run these numbers on whatever schedule and prices you want to test, including the unfavourable paths. Test both directions, because a tool run only on a chart that went up is a machine for producing confidence. Nothing here is advice about which to use, or about whether to buy anything at all — see our disclaimer.

    Key takeaways

    • Fixed-amount buying gives an average cost equal to the harmonic mean of prices, always at or below their average.
    • In the worked falling-then-recovering example: 44.44 average cost against a 56.25 average price, 21% better.
    • On a rising path, buying at once won 700 to 531.67 — the averaging mechanism worked and the outcome was still worse.
    • Averaging optimises entry price; buying at once optimises time exposed. Which wins depends on the path.
    • Most people investing from income are not making this choice at all — monthly cashflow decides it for them.
    • At roughly 0.23% per purchase, twelve instalments cost about 2.8% and can erase the averaging benefit.
    • Averaging in changes the path, not the destination. It is not risk reduction, and it does not help if you are wrong about the asset.
  • Why the same chart looks bullish and bearish at the same time

    Why the same chart looks bullish and bearish at the same time

    The short version

    A chart is not a picture of the market; it is a picture of the market filtered through one timeframe. The same price history genuinely looks like an uptrend on one interval and a downtrend on another, and neither is wrong — they answer different questions. Almost every unresolvable chart argument is two people using different intervals without saying so. Decide your horizon first, then open the chart.

    Two competent people can look at the identical price history and reach opposite conclusions, with neither making an error. This is not a flaw in technical analysis; it is a direct consequence of what a chart is. Understanding it removes most of the confusion in chart-based discussion.

    A chart is a lossy summary

    Price is a continuous stream of transactions. A chart takes that stream, divides it into intervals, and reduces every interval to four numbers: the first trade, the highest, the lowest and the last. Everything else is discarded.

    That reduction is what makes a chart readable. It also means the interval is a parameter you chose, and it determines what survives. Move from an hourly to a daily view and twenty-three of every twenty-four data points vanish. The remaining one is not more true. It is a different summary of the same reality.

    Two things follow. Structure appears and disappears with the interval — a clear pattern on one is invisible on another, and neither is an illusion. And a single dramatic candle usually contains a long, messy sequence that a shorter interval would have shown as several distinct moves.

    Why contradictory readings are both correct

    Consider an asset that fell sharply over three months and has risen steadily for two weeks. On a daily chart the fall dominates and the recent rise is a modest bounce inside a downtrend. On a four-hour chart the rise is a clean series of higher lows — an uptrend.

    Both descriptions are accurate. They answer different questions: “what has the last quarter done” and “what has the last fortnight done”. The disagreement only exists because the timeframe was left implicit, which it almost always is.

    This is why “is bitcoin in a bull market” has no answer without a horizon attached. Over ten years, over one year, and over one month it can be simultaneously up, down and flat. Every one of those statements is checkable and true.

    The trap: shopping for the timeframe that agrees with you

    Here is where a descriptive problem becomes an expensive one.

    Given enough intervals — one minute, five, fifteen, hourly, four-hour, daily, weekly, monthly — at least one will support almost any view at almost any moment. So if you open a chart wanting a conclusion, you will find a timeframe that provides it, and it will look like analysis rather than selection.

    The tell is switching intervals after forming a view. Someone who is long and looking at a weakening hourly chart moves to the daily “for context”. Someone who wants to buy and sees an ugly daily chart moves to the fifteen-minute “for an entry”. Both feel like diligence. Both are the same act: retaining the picture that agrees and discarding the ones that do not.

    The fix is procedural, not analytical. Decide your horizon before you open the chart, based on how long you intend to hold and how often you can genuinely pay attention — then read that timeframe and accept what it says. If you change interval, change it deliberately and note why, so you can see later whether the reason was a reason.

    What each timeframe is actually for

    Interval Answers Signal-to-noise
    1-5 minute Where is liquidity right now Very low; mostly microstructure
    15 minute – 1 hour What is today’s session doing Low; dominated by positioning
    4 hour What is this week’s move Moderate
    Daily What is the current phase Higher; the common reference
    Weekly What is the multi-month trend High, but slow to update
    Monthly What is the multi-year picture Highest, and nearly useless for timing

    Read the right-hand column carefully. Signal quality improves as you go down, and responsiveness collapses. There is no interval that is both reliable and fast — that trade-off is the whole difficulty, and any method claiming to have solved it has instead hidden it.

