Cliff vs Linear Vesting: What Each One Does to Your Order Book

This article was written jointly by FinDaS, a tokenomics design firm, and EchoTrade, a crypto market maker. The choice between releasing an allocation in one block and releasing it continuously is made in a design document, months before anyone sees an order book, and its consequences land in the book. Designers and market makers rarely argue it in the same room.

The design side: the release is one knob

Hristo Piyankov, Lead Token Economist, FinDaS

Take an allocation with a twelve-month cliff. Nothing is released for a year, and then all of it becomes available on the same day. Give it a twelve-month cliff and thirty-six months of vesting instead, and nothing is released for a year, then 1/36 of it each month. Release those thirty-six months as a continuous stream rather than monthly steps, and it becomes a per-block rate. Those are three settings of one knob, and the cliff is identical in all three. What changes is how much becomes sellable on the largest single day, against what the market trades on an average day, and that is the question "cliff or linear" is really asking.

Let's say the supply is a billion tokens, 10% goes to investors, and the token trades at ten cents on two million dollars of daily volume. That is a hundred million tokens locked and twenty million traded a day. The twelve-month cliff on its own puts five days of volume into one morning. With thirty-six months of vesting, each monthly step is fourteen percent of a day. As a daily stream over the same three years, it is half of one percent.

The choice is almost never made on that basis. Cliffs exist for two good reasons, and neither of them is the market. The first is retention: a cliff is the only structure that makes a specific date mean something to the person waiting for it. The second is verifiability. An investor can check a cliff. It is a credible commitment a founder can offer and a lawyer can draft in one clause, whereas "we will behave responsibly with our tokens" is not.

The cliff serves both on its own. Nothing about retention or verifiability needs the whole allocation to arrive the day the cliff ends. That habit usually comes from equity compensation, where a one-year cliff vests a quarter of a grant at once, and it traveled into token vesting more or less unexamined. And the schedule is usually signed before anyone has asked a liquidity question.

What founders are not told is what releasing everything at the cliff costs on the other side. The obvious answer is concentration, and that is only half of it. The larger cost is that it creates a scheduled coordination point. Every holder in that cohort knows the date and gets the same opportunity on the same morning. You have taken a population with different intentions, different cost bases and different time horizons, and handed them a single Schelling point. Linear vesting removes that coordination point without removing one token of supply.

There is a design that keeps the cliff and removes the coordination point, and it is what I usually recommend. A short cliff covering the retention window, typically six to twelve months, then a continuous stream rather than monthly steps. Cohorts staggered so no two land in the same quarter. Smallest cohort unlocking first, because the first unlock is where the market learns how much of a tranche actually gets sold, and you want that lesson to be cheap. The same reasoning applies one step earlier, to how much supply there is to release.

I am not certain about one part of that, so I put it to EchoTrade. Streaming looks better from the design end because it removes the event. My instinct is that it is worse from the desk's end, because a stream never gives the book a quiet window: inventory drifts one way continuously and there is never a date to pre-position against. If that is right, "just stream it" moves cost off the token holder and onto the market maker. EchoTrade disagrees, and the reasoning changed my view.

The shape of a vesting schedule does not change total dilution by a single token. Anyone selling linear vesting as "better tokenomics" is describing a preference about volatility, not a claim about value. If demand for the asset is not there, linear vesting distributes the decline more evenly across more days. That is the whole of what it does, and it is still worth doing, because the avoidable damage is genuinely avoidable. When the problem is demand rather than shape, the fix is on the demand side of the design.

The market side: what each shape does to the book

Mans, CEO, EchoTrade

Hristo's framing is the right one, so I will take it as given: a schedule is a largest-single-day quantity against an average day's volume. From the book, that number decides almost everything, but the two shapes fail in different ways and it is worth being exact about how.

What a cliff does. A cliff is a known date with an unknown quantity. The market can read the schedule; it cannot read the intent of the wallets receiving the tranche. So the days before a cliff are spent pricing a guess. Bids thin as holders reduce exposure ahead of supply, short interest builds where venues allow it, and by the morning of the date the book is already leaning one way. Then the tranche lands and the realized selling is compared against the guess. If less sells than feared, the book recovers quickly. If more, the selling meets bids that were withdrawn days earlier, and the price gaps through the levels that should have been there. The characteristic failure of a cliff is not the size of the supply. It is that the supply arrives on a book that pre-emptied itself.

What a stream does. A stream is the opposite problem: a known quantity with no date. Every day a small, predictable amount becomes sellable, and some fraction of it reaches exchanges. There is no event to price, so there is no pre-positioning, and spread stays close to normal because nobody is charging a premium for event risk. The desk absorbs a rate, works that inventory back out into natural demand across venues, and repeats. What the stream costs is exactly what Hristo identified: the book never gets a quiet window. Inventory drifts one direction continuously, and the desk is always slightly longer the token than it was yesterday.

The question I was asked, and the answer. Is a permanent daily flow easier for a desk than a single date it can prepare for? My instinct is the reverse of Hristo's, and the reason is that predictability beats magnitude. The difficulty of a cliff was never the date. It was that the date is known and the quantity is not, which means the desk has to fund depth for the worst case and hold it idle for the best one. A stream removes the unknown. A known small quantity every day is hedgeable, workable and cheap to carry. A known date with an unknown quantity is a binary event, and insuring against a binary event is expensive. That idle capital is a real cost, and it is part of what a market maker costs on any engagement where a cliff is on the calendar.

