Vesting schedule

The three parameters that define any schedule

The cliff is the initial period in which nothing releases at all. The duration is the window over which locked supply distributes once the cliff has passed. The cadence is how often a release actually happens inside that window. Set those three per bucket and you have described the shape of every unlock curve in the design.

Monthly cadence is the working default. Arbitrum's team, contributor and investor tokens unlock over four years starting 16 March 2023, with the first unlock on 16 March 2024 and monthly unlocks after that, which is a plain cliff-plus-monthly-linear schedule stated in the project's own documentation.¹ Daily or per-second release is available onchain and gets used, mostly through streaming contracts rather than batch transfers.

A fourth parameter belongs in the same table even though it sits outside the schedule proper: the launch-day unlock. A bucket with a 12 month cliff and a 5 percent release at the token generation event is not a locked bucket, and treating it as one is how a launch-day float calculation goes wrong.

Linear, milestone and hybrid, and what each one buys

Linear vesting releases equal value per unit of time after the cliff. It is what almost every real schedule uses, and there is a mechanical reason for that beyond preference. OpenZeppelin's VestingWallet, the audited reference contract most vesting deployments are built on or modeled after, implements strictly linear time-based release with no native milestone logic at all.⁴ The default is linear because linear is what the standard primitive does.

Milestone vesting ties releases to delivery instead of the calendar: mainnet ship, an audit passing, a usage threshold. It aligns better with what everyone claims to want, and it is rare, because someone has to attest onchain that the milestone was met. That attestation is an oracle with a human behind it, and now your vesting contract has a trusted party and a dispute surface. Teams that want milestone vesting usually get it through a custom contract or a lockup platform supporting per-beneficiary schedules.⁵

Hybrid schedules are the practical compromise. A calendar vest runs underneath, with a portion of each bucket gated behind delivery, or a milestone accelerates a vest that would otherwise run its full term. In our experience the hybrid version survives contact with lawyers more often than pure milestone vesting does, because the downside case is still a schedule everyone can compute.

How a schedule is actually enforced onchain

Three enforcement models are in use, and they are not equivalent. The first is a vesting contract holding the tokens, with the beneficiary calling a release function to pull whatever has accrued. VestingWallet works this way: it stores a beneficiary, a start timestamp and a duration, computes the vested amount as a proportion of elapsed time, and exposes release for the beneficiary to claim.⁴ Anyone can read the contract state and verify the terms without asking the team.

The second is a streaming protocol. Sablier creates a stream with a start and end timestamp and a deposited amount, and the claimable balance accrues continuously per second rather than in discrete monthly steps.⁵ The economics are the same shape. The claimable balance on any given day is not.

The third is a multisig making batch transfers on a stated schedule. This is the weakest form. The schedule is a promise rather than a constraint, verification requires trusting the team's reporting, and the signers can move early. If a project's vesting is enforced this way, the schedule in the deck is documentation of intent rather than a mechanism.

Which model you pick is a disclosure decision as much as a technical one. A contract address that anyone can read is a credibility asset during a raise, and it costs nothing to have on day one.

Arbitrum's schedule, and the day it came due

Work the arithmetic on a real one. Arbitrum's team, contributor and investor allocations vest over four years from 16 March 2023, with nothing released until the first unlock exactly twelve months later and monthly unlocks after that.¹ A four year vest behind a one year cliff means that at the moment the cliff breaks, one quarter of those buckets is claimable in a single event rather than a monthly slice.

That is what happened. On 16 March 2024 Arbitrum cliff-unlocked 1,111,750,000 ARB, worth about $1.24 billion at the prevailing $1.12, an amount equal to 87 percent of the token's then-circulating supply of 1.275 billion.⁶ One date, one contract, and the tradable supply of the token nearly doubled.

None of that was hidden. The terms were published a year in advance, the date sat in every unlock calendar, and the size was arithmetic anyone could do. The lesson is not that the schedule was badly designed. It is that a one year cliff on a four year vest concentrates twelve months of accrual into a single day, and that concentration is a design choice rather than an accident.

The same shape recurs across the cohort. Optimism locked core contributor and investor tokens for one year before three years of linear vesting.³ Sui blocked initial investors from transferring any of their stake until its one year cliff ended in May 2024.² Three projects, three different chains, one convention nobody has published a dataset to justify.

The alternative shapes are available and cost nothing to specify. A shorter cliff with a longer tail. A partial release at the cliff with the remainder resuming monthly. Cliff dates staggered across buckets by a quarter. Each spreads the same total supply across more days.

What unlock pressure actually does

An unlock does not sell anything. It changes what is capable of being sold, which the market prices in advance because the date and the size are both public. This is why the observable price effect frequently lands before the unlock rather than on it, and why a team announcing that recipients have committed to hold rarely moves anything: the capacity is what gets priced, not the stated intention.

Two schedule properties drive how hard that pricing bites. Size relative to circulating supply, not relative to total supply, because circulating supply is what sets the depth available to absorb it. And whether other buckets release inside the same window, since the market sees one aggregate number arriving in a month and not four separate governance decisions.

We do not publish a percentage for how much circulating supply a given unlock adds on average, because no cross-project dataset supports one. The named, dated events are what can be cited, and they vary enormously. Arbitrum's cliff event at 87 percent of circulating supply is documented.⁶ It is one event, not a benchmark.

Modeling the buckets together instead of one at a time

A team bucket with an 18 month cliff and a 36 month vest looks unremarkable alone. Put an investor bucket with a 6 month cliff and an 18 month vest beside it and the investor curve peaks well before team vesting has started to bite. Add a liquidity bootstrap tranche also releasing at month six and two buckets now peak together, in a month where nobody planned for them to.

That compounded release is the thing a single-bucket view structurally cannot catch, and it has a name once several cliffs land inside the same window. The fix is procedural rather than clever: build the combined month-by-month model before any single bucket's terms are finalized, then negotiate against the model.

The order that produces failures is the common one. Each bucket gets negotiated with its counterparty on its own merits, the tables get combined afterwards, and three or four months turn out to carry several large releases at once. By then every one of those terms is in a signed agreement.

What we settle before any bucket is negotiated

Four things, in order. Total supply and launch-day float, because the schedule's job is defined relative to the float it is protecting. The enforcement model, since a contract address is a different negotiating position from a promise. The combined calendar, with every cliff end date and every bucket's monthly release laid out to month 48. And the specific months where releases overlap, with a written reason for each overlap that survives.

Then, and only then, the per-bucket terms. Across 100+ projects and $100MM+ in combined raises, the schedules that hold up are rarely the sophisticated ones. They are the ones where somebody built the aggregate model first and had a number to point at when an investor asked for a shorter cliff.

A vesting schedule manages the supply side of a market. It cannot create demand, and it cannot compensate for a business with no revenue path underneath it. The token is infrastructure. The business is the engine. Nothing on this page is a recommendation to buy, sell or hold any token.