    Two related distortions

    The axis. A linear axis gives equal space to equal absolute moves; a logarithmic axis gives equal space to equal percentage moves. For an asset that has changed by orders of magnitude, a linear chart compresses all early history into a flat line and makes recent volatility look unprecedented. Neither axis is dishonest, but the choice materially changes the impression, and for a long history the logarithmic view is usually the more informative one.

    The window. Where a chart starts is a choice, and starting at a peak or a trough produces very different pictures of the same asset. When you see a chart in an article, check the start date before the shape.

    How to use multiple timeframes without fooling yourself

    Looking at more than one is legitimate — the discipline is assigning each a fixed job in advance and not letting them trade places.

    A workable arrangement: use a slower interval to establish context, and a faster one only for execution timing, never to overrule the context. So the daily chart decides whether you are interested at all, and the four-hour decides when — but the four-hour never gets a vote on whether. Write it down, because the temptation to let the fast chart overrule the slow one arrives exactly when the fast chart is most exciting and least informative.

    The broader point: a chart cannot tell you what will happen. It shows what has happened, filtered through choices you made. Being explicit about those choices is the difference between reading a chart and being read by it — which is the same argument we make in how to read a crypto chart without fooling yourself and in why support and resistance are not lines on a chart. Nothing here is advice; see our disclaimer.

    Key takeaways

    • A chart reduces every interval to four numbers. The interval is a parameter you chose, and it decides what survives.
    • Contradictory readings on different timeframes are usually both correct — they answer different questions.
    • “Is this a bull market” has no answer without a stated horizon.
    • Given enough intervals, one will support any view. Switching interval after forming a view is the tell.
    • Decide your horizon before opening the chart, based on holding period and attention available.
    • Signal quality rises and responsiveness collapses as intervals lengthen. No interval is both reliable and fast.
    • Give each timeframe a fixed job in advance: the slow one decides whether, the fast one only decides when.
  • The real cost of a trade: fee, spread and slippage

    The real cost of a trade: fee, spread and slippage

    The short version

    A trade costs the commission, plus half the bid-ask spread, plus slippage from walking the order book, plus any funding or transfer costs. On a venue advertising 0.10%, a realistic total is around 0.23% one way — more than double the headline. Compounded over twenty round trips, that removes close to 9% of an account before a single directional decision has been judged.

    Fee comparisons between exchanges usually put a commission table side by side and declare a winner. That table is the smallest of the costs for most people, and the venue with the lowest advertised fee is frequently not the cheapest place to trade.

    Here is the full cost, with the arithmetic done rather than asserted.

    Cost 1: the commission

    Two rates, and the distinction is worth money. A maker fee applies when your order rests in the order book and someone else trades against it — you supplied liquidity. A taker fee applies when you trade against a resting order — you removed liquidity. Taker fees are higher, often by two or three times, and some venues rebate makers entirely.

    Practically: a market order always pays taker. A limit order placed away from the current price and left to fill pays maker. If your strategy tolerates waiting, this alone can halve your commission — and it is a choice most people make unconsciously by defaulting to market orders.

    One warning about limit orders. A limit order priced through the spread — a buy above the current offer — executes immediately against resting size and is charged as a taker order, because it removed liquidity. The label on the order type is not what determines the fee; whether you supplied or removed liquidity is. Some venues offer a post-only flag that cancels the order rather than letting it cross, which is the only reliable way to guarantee the maker rate.

    Cost 2: half the spread

    The spread is the gap between the best bid and the best offer. It is not a fee and appears on no statement, which is exactly why it goes unnoticed.

    Take a mid price of 100.00 with a bid of 99.95 and an offer of 100.05. The spread is 0.10, or 0.10% of mid. If you buy at the offer you pay 0.05% above mid; if you then sell at the bid you give up another 0.05%. So a round trip pays the full spread — 0.10% — with no fee involved at all.

    This is why a venue with a slightly higher commission and a much tighter spread is often cheaper in practice, and why comparing fee schedules in isolation is close to meaningless.

    Cost 3: slippage

    The quoted offer is only good for the size resting at it. If your order is larger, it consumes that level and continues into the next, and the next. Your average fill is worse than the price you saw. That difference is slippage, and it grows with your size relative to the book’s depth.

    Two things make it much worse. Thin markets: on a small asset, an ordinary order can be a large fraction of the visible book. And volatile moments — which is when people most often use market orders, so the cost peaks exactly when it is least affordable.

