How Veld works

A short, plain-English summary of the rules every Veld node enforces. No hidden fees, no off-protocol math. If you want the full technical derivations, formulas, and edge-case behaviour, those live in the whitepaper.

1. Block reward & emission

The network produces a new block roughly every 60 seconds and mints a fixed reward of about 1.9 VELD. That works out to ~2,739 VELD per day across the network. Total supply is capped at 21,000,000 VELD — once that's mined, no more is ever created.

2. Where each block reward goes

Every block reward is split four ways. Every node enforces this split — a miner who tries to keep more than 50% has their block rejected.

Transaction fees from a block flow to the vault on top of the 20% block share.

3. How vault distributions work

Every 144 blocks (~2.4 hours) the vault pays out a portion of its inflow during that cycle to every active staker. Your share is proportional to your stake × your multiplier, normalised against the rest of the active stakers.

Higher multipliers get a bigger slice; everyone else still gets paid. The full payout math is in the whitepaper.

4. Vault drain protection — vault never drains, always grows

The vault is the network's long-term staking reserve. A consensus-enforced rule guarantees the vault holds a positive growth rate every cycle, even at scale:

• Inflow cap: at most 90% of the vault inflow during a cycle (~144 blocks of coinbase splits + transaction fees + every-100th-block bonuses) is paid to stakers. The remaining 10% stays as principal — the vault grows by that 10% every cycle, forever.
• Backstop ceiling: never more than 8% of the vault balance distributes in one cycle. (Defence-in-depth; the inflow cap binds first in normal operation.)
• Pool never freezes: distributions fire whenever there's at least one active staker.

These limits are a hard guarantee enforced at consensus time. The multiplier system can only redistribute the cycle budget between stakers; it can never enlarge it. Because at most 90% of inflow leaves per cycle, the vault grows monotonically — an early-staker bonanza is impossible by design, and the post-emission-era fee economy inherits a substantial principal that funds stakers indefinitely.

5. Concentration limit

No single staker can receive more than 75% of any one distribution cycle, no matter how big their stake or multiplier is. If their share would exceed 75%, the excess stays in the vault for the next cycle — it is never redistributed back to other stakers.

At realistic network scale (many active stakers) the limit is dormant: every staker's share is well under 75% and the full cycle distributes. The 75% number is set so a max-lockup staker still earns the full multiplier they were promised; the math is in the whitepaper.

6. How the staking multiplier is built

Your staking multiplier is the value that boosts your share of every vault distribution. It comes from two ladders:

• Mining tier — rewards you for being an active miner over time (see §8).
• Lockup tier — rewards you for committing your stake longer (see §7).

The two multiply together. The result is capped at 3.00×, and that ceiling is reachable only at Diamond mining tier — even max Platinum + a 90-day lock tops out at 2.70×. Doing both pays the most.

7. Stake lockup tiers

When you stake, you pick how long to lock it. Longer commitment = bigger multiplier. You can't unstake before the lockup expires.

You can hold multiple stakes at the same address with different lockups. Each one unlocks on its own schedule — adding a new stake doesn't reset older ones.

8. Mining tier ladder

Mining tiers reward consistent mining over time. A day counts as active for your address if you mined at least one block during it — it doesn't matter whether you found 1 or 100. Your tier is recomputed every block from a rolling window of recent days.

There's no permanent tier. Mine consistently to earn it; if you stop, the tier drops as old active days roll out of the window. The full rolling-window logic is in the whitepaper.

9. Validators & the validator pool

Validators run a daemon that signs each new block — an endorsement. The validator pool collects 10% of every block's reward and pays it out every 144 blocks (~2.4h) in proportion to each validator's endorsement count over the trailing 144-block window. Idle validators earn nothing for that cycle.

• System unlock. The validator system stays locked until at least 50,000 VELD is staked network-wide. Until then no endorsements are accepted and the pool simply accumulates.
• Register on-chain. You submit a VELD_VALIDATOR|REGISTER transaction binding your ML-DSA-65 public key. A matching DEREGISTER exits cleanly.
• Bond the minimum. Registration requires posting the minimum validator bond (see §10). The live bring-up threshold is small for testing; at the 2026-08-30 mainnet launch it rises to 10,000 VELD. The live value is shown on the Validators page.
• Stay online. Endorsement share is your count ÷ the network's count over the last 144 blocks — consistent uptime is what pays.

Endorsement is one of three independent reward streams: the co-mining lottery is a flat draw (§11), the vault pays every staker (§3), and only the validator pool is endorsement-gated. They do not overlap and you can earn from more than one.

10. Validator bond, slashing & yield escrow LIVE

A validator's registration bond is not just a balance check — it is real, slashable capital. This whole system is active on-chain today (a single hard fork enabled custody, slashing, and the reporter bounty together).

