Validator Net Yield Comparison

Net yield is the actual APR a staker earns on a given validator, computed as `gross_apr × uptime_fraction`. Gross APR is what the chain's reward formula would pay if the validator never missed a slot; uptime is the fraction of slots the validator actually voted/produced in the measurement window. For Solana the bench uses Stakewiz's `total_apy` for gross (which already includes Jito MEV tips), multiplied by the validator's reported uptime. For Hyperliquid it uses `predictedApr.day` × `stats[day].uptimeFraction` from `/info validatorSummaries`. The unit is basis points (1% = 100 bps) to stay consistent with the rest of OpenChainBench.

Two reasons. (1) ~42 of the top 200 tracked Solana validators are currently emitting 0% APY, some have commission set to 100% (operator takes everything, delegator gets 0), others have zero leader slots in the current epoch (no block-production rewards earned during the measurement window). These are legitimate, transient states that count toward the chain's median by definition. (2) Average validator commission on the tracked set is ~24%, which already eats into the headline 'staking APY' number quoted by chain marketing. If you want 'yield I'd get if I picked a sane validator', look at the per-validator gauges or wait for the leaderboard to expose p75+, those numbers cluster around 5.9-6.0% net.

Yes, the direction is what it looks like: Solana ~5.6% > Hyperliquid ~2.2%. Three reasons. (1) Hyperliquid pays staking rewards in HYPE only, there is no inflation analogue to Solana's ~5% protocol issuance. (2) No MEV layer exists at the validator level (the centralised sequencer captures order-flow value at protocol level, not validator level), so there's no MEV-tip surcharge to add. (3) The reward pool is split across ~30 validators evenly when active, so per-validator APR is structurally flat, top operators all sit at ~225 bps with very tight spread. A staker chooses a Hyperliquid validator on uptime / jail risk / governance alignment, not on yield differentiation.

MEV is INCLUDED in the gross APR figure, not subtracted from it. For Solana, Stakewiz's `total_apy` already folds Jito MEV tips into the headline APR by design, this bench inherits that aggregation rather than re-computing. The `ocb_validator_mev_share_bps` gauge is exposed separately for transparency (so you can see e.g. 'Helius's 600 bps APR is 80 bps MEV + 520 bps staking') but it does NOT get subtracted to produce a 'staking-only' number. For Hyperliquid, no separate MEV layer exists at validator level (the centralised sequencer captures value upstream), so the MEV gauge is reported as 0 by convention. This is the same `net_yield` definition every staking dashboard uses, Stakewiz, Solana Beach, Hyperliquid Stats, etc.

Two legitimate causes, both visible in the live data. (1) Commission set to 100%: the validator operator has configured their vote account to route 100% of staking rewards to themselves, so delegators receive 0%. This is common for operators running their own treasury stake (Binance, exchange custody, etc.) where 'delegators' is a fiction, there's only the operator's own SOL. (2) Zero leader slots in the current epoch: Solana's leader schedule is stake-weighted but randomised per epoch, so a small validator can statistically draw zero slots in a given ~2-day epoch and earn zero block-production rewards. Both states are transient: commission can be lowered, leader slots cycle every epoch. The bench surfaces them as 0 rather than filtering because they ARE part of the live validator set.

Ethereum is deferred to v2. Honest validator-level APR on Ethereum requires either (a) running a beacon-node connection (~1 TB disk, ongoing sync cost) to read per-validator effective balance and attestation history, or (b) scraping every MEV-Boost relay's API for proposer-level payment data (Flashbots, BloXroute, Eden, Aestus, Manifold, etc., each with its own schema). Neither is a 5-minute HTTP poll like Stakewiz or Hyperliquid `/info`. The v1 scope was 'free no-key APIs that expose the full active validator set in one call'; Ethereum doesn't have that. v2 will add beacon-node ingestion or relay aggregation once the harness footprint can afford it.

Stakewiz (`api.stakewiz.com`) is a community-maintained Solana validator analytics service that ingests on-chain vote-account state, commission history, leader schedule and reward distribution, then publishes a clean `/validators` JSON with `total_apy` already computed as a 30-epoch trailing estimate. Jito Kobe (`kobe.mainnet.jito.network`) is Jito Labs' public API exposing per-validator MEV-tip earnings broken down by epoch. Both are free, key-less, and well-known in the Solana validator-ops community, this bench treats them as the canonical source for Solana validator APR the same way most Solana staking dashboards do (Solana Beach, Marinade UI, etc., all consume one or both).

Prometheus cardinality budget. The full Solana validator set is ~3000; tracking all of them across `net_yield`, `gross_yield`, `mev_share`, `commission`, `uptime`, `stake_usd`, `jailed` would land ~21k label combinations on the OCB Prom, vs ~1.6k for the top 200. The top 200 by activated stake represent ~90% of staked SOL by USD value, so the chain median is representative of where real delegated stake sits, not skewed by the long tail of <1k-SOL validators that have negligible delegate base. The cap is a knob, if Prom capacity grows, it lifts to 500 or 1000.