    Adding it up honestly

    A 10,000 trade on a venue advertising a 0.10% taker fee, with the spread above and modest slippage:

    Component Rate Cost
    Taker commission 0.100% 10.00
    Half the spread 0.050% 5.00
    Slippage 0.080% 8.00
    Total, one direction 0.230% 23.00
    Round trip 0.460% 46.00

    The real cost is 2.3 times the advertised one. And the compounding is the part that surprises people: at 0.46% per round trip, twenty round trips leave you with 0.9954 raised to the power of 20, which is 0.912. Nearly 9% of the account has gone on costs alone — before any judgement about direction has been assessed.

    That single calculation is the strongest argument against high trading frequency that exists, and it requires no view on markets whatsoever.

    Why the fee tier you were shown is probably not yours

    Almost every venue publishes a tiered schedule where the rate falls as monthly volume rises, and the headline number in comparisons is often taken from somewhere down that table rather than the top of it. Two things follow.

    First, check which tier you are actually in. The entry-level rate is frequently several times the rate quoted in a review, and the volume needed to reach the middle of the table is typically far beyond what an individual generates in a month. The relevant number is the one on the row you occupy.

    Second, notice the incentive the structure creates. A schedule that rewards volume rewards trading more, and trading more is precisely what the arithmetic above says destroys accounts. A tier discount of a few basis points is worth having; it is not worth a single extra round trip taken to qualify for it, because the round trip costs more than the discount saves. Some venues also discount fees when you hold or pay in their own token, which converts a cost saving into exposure to that token’s price — a trade you may not have intended to make.

    The costs that hide outside the trade

    Deposit and withdrawal. Crypto withdrawal fees are sometimes a flat amount well above the actual network cost. Fiat rails vary from free to punitive. If you move funds often, this can exceed your trading costs.

    Conversion. Trading a pair not quoted in your own currency means an implicit conversion, usually at a spread rather than a stated fee. Two trades to reach a position means two conversions.

    Funding, on perpetuals. Holding a leveraged perpetual position means paying or receiving funding periodically. This is a recurring cost that scales with notional, not with margin, and it can dwarf commissions on a position held for days. We work through the numbers in perpetual futures explained.

    Price impact on-chain. On an automated market maker the equivalent of slippage is set by the pool curve and can be far larger than any fee, as the worked example in lending pools and market makers in plain terms shows.

    Measuring your own cost, which nobody does

    All of the above is an estimate. Your actual cost is measurable, and the method is simple enough to do by hand.

    Before you send an order, note the mid price — halfway between the best bid and offer. After it fills, compare your average fill price to that number, and add the commission. The gap is what the trade genuinely cost you, and it captures the spread and the slippage together without needing to separate them.

    Do this for ten trades and you will have something far more valuable than any published fee comparison: your own realised cost, at your own size, on the venue you actually use, in the conditions you actually trade. Two things usually emerge. The cost is larger than expected. And it varies enormously with when you traded — the same order in a quiet hour and in the first minute of a violent move are different transactions at very different prices.

    Keep the record in whatever you already use. The point is not sophistication; it is that a number you measured beats a number you were quoted.

    How to reduce it, in order of effect

    Trade less. This dominates everything else on the list, by the arithmetic above.

    Then: use limit orders where you can, with post-only if the venue offers it, to pay maker instead of taker and to avoid slippage entirely; check depth rather than headline volume before sizing; avoid trading in the first minutes of a violent move, when spreads widen and depth vanishes; and consolidate transfers rather than moving funds repeatedly.

    What matters is that these are all controllable. Direction is not. Costs are the part of the outcome you can actually decide, which is a good reason to measure them properly rather than accept the number on the marketing page.

    Key takeaways

    • The advertised commission is usually the smallest of three costs.
    • A round trip pays the full bid-ask spread even if the fee were zero — and the spread appears on no statement.
    • A 0.10% venue realistically costs about 0.23% one way once spread and slippage are included.
    • At 0.46% per round trip, twenty round trips remove nearly 9% of an account before any directional call is judged.
    • A limit order that crosses the spread is charged as a taker order. Only post-only guarantees the maker rate.
    • Check which fee tier you actually occupy — and never take an extra trade to qualify for a discount.
    • Measure your realised cost against the mid price before you traded. A number you measured beats one you were quoted.
  • Perpetual futures: funding, leverage and the liquidation price

    Perpetual futures: funding, leverage and the liquidation price

    The short version

    A perpetual future never expires, so instead of settling at maturity it uses a periodic funding payment between longs and shorts to hold it near the spot price. Leverage sets how far price can move before you are liquidated: at 10x that is roughly 9.5%, at 25x roughly 3.5%. Funding looks trivial per payment and is not — 0.01% every eight hours is about 11% a year on notional, which at 10x is nearly 110% a year against your margin.