The bond (custody)

At registration the minimum bond is sent into a sigless custody vault — an address whose key nobody holds, so the principal cannot be moved by anyone. It sits there as collateral for as long as you validate and is returned to you in full on a clean deregister, at the next 144-block settlement boundary.

Slashing — equivocation is permanent

Signing two different block hashes at the same height (equivocation / double-sign) is cryptographically provable. Anyone can submit that proof. A verified slash is permanent: the offending public key is banned and can never re-register, and the custodial bond is confiscated and split deterministically:

Validators that registered before the bond hard fork are grandfathered: they hold a logical lock rather than movable custodial principal, so they face the permanent ban + stake-lock extension but no on-chain confiscation. New registrations are fully custodial.

Yield escrow (D′)

Idle collateral is wasteful, so the bonded principal also earns vault yield at the top lockup tier. That yield is not paid out immediately — it is escrowed and vests on a rolling delay:

• Vested yield is released to the validator on the normal settlement schedule.
• If the validator is slashed, the still-unvested tail is clawed back along with the bond.
• Already-vested income is kept — behave for years then slip once and you only forfeit the recent unvested tail around the offence, not your whole earning history.

Live custody-vault and yield-escrow balances, per-validator bond status (custodial / grandfathered / slashed), and the slashing-evidence ledger are all on the Validators page. Full vesting schedule and confiscation derivation are in the whitepaper.

11. Co-mining lottery

Mining doesn't have to win the block to pay you. The 20% co-mining pool collects on every block and pays out every 100 blocks through a simple uniform lottery:

• Mine the chain. Your node automatically submits "near-misses" — hashes that came close to the target but didn't quite win the block.
• Have the minimum stake locked at your mining address. (Same stake as for vault distributions, no double-lock.)
• Submit at least one valid near-miss in the 100-block window — that's it.

At the end of each window the pool is split equally among a fixed number of randomly-chosen eligible miners. The draw is flat: every eligible miner has exactly the same chance, no matter how many near-misses they submitted — one near-miss qualifies you the same as a thousand. No carry-over between windows; each cycle starts fresh.

Why flat? A weighted lottery would let a hashpower-heavy farm sweep the pool by pumping near-miss volume. An equal-chance draw caps any single miner's expected take per cycle and pushes payouts toward the long tail of small, independent miners — which is the point of the co-mining stream. The exact winner count, draw-randomness derivation, and consensus enforcement are in the whitepaper.

12. Supply & emission

• Hard cap: 21,000,000 VELD — ever.
• Block reward: ~1.9 VELD per block, flat — no halving.
• Issuance era: ~21 years to mine the full 21M.
• After the cap: block subsidy ends; miners + validators keep earning indefinitely from transaction fees.

Over the issuance era, miners (block-winner + co-mining pool) collectively earn ~70% of total supply, the vault accumulates ~20% for stakers, and the validator pool accumulates ~10% for endorsers. Post-cap fee-split details are in the whitepaper.

13. Privacy & safety

• Your private key never leaves your machine. The wallet signs every transaction locally; only the signed bytes go on the wire.
• No deep rewrites. Hard-coded checkpoints plus a 100-block reorg cap mean nobody can rewrite chain history older than ~100 minutes.
• Vault drain protection. The 90% inflow cap (vault retains ≥10% of every cycle's inflow as principal), 8% balance backstop, and 75% per-staker concentration limit collectively make vault drain impossible — the reserve grows monotonically.
• Coinbase maturity. Newly-mined VELD (including co-mining and validator-pool payouts) is spendable only after 100 confirmations. Stops reorg-and-spend attacks.
• Post-quantum signatures. Every transaction is signed with ML-DSA-65 (NIST FIPS 204), so a future quantum computer doesn't invalidate a single UTXO.

14. Reward promos

Validator governance can activate a temporary reward boost for all miners and stakers. The bonus comes from the vault (no new VELD is printed, so the 21M cap stays intact). Strict caps prevent abuse:

15. Difficulty retargeting

The chain aims at a 60-second block time. Mining difficulty re-tunes itself once every 144 blocks based on how fast the previous window of blocks was found. If miners are finding blocks too fast, difficulty rises; too slow, it falls.

Difficulty can change by at most 1.5× or 0.5× per retarget, so a hashrate spike or drop can never push the chain off balance suddenly. If the network briefly stalls, mining waits for the next 144-block retarget boundary — LWMA reads the previous window's timestamps and re-targets to the current hashrate without any special-case eased-bits rule.

The retarget algorithm is LWMA-1 (Linear Weighted Moving Average). Full math, clamp rationale, and consensus-enforcement details are in the whitepaper.