So no, a stream does not move cost onto the desk. It moves cost from one line to another: capital held idle for one day becomes inventory worked continuously across many. The second is smaller and productive. Where Hristo's instinct does hold is on the compensation model, and this is the part the article should say plainly. Under a retainer, continuous inventory work is a visible operating cost the project pays for, and the desk has no reason to prefer one shape over the other. Under a token loan with an option, the desk carries that cost itself, and a cliff with a known date is a defined event it can position around. Whether a desk finds "just stream it" expensive depends less on the book than on whether it is paid by retainer or a token loan. We work on a retainer only, so we have no stake in which shape a project picks, and that is what lets me answer this honestly.

What I have seen. Two tokens of similar size, similar float, and first tranches within a few percent of each other as a share of supply. One released it as a single cliff; the other had converted the same cohort to a daily stream a few months before the date. On the cliff, bid depth within 2% of mid fell by more than half in the four days before the date on the largest venue, before a single unlocked token had moved. On the stream, depth on the same measure stayed within its normal range throughout the equivalent period. Same supply, very different books waiting for it. The only difference was whether the market had a date to prepare for.

On coordination versus concentration. Hristo invited the push-back, so here it is, and it is a partial agreement. From the book, concentration is what does the damage: a tranche larger than resting depth gaps the book whatever anyone was thinking. But coordination is what makes the concentration worse than its size, because the pre-positioning happens on the bid side, before the supply arrives. A cliff does not just deliver a large quantity. It delivers a large quantity onto a book that the same information has already hollowed out. Concentration sets the size of the test. Coordination shrinks the book that has to pass it.

The mirror limit. Can a desk make a cliff behave like a stream? Partly, and only with cooperation. What a desk can do is fund depth in advance so the tranche is absorbed by resting orders rather than gapping through empty levels, hold spread and cross-venue consistency through the flow, and, where a large holder is willing, work their exit over days rather than watch it hit the book at once, which turns that holder's cliff into a de facto stream. Where it stops is exactly where Hristo's limit stops: a desk cannot change when supply becomes sellable, cannot make a holder who wants out at market wait, and cannot absorb a tranche larger than the capital the project funded into the book. It changes how supply meets the market, never how much of it there is. That is the whole of what market making can do here, and it is still worth doing, for the same reason Hristo gives: the avoidable part is genuinely avoidable.

When a cliff without vesting is right anyway

A cliff without vesting is a mistake as a default, which is what it usually is, and the right instrument in four situations.

Retention is not on the list. The cliff does the retaining, whichever release follows it.

So which should I pick?

From the design end. Keep the cliff and stream what comes after it, unless the release is one of the four cases in the table above. Anything else released in one go is released that way by inheritance rather than by decision. Where releasing everything at the cliff is unavoidable, make that allocation small and keep it clear of the other cohorts' dates.

From the book. Stream it, and for a reason the design end does not see. A stream is cheaper for a desk to carry, because a known small quantity every day is hedgeable and a known date with an unknown quantity is a binary event. The exception is the compensation model: a desk paid in a token loan with an option has an interest in the event, and a desk on a retainer does not. Ask which one you are buying before you ask what it thinks of your schedule.

FAQ

What is the difference between cliff and linear vesting?

An allocation with a twelve-month cliff releases nothing for a year and then all of it at once. One with a twelve-month cliff and thirty-six months of linear vesting releases nothing for a year and then 1/36 of the allocation each month, and most allocations are built that way. What matters for the market is the difference on the largest single day, measured against average daily traded volume.

Is linear vesting better than a cliff?

For the order book, usually yes. A cliff on its own gives every holder in a cohort the same opportunity on the same morning, and linear vesting removes that coordination point. For retention, the cliff does most of the work either way, and with the same cliff in front, vesting keeps part of the allocation binding for longer. Most schedules should be a short cliff followed by a stream.

What is the right cliff length for a team allocation?

Six to twelve months covers the retention window in most cases, followed by a continuous release rather than monthly steps. The length matters less than two other things: that the team cliff does not end on the same date as the investor cliff, and that whatever unlocks first is small enough that the market learns cheaply how much of a tranche actually gets sold.

My schedule is monthly. Is that a stream?

Not necessarily. A monthly release is a one-day event repeated every month, and every date is on the public calendars. Divide one month's release by average daily volume. Well under a day and the distinction is academic; approaching a full day and you have twelve scheduled supply events a year rather than a stream.

Can a vesting schedule be changed after TGE?

Contractually, rarely, and not unilaterally. What can still change without reopening the contract is which cohort unlocks first, whether cohorts can be staggered by consent, whether a large holder will take an OTC exit with a lock instead of selling into the book, and whether the desk has the calendar. Re-launching a token is the harder route and the one to avoid needing.

Does a market maker prefer cliff or linear vesting?

A retainer desk has no structural preference, and it should say so. A cliff is a capital problem: depth has to be funded for the worst case and sits idle for the best one. A stream is an inventory problem: a small known quantity absorbed every day and worked back into natural demand. The second is cheaper to carry and easier to hedge. A desk paid by token loan may prefer a cliff, because a known date is an event it can position around, which is a preference about the desk's own economics rather than about the token.

If my schedule is already a cliff, what can the market maker actually do about it?

Fund depth ahead of the date so the tranche meets resting orders rather than empty levels, hold spread and keep venues consistent through the flow, and where large holders are willing, execute their exits over days instead of at once. What it cannot do is change when the supply becomes sellable or absorb more than the capital the project has put into the book. The preparation has to start weeks out, and the vesting calendar has to be with the desk from onboarding, not the week before.

For the market maker's side of the first unlock after TGE, read The First Unlock After TGE: What Happens to Your Book on EchoTrade's blog.