    Perpetual futures are the highest-volume instrument in crypto, and they are structurally unlike anything in traditional markets. Two mechanisms — funding and liquidation — determine essentially everything about how they behave. Both are arithmetic.

    Why “perpetual” required inventing something

    A conventional future has an expiry date, and that date does the work of keeping it honest: at settlement the contract must converge on the spot price, because it settles against it. Arbitrage enforces convergence in advance.

    Remove the expiry and you remove the anchor. A perpetual could drift arbitrarily far from spot with nothing to pull it back. The solution is the funding rate: a payment made directly between holders of long and short positions at regular intervals, commonly every eight hours.

    The direction follows the deviation. If the perpetual trades above spot, longs pay shorts — being long becomes expensive and being short is paid, which pushes the price back down. If it trades below spot, shorts pay longs. The exchange is not a party to this; it moves between users.

    Funding: small numbers, large annualised cost

    Funding rates are quoted per interval, which makes them look negligible. Annualise them and the picture changes completely. With three payments a day:

    Funding per 8h Per day Per year on notional Per year on margin at 10x
    0.01% (typical) 0.03% 10.95% 109.5%
    0.05% 0.15% 54.75% 547.5%
    0.10% 0.30% 109.5% 1,095%

    The final column is the one that matters and the one nobody quotes. Funding is charged on your position’s notional value, not on the margin you posted. At 10x leverage your notional is ten times your margin, so the cost relative to the capital actually at risk is ten times the headline rate.

    The first row is an ordinary, unremarkable funding rate. Held for a year at 10x it costs more than your entire margin. This is why perpetuals are instruments for short holding periods, and why using one as a substitute for owning the asset is expensive in a way that is easy to miss — the payments are small, frequent, and never appear as a single alarming number.

    Funding also tells you something. Persistently positive funding means the market is crowded long, and crowded positioning is what makes liquidation cascades possible. It is a positioning gauge, not a directional signal.

    Leverage: what the multiple actually buys

    Leverage means posting margin worth a fraction of the position. At 10x, 1,000 of margin controls 10,000 of exposure. Gains and losses accrue on the 10,000.

    The consequence people underestimate is not the amplification — that part is obvious — but how little room remains:

    Leverage Initial margin Adverse move to roughly liquidate
    2x 50% ~49%
    5x 20% ~19%
    10x 10% ~9.5%
    25x 4% ~3.5%
    50x 2% ~1.5%
    100x 1% ~0.5%

    The arithmetic is simply the initial margin percentage minus the maintenance margin the venue requires. Set that against reality: bitcoin routinely moves several percent in a day. At 25x, an ordinary session is enough. At 100x, ordinary noise — a spread widening, a brief wick — will do it. High leverage does not increase your expected return; it increases the probability that ordinary volatility removes you before your idea has had time to be right or wrong.

    Liquidation: how the position actually ends

    Your position is closed when equity falls to the maintenance margin. Three details decide how much you lose.

    The liquidation price is computed from a mark price, not the last trade. Venues use an index across several spot markets specifically so a single venue’s wick cannot trigger liquidations. This protects you, and it means your liquidation level may not correspond to any price you saw printed.

    Liquidation is a market order. It closes at whatever the book offers. In a fast move that is meaningfully worse than the liquidation price, which is why realised losses often exceed the calculation.

    Isolated versus cross margin changes what is at stake. Isolated margin risks only what you assigned to that position. Cross margin uses your whole account balance as collateral — which delays liquidation but puts everything behind one trade. Choosing cross margin without understanding this is a common and expensive mistake.

    Venues also maintain an insurance fund for cases where a liquidation closes worse than bankruptcy price, and some operate auto-deleveraging that can close profitable opposing positions when that fund is exhausted. It is rare. It is not impossible, and it means an extreme event can affect you even when your own position was fine.

    The cascade, which is the whole risk picture

    Put the pieces together. Leverage concentrates liquidation prices at predictable levels. Price reaching one triggers forced market selling. That selling pushes price to the next cluster. Each stage feeds the next, and the mechanism requires no manipulation — just crowded positioning and a starting push.

    This is why the largest single-day moves in crypto tend to be liquidation events rather than news events, and why they so often retrace substantially within a day. Nothing changed about the asset; positions were removed.

    The honest summary

    Perpetuals are well-engineered instruments that do exactly what they say. They are also the fastest route to total loss available in this market, because funding erodes a position quietly while leverage removes the room to be temporarily wrong.

