Introducing validator retainment and compound rates
Polkadot nominators tend to think in years. They bond DOT, select validators and accept the trade-offs of staking because they believe in the network over the long term.
Validators have a different set of day-to-day decisions. They earn DOT through staking and validator incentive programmes, then decide whether to compound it, hold it, move it elsewhere or sell it. Selling rewards can be entirely legitimate: operators have infrastructure, staff and tax costs. But when a validator consistently sells a large share of its rewards while its nominators continue to hold, the two groups may no longer have the same economic exposure.
Until now, that difference has been difficult for a nominator to see at a glance. A validator's rewards and balances are public, but turning many accounts, identities, eras and transactions into useful retainment signals is not a trivial task.
That is what validator retainment statistics are designed to do. They give nominators a consistent way to compare how rewards circulate through an operator's identity graph and how much of them are added back to self-stake.
Retainment estimates how much of an operator's validator rewards remained within its publicly linked on-chain identity over a monthly period. A related compound rate estimates how much of one validator's own rewards appeared as additional self-stake. Together, these metrics give nominators a consistent way to compare how rewards circulate through an operator's identity graph and how much of them are added back to self-stake.
These metrics help nominators ask a question that was previously hard to answer: does this validator appear to be keeping meaningful economic exposure to the DOT it earns?
Three related measurements
The feature separates three ideas that are easy to conflate.
| Measurement | What it asks | Scope |
|---|---|---|
| Retainment rate | How much of the rewards earned by this identity graph is reflected in its higher total DOT balance? | The validator's complete People Chain identity graph |
| Compound rate | How much of this validator's own rewards is reflected in higher validator self-stake? | One validator stash |
Retainment is deliberately broader than compounding. DOT does not have to be bonded as self-stake to be retained: it may remain liquid or reserved in any account connected to the same identity. Compounding is narrower, and focuses specifically on increased self-stake.
Start with reproducible monthly snapshots
The calculation begins with two historical snapshots, normally taken at consecutive UTC month boundaries.
For each month, the snapshot process finds the first canonical Asset Hub block at or after 00:00 UTC on the first day. That block supplies the authoritative timestamp, active era, registered validator set and account balances. The process then selects the latest indexed People Chain block at or before that exact Asset Hub time. The People Chain view is therefore never allowed to run ahead of the balance view.
The Asset Hub and People Chain block numbers are stored with the snapshot. This makes the result reproducible: it identifies the precise states used rather than relying on an approximate wall-clock date.
Identity graphs
A stash address rarely tells the whole story of a validator operation.
On People Chain, an account may have a super-identity, sub-identities below it and further sub-identities below those. Several validator stashes may belong to the same root identity. Other linked accounts may not validate at all but can still hold the operator's DOT.
For each registered validator, the snapshot process follows the super-identity relationship up to its root, then recursively expands every sub-identity beneath that root. The resulting graph includes validator and non-validator accounts. Every account is counted once, and its free and reserved balances are added to the graph total.
If rewards leave Validator A's stash but arrive in the publicly linked treasury account, a stash-only calculation would say the rewards disappeared. The graph calculation correctly recognises that the DOT remained within the same on-chain identity.
When multiple validator stashes share a graph, their validator rewards are combined for the retainment calculation. As a result, validators in a continuous shared graph receive the same graph-level retainment rate, while each still receives its own compound rate.
Identity relationships can change between snapshots. If the same graph root persists and still contains the validator, the full graph is used. If that continuity breaks, it falls back to the validator's own account balance at both boundaries. If no reliable opening balance exists, the interval is omitted instead of producing a misleading number.
Deriving the retainment rate
Once two consecutive snapshots exist, the graph's net balance movement is measured.
Net inflow = closing graph balance − opening graph balance
It also totals the ordinary validator rewards and validator incentive rewards earned by every validator in the closing graph during the interval. The opening era is excluded and the closing era is included, preventing the boundary reward from being counted twice across adjacent months.
The retained amount is the positive graph balance movement, capped at the graph's rewards:
Retained rewards = the smaller of positive net inflow and graph rewards
Retainment rate = retained rewards ÷ graph rewards × 100
The lower bound matters because a falling graph balance cannot represent retained rewards. The upper bound matters because a graph can receive DOT from somewhere other than validator rewards. External deposits may increase the balance, but they cannot push retainment above 100%.
