Validator Hierarchy: Methodology FAQ
How the data was pulled, why we rank on realized return rather than take rate, and how every figure in the report can be reproduced from public on-chain data.
Where do these numbers come from?
Every figure comes from the Taostats public API endpoint /api/dtao/validator/yield/history/v1, which reports daily nominator yield net of validator take. We pulled it for 25 validators across all 129 subnets including root, for 365 days ending June 12, 2026: 860,642 daily snapshots in total. Anyone with a Taostats API key can pull the same data. The full dataset is retained and available for verification.
How did you pick these 25 validators?
They are the 25 largest commercial root validators by global nominator count, filtered to operators active on at least 50 subnets. That filter deliberately excludes subnet-primary validators such as Chutes and Macrocosmos, which validate a single subnet and do not compete for general nominators. The 25 we included hold the overwhelming majority of all staked root TAO, so this is not a sample, it is close to the whole relevant population.
Why rank on realized return instead of advertised take rate?
Because take rate is what a validator charges, and realized return is what a nominator actually earns. The entire point of the report is that these are different numbers. Take is one input; validator performance (weight-setting, uptime, the share of emissions the network awards) is another. Ranking on realized return captures both, which is what matters to your stake.
Did you cherry-pick a 90-day window that flatters or punishes certain validators?
We published six windows: 7, 30, 60, 90, 180, and 365 days, and we show every validator's rank in all of them. The headline uses 90 days. The extremes are stable across all six windows: the top two are first or second everywhere, the bottom two are 22nd to 25th everywhere. If a validator looked bad at 90 days but good at 180 and 365, that would be noise and we would say so. For the top and bottom tiers, it is signal.
My validator charges a low take but ranks poorly. Isn't that unfair?
It is exactly the finding. A low take does not guarantee a high return. Cortex Foundation charges 0% and still finishes 18th of 24, because its gross performance is lower. We publish a decomposition that splits each validator's shortfall into the part caused by its fee and the part caused by earning less gross yield, so you can see precisely where the gap comes from. For several poorly-ranked validators, the fee is a minor factor next to performance.
Are you saying validator X is bad, or dishonest?
Neither. We are reporting that nominators staking with a given validator earned a measurable amount less than they would have with the best in class, over a defined window, from public data. That is a measurement, not an accusation. Validators may offer value beyond yield (insurance, custom infrastructure, integrations) that justifies a return gap for some nominators. We make no judgment on those. We report the realized financial outcome and nothing more.
Why root only? What about subnet staking?
Root staking is the cleanest possible comparison: every commercial validator validates root identically, so the only variable is the operator. Subnet staking is dominated by the alpha token's price movement rather than validator yield, which our first report (The Yield Illusion) covered in detail. Validator choice matters less on subnets because price swamps it. We did pull subnet-level data for all 25 validators and the broad picture is consistent with the root ranking, but root is where validator selection is the whole game, so that is what we lead with. A cross-subnet breakdown is reserved for a follow-up.
How is the cost-per-stake figure calculated?
It is the gap in annualized percentage points between a validator and the best in class, applied to a given stake size for one year. If the best validator returns 8.86% and another returns 4.45%, the gap is 4.41 points, which on $10,000 is $441 per year. Because the APY figures are already annualized, this is a direct multiplication. It does not assume compounding, so if anything it understates the multi-year cost.
Why didn't you contact validators before publishing?
Research is published cold. Selective pre-notification would create an unfair information advantage and compromise the independence of the work. Validators are welcome to respond publicly with corrections or additional context, and we will engage substantively with anything that materially affects the data or methodology. The full dataset is available so any disagreement can be settled with numbers.
Will you update this?
Yes. We intend to refresh the Validator Hierarchy quarterly. A recurring measurement also reveals which validators improve in response to being measured, which is part of the point. If publishing realized returns pushes underperforming validators to set better weights or lower their take, the report has done its job.
Still have a question?
If something here is unclear or you believe a figure is wrong, we want to know. The full dataset and analysis code are retained for verification. Reach us in the FlowSniper Discord.