    If you use them, know your liquidation price before you open, size from that rather than from conviction — our position size calculator does the arithmetic, and how to size a crypto position explains the reasoning — and annualise the funding rate before assuming it is trivial. Nothing here is a recommendation to trade these; total loss is a realistic outcome. See our disclaimer.

    Key takeaways

    • Perpetuals have no expiry, so a periodic funding payment between longs and shorts holds them near spot.
    • Funding is charged on notional, not margin — so at 10x the cost against your capital is ten times the headline rate.
    • A typical 0.01% per eight hours is about 11% a year on notional, and nearly 110% a year against margin at 10x.
    • Liquidation arrives after roughly a 9.5% adverse move at 10x, 3.5% at 25x and 0.5% at 100x.
    • The liquidation price uses an index mark price, so it may not match any price you saw printed.
    • Liquidation executes as a market order, so realised losses frequently exceed the calculated level.
    • Cross margin puts your entire balance behind one position. Isolated margin risks only what you assigned.
  • How to judge an exchange, and what a review cannot tell you

    How to judge an exchange, and what a review cannot tell you

    The short version

    Exchange reviews grade fees, interface and asset selection because those are observable. The failures that actually destroy accounts — reserves that do not match liabilities, commingled customer funds, a withdrawal freeze — are invisible from outside until they are not. Judge the structural things you can verify: who holds the assets, under what licence, with what proof, and how it behaves under stress. And treat any score out of ten as a review of the storefront.

    We are going to start with an admission, because it shapes everything that follows. Coinpric does not currently publish exchange reviews with scores, and this article explains the reasoning rather than working around it. The short version is that the most decision-relevant facts about an exchange are not observable from the outside, and a score built only from the observable ones creates false confidence in precisely the wrong direction.

    What a review can genuinely establish

    Plenty is checkable, and worth checking:

    • Published fee schedule — the maker and taker rates, tiering, deposit and withdrawal costs.
    • Asset selection and pairs — what is listed and what it can be traded against.
    • Observable liquidity — visible order-book depth, and the spread at a given size.
    • Interface and reliability — measurable over time, including during volatile sessions.
    • Licensing status — verifiable directly against a regulator’s public register.
    • Stated custody arrangement — who holds assets, in what proportion, per the exchange’s own claims.
    • Documented incident history — outages, halts and losses that became public.

    A review covering these honestly is useful. Note that all of it describes the product, and the risks that ruin people are not product risks.

    What no reviewer can see

    Here is the gap, stated plainly.

    Whether assets exist in the stated amount. A reviewer sees a balance in an interface. That is a database row. Whether the corresponding assets are actually held, unencumbered, is not observable from outside — and historically, when exchanges have failed, this is the thing that was wrong.

    Whether customer assets are segregated. An exchange can hold enough in aggregate while commingling customer funds with its own operations, so a loss elsewhere in the business becomes a customer loss. From the outside, a healthy exchange and one quietly running a deficit look identical.

    Whether affiliated entities are lending the assets. Group structures can be complex enough that customer assets support activity in a related company. This is not visible in an interface and frequently not visible in financial statements either.

    How withdrawals behave in a crisis. Withdrawals work smoothly right up until the moment they do not, and the transition is often a matter of hours. A review conducted in calm conditions tells you nothing about this.

    Concentration and counterparty exposure. Where an exchange’s own risk sits — which lenders, which market makers, which banks — is essentially never disclosed.

    The one partial answer: proof of reserves

    Some exchanges publish cryptographic attestations letting customers verify their balance is included in a total, and demonstrate control of addresses holding that total. This is a real improvement and worth seeking out. But understand exactly what it does and does not cover:

    What it can show What it still cannot show
    Assets are controlled at a moment in time That they are not borrowed for the snapshot
    Your balance is in the claimed total Total liabilities, unless separately proven
    A specific point-in-time position Anything about the following day
    Control of on-chain assets Off-chain obligations, debts and encumbrances

    The core limitation: assets are provable on-chain, liabilities generally are not. A proof of reserves without a corresponding proof of liabilities shows that assets exist, not that they are sufficient. Some schemes address this; many do not, and the difference is rarely explained in the announcement.

    What to actually check yourself

    Verify the licence at the source. Do not accept a logo on a website. Regulators publish searchable registers; find the entity name and confirm it, and check that the licensed entity is the one you will actually contract with. Group structures sometimes route customers to a different company from the licensed one.