Consider a graph that starts a month with 10,000 DOT, earns 800 DOT in validator rewards and ends with 10,600 DOT:
| Item | Amount |
|---|---|
| Opening graph balance | 10,000 DOT |
| Closing graph balance | 10,600 DOT |
| Net inflow | 600 DOT |
| Graph rewards | 800 DOT |
| Retained rewards | 600 DOT |
| Retainment rate | 75% |
The graph is 600 DOT larger after earning 800 DOT, so the measurement attributes at most 600 DOT of the rewards to retainment.
Now suppose the graph ended at 11,000 DOT. Its 1,000 DOT increase is greater than its 800 DOT of rewards. The retained amount would be capped at 800 DOT and the rate at 100%, because the extra 200 DOT may have come from another source.
If a graph earned no validator rewards during the interval, its retainment rate is unavailable rather than zero. There is no meaningful reward denominator for that period.
Deriving the compound rate
The compound calculation asks a narrower question: did a particular validator increase its own stake?
For each validator, its first and last recorded self-stake across the monthly era interval is compared. It then reconciles that change against only that validator's own ordinary and incentive rewards:
Self-stake change = ending self-stake − starting self-stake
Compounded rewards = the smaller of positive self-stake change and validator rewards
Compound rate = compounded rewards ÷ validator rewards × 100
If a validator earned 300 DOT and its self-stake rose by 120 DOT, its compound rate is 40%. If its self-stake rose by 500 DOT, the compounded amount is capped at the 300 DOT of rewards and the rate is 100%. If self-stake fell, the compounded amount is zero.
As with retainment, a period with no validator rewards has no meaningful compound rate and is reported as unavailable.
Reading retainment and compounding together
The two rates are most useful side by side.
| Retainment | Compounding | One possible interpretation |
|---|---|---|
| High | High | Rewards largely remain in the identity graph, with a meaningful share added to self-stake |
| High | Low | DOT remains in linked accounts but is mostly liquid, reserved or otherwise not added to validator self-stake |
| Low | High | The validator added to its own stake, while other balance movement across the wider graph reduced net retainment |
| Low | Low | The graph did not grow in line with rewards and little appeared as additional self-stake |
These are prompts for investigation, not claims about intent. In particular, a low retainment rate does not prove that DOT was sold. Rewards may have funded operations, taxes or community activity; they may have moved to an account that is not linked on People Chain; and other withdrawals from the graph may offset retained rewards. Conversely, a high rate can coincide with unrelated deposits, although the 100% cap limits their effect on the reported result.
The measurement tracks changes in balances, not the lineage of individual DOT. It also observes month boundaries, so transfers that happen and reverse entirely between snapshots are not visible. The identity graph is only as complete as the public on-chain relationships at the selected People Chain block, and self-stake changes can include funds added or removed for reasons unrelated to rewards.
That context is why the underlying period, reward totals, net inflow, self-stake change, identity graph and source block numbers are exposed alongside the rates. A transparent metric should be inspectable, not merely asserted.
What this gives nominators
Retainment and compound rate metrics add a new alignment lens to validator selection.
A nominator can now see whether an operator's publicly linked DOT balance has tended to rise with its rewards, distinguish general holding from self-stake compounding and compare the latest result across validators without manually reconstructing identities and monthly balance histories.
The signal is strongest when used as one part of a broader decision: validator performance, identity, operational history, decentralisation and risk are still important. Retainment adds something those measures do not—the operator's apparent continuing exposure to the asset that nominators are being asked to lock for the long term.
Where retainment can go next
Monthly snapshots are the beginning. Several extensions could make the signal more useful:
- Longer time horizons. Rolling three-, six- and twelve-month views could separate durable treasury behaviour from one-off monthly expenses or balance movements.
- A standalone retainment dashboard. Graph history, identity changes and the relationship between retainment and compounding could be explored in one place.
- Automated DOT distribution reports. Periodic reports could summarise rewards earned, DOT retained in the identity graph, DOT added to self-stake and balance that left the graph, with clear caveats where destination or intent is unknown.
The goal is not to prescribe what every validator must do with every reward. It is to make an important part of validator–nominator alignment visible. With retainment and compound rate metrics, nominators can finally bring reward behaviour into the same evidence-based selection process they already apply to commission, performance and identity.