    Read the terms for the custody language. You are looking for whether assets are held in trust or as the exchange’s property, what happens to you in insolvency, and whether the exchange may lend or pledge your holdings. This is usually stated, and almost never read.

    Test a withdrawal early and small. Before size matters, confirm the path works end to end. Also confirm the limits, because they are frequently far lower than the deposit limits.

    Look for the specific failure of transparency. Not “is there marketing”, but: is the legal entity named, is a jurisdiction stated, is there an auditor, and are the reserve attestations recurring rather than one-off?

    The conclusion we actually hold

    Exchange risk is not primarily a choice between good and bad venues. It is a structural consequence of holding assets at a venue at all — you are exposed to that company for as long as your assets sit there, and no amount of interface quality changes that.

    Which points at the only mitigation that does not depend on trusting a review: reduce the exposure. Hold at an exchange what you need there for trading and settlement, and hold the rest where the failure of any single company cannot reach it. We cover what that involves, honestly including the new risks it introduces, in what “not your keys, not your coins” really means.

    When we do publish exchange assessments, they will state which claims we verified, how, and on what date — and they will say plainly which risks remain unobservable. Our editorial guidelines and affiliate disclosure set out the commercial position: there are currently no affiliate relationships on this site, and a commercial relationship will never move an assessment.

    Key takeaways

    • Reviews grade fees, interface and liquidity because those are observable. Account-destroying failures are not.
    • Whether assets exist in the stated amount, unencumbered and segregated, cannot be seen from outside.
    • Withdrawals work perfectly until they stop, so a review conducted in calm conditions cannot test the thing that matters.
    • Proof of reserves is a real improvement but proves assets, not sufficiency, unless liabilities are proven too.
    • Verify a licence on the regulator’s own register, and check the licensed entity is the one you contract with.
    • Read the terms for whether assets are held in trust, and whether they may be lent or pledged.
    • Exchange risk comes from holding assets at a venue at all. The only reliable mitigation is holding less there.
  • What a token actually proves about ownership

    What a token actually proves about ownership

    The short version

    A token proves one thing precisely: that a particular ledger entry is controlled by a particular key. It does not prove ownership of whatever the token refers to, because a chain cannot enforce rights over off-chain things. Rights come from a licence or contract, and most tokens are sold without one. The technology is sound; the claim usually made for it is not.

    The interesting question about non-fungible tokens is not whether they are valuable. It is what one actually establishes — because that has a clear answer, and it is narrower than almost all of the surrounding discussion assumes.

    The mechanism, precisely

    A conventional token is fungible: any unit is interchangeable with any other, like a coin. A non-fungible token is one whose units are individually distinguishable, so a contract can track each one separately and record which address controls it.

    What the blockchain establishes, with certainty and without trusting anyone: that a specific token identifier within a specific contract is currently assigned to a specific address, and the complete history of that assignment. That is genuine and useful. It is also the entire extent of it.

    Crucially, the artwork, video or document is almost never on the chain. Storing large files on-chain is prohibitively expensive, so the token holds a pointer — a piece of metadata containing a link. The token is a ledger entry with an address attached to it.

    The three-link chain, and where it breaks

    For “I own this thing” to hold, three links must all hold:

    Link Guaranteed by Strength
    Your key controls the token Cryptography and consensus Very strong
    The token points at the file A link in metadata Only as durable as the link
    Controlling the token grants rights A licence or contract, off-chain Usually absent

    The first link is excellent. The second is where the technical fragility lives. The third is where the conceptual problem lives — and it is much the more important of the two.

    Link two: pointers rot

    If the metadata contains a conventional web address, then the token depends on someone continuing to pay for that server. When it lapses, the token remains perfectly valid and points at nothing. The ledger entry is intact; the reference is dead.

    Content-addressed storage improves this substantially: the pointer is a hash of the content, so any copy can be verified as the right file and the reference cannot silently be swapped for a different image. But it still requires that at least one participant is storing the file. Content addressing guarantees integrity, not availability. If nobody keeps a copy, a verifiable pointer to an absent file is what you have.

    A small number of projects store the artwork fully on-chain, usually by generating it from code. Those genuinely do not have this problem. They are the exception.

    Link three: a chain cannot enforce rights over off-chain things

    This is the one that matters, and it is a limit of law rather than of engineering.

    A blockchain can enforce anything inside its own state — it can guarantee that only your key moves that token. It has no ability whatsoever to affect the world outside. It cannot stop anyone copying an image, cannot prevent the creator selling the same rights elsewhere, and cannot give you standing to object.

    Copyright, in essentially every jurisdiction, sits with the creator unless it is transferred by a written agreement. Buying a physical painting does not give you the right to reproduce it, and buying a token is not different. So what you actually get depends on the licence the seller granted — and the range is enormous:

    • No licence at all. Extremely common. You control a ledger entry. You have no rights over the image.
    • A personal-use licence. You may display it. You may not make commercial use of it.
    • A broad commercial licence. Some projects grant genuinely wide rights, and these are the exception worth noticing.
    • Copyright assignment. Rare, and requires a proper written instrument.

    The token itself carries none of this. Two visually identical tokens from two projects can convey completely different rights, and nothing on-chain distinguishes them. The only way to know is to read the terms — which many collections never published.

    The consequence people find hardest

    Because minting is permissionless, anyone can create a token pointing at anyone’s work. There is no gatekeeper, no verification step, and no rights check. So a token pointing at an image is not evidence that the minter had any relationship to it.

    Platform verification badges are a social layer bolted on top precisely because the technical layer cannot answer the question. When people say a token “proves provenance”, what it proves is the provenance of the token — the chain of custody of the ledger entry — which is a different and much smaller claim than the provenance of the work.

    What this technology is actually good for

    The narrowness is not a criticism. A verifiable, transferable, uniquely identified ledger entry that no issuer can revoke is a genuinely useful primitive, and it works best where the entry is the thing rather than pointing at something else.

    Fully on-chain generative artwork qualifies: the token and the work are the same object. So does anything where the right lives inside the same system — access to an on-chain application, a position in an on-chain protocol, an identifier used by contracts that can read it. In those cases all three links collapse into the first one, which is the strong one.

    Where the token points outward at a physical object, a legal right or a service, the chain is doing a small part of the job and documents are doing the rest. That can work perfectly well. It just is not the chain that makes it work.

    Two questions before you buy anything

    Where is the content stored, and does it survive the project’s own website going away? And what licence am I granted in writing — with the answer being a document you have read, not an assumption from how the collection is described.

    If the answer to the second is “none”, that is a legitimate thing to buy knowingly. It is a bad thing to discover afterwards. Nothing here is advice about whether to buy any of it; see our disclaimer.

    Key takeaways

    • A token proves that a ledger entry is controlled by a key. That claim is strong, and it is the only strong one.
    • The artwork is almost never on-chain — the token holds a pointer, and a conventional web link can simply die.
    • Content addressing guarantees integrity, not availability: a verifiable pointer to a file nobody stores is still nothing.
    • Copyright stays with the creator unless transferred in writing. Buying a token grants no rights by default.
    • Two identical-looking tokens can convey entirely different rights, and nothing on-chain distinguishes them.
    • Minting is permissionless, so a token pointing at an image is no evidence the minter had any right to it.
    • The technology works best where the entry is the thing — fully on-chain art, or rights that live inside the same system.
  • Lending pools, market makers and liquidations, in plain terms

    Lending pools, market makers and liquidations, in plain terms

    The short version

    Most of decentralised finance reduces to three mechanisms. A lending pool lets anyone deposit and anyone borrow against collateral, with a rate set by how much of the pool is in use. An automated market maker holds two assets and quotes a price from their ratio, which means every trade moves the price along a curve. A liquidation sells collateral when it stops covering a debt. None of them require a counterparty who agrees with you — and none of them can pause when things go wrong.

    Decentralised finance is often introduced through its vocabulary, which is the wrong order. The vocabulary is intimidating and the mechanisms are not. There are essentially three, they are built from arithmetic you can follow, and understanding them explains most of what happens when this part of the market breaks.

    Mechanism 1: the lending pool

    Ordinary lending needs a lender, a borrower and terms. A lending pool removes the matching problem: depositors put assets into a shared contract, borrowers take assets out against collateral, and nobody negotiates with anybody.

    The interest rate is set by a formula, not a market maker. The input is utilisation — the share of the pool currently borrowed. Low utilisation means a low rate, because idle capital needs tempting borrowers. High utilisation means a high rate, because the pool needs to attract deposits and discourage further borrowing. Most implementations add a kink: beyond some level, the rate climbs steeply to protect depositors’ ability to withdraw.

    Two consequences follow directly and matter more than any headline yield. First, the rate you see is not a rate you are promised — it recalculates continuously as others deposit and borrow, so a quoted yield is a snapshot. Second, if utilisation approaches its ceiling, depositors cannot all withdraw at once, because the assets are lent out. The high rate is the system asking for help, not a reward for cleverness.

    Borrowing is always over-collateralised: you post more value than you take. There is no credit assessment because there is no identity — the collateral is the entire underwriting.

    Mechanism 2: the automated market maker

    A conventional venue matches buyers with sellers in an order book. An automated market maker holds a reserve of two assets and quotes a price derived from the ratio between them, so it will always trade — no counterparty needed.

    The most common rule keeps the product of the two reserves constant. Take a pool holding 100 of an asset and 200,000 units of a stable asset. The product is 20,000,000, and the implied price is 2,000. Now buy 10 units:

    Step Value
    Reserves before 100 and 200,000
    Constant product 20,000,000
    Implied price before 2,000
    Asset reserve after buying 10 90
    Stable reserve must become 20,000,000 ÷ 90 = 222,222.22
    You therefore pay 22,222.22
    Your average price per unit 2,222.22
    Price impact 11.1%

    You wanted 10% of the pool and paid 11.1% above the starting price. Nothing went wrong; that is the curve. This is why slippage on a small pool is brutal and why pool depth matters far more than any advertised fee. The fee might be 0.3%; the price impact above is more than thirty times that.

    It also explains why arbitrage is structural rather than parasitic. The pool has no idea what the asset is worth — it only knows its own ratio. When the outside price moves, the pool is stale until somebody trades against it, and that somebody takes the difference. Providing liquidity to a pool means systematically selling the asset that is rising and buying the one that is falling. That is the honest description of what is often labelled “impermanent loss”, and there is nothing impermanent about it if prices do not come back.

    Mechanism 3: liquidation

    Every over-collateralised position has a threshold at which the collateral no longer comfortably covers the debt. Cross it, and anyone may repay part of the debt in exchange for a discounted slice of the collateral. The discount is the incentive; that is the whole design.

    This works well in ordinary conditions and is exactly where the system is fragile in bad ones, for three reasons that all arrive together:

    • Price feeds. A contract cannot see market prices; it is told them. During violent moves feeds can lag, and a stale feed either liquidates positions that were fine or fails to liquidate ones that are not.
    • Block space. Liquidations are transactions and compete for the same congested blocks as everything else. A crash is precisely when fees spike, and a liquidation that cannot be included does not happen.
    • Buyers. The discount only helps if someone wants the collateral. In a sharp decline the people who would normally take the other side are managing their own positions.

    When all three bind at once, the protocol ends up holding debt worth more than the collateral backing it. See liquidation for the mechanics, and how forced selling interacts with resting orders for why the cascade accelerates.

    What “decentralised” does and does not remove

    It genuinely removes some things. There is no application to be approved, no minimum balance, no counterparty who can decline you, and the rules are readable code that executes identically for everyone.

    It does not remove risk — it substitutes one set for another:

    Removed Introduced instead
    Counterparty refusing you Contract bugs, which execute perfectly as written
    Discretionary account freezes No recourse, ever, including for your own mistakes
    Opaque internal risk models Dependence on an external price feed
    A firm that can pause trading Nobody who can pause anything mid-crisis
    Credit assessment Mandatory over-collateralisation

    That last row on the left is worth sitting with. The ability to halt is usually described as a flaw of traditional finance, and sometimes it is. It is also a circuit breaker. A system that cannot stop will keep liquidating into a market with no buyers, exactly as instructed, until there is nothing left to liquidate.

    The three questions worth asking of any protocol

    Where do prices come from, and what happens if the feed is wrong or late? What is the collateral, and is it correlated with what has been borrowed against it — because correlated collateral means the crash and the liquidation arrive together? And who can change the rules, how quickly, and with whose approval?

    None of this is a recommendation to use any of it. These are high-risk systems in which total loss is a realistic outcome, including through mechanisms that involve no error on your part. Our disclaimer sets out the full position.

    Key takeaways

    • Lending pool rates are formulas driven by utilisation, so a quoted yield is a snapshot, not a promise.
    • High utilisation means depositors cannot all withdraw — the high rate is a warning, not a reward.
    • A constant-product pool moves price along a curve: buying 10% of a pool cost 11.1% above the start price in the worked example.
    • Price impact frequently dwarfs the trading fee, so pool depth matters far more than the advertised rate.
    • Providing liquidity means systematically selling what rises and buying what falls. “Impermanent” is optimistic.
    • Liquidations depend on price feeds, block space and willing buyers — all three strained at the same moment.
    • Decentralisation removes discretion, including the discretion to stop. Nothing can pause mid-crisis.