Introduction
SAM was built to help Solana stake flow to validators that deliver the best returns, ensuring stakers consistently earn the best APY on the market. As market conditions have evolved, some of SAM’s original assumptions around stake movement and validator incentives have shifted, creating a need to refresh the system so it continues to perform as intended.
This proposal updates SAM by (1) letting bids determine stake priority so higher-value validators receive stake sooner, (2) introducing a Bond Risk Reduction Mechanism to better align bond maintenance incentives with re-delegation costs, (3) removing the largely unused MNDE Enhanced Stake mechanism to simplify the system, and (4) increasing per-validator stake caps through a controlled rollout.
Together, these changes make stake allocation leaner and reduce wasted re-delegation costs. The result: a better SAM and higher APY for stakers.
Proposal
If passed, this proposal would mandate and require of Marinade Labs to implement the necessary changes to Marinade Protocols so that the following are satisfied:
-
Allow validators to win stake priority in the auction
With this change, validators will become able to directly influence who Marinade delegates to first.
-
Introduce a Bond Risk Reduction Mechanism
To improve protection of stakers’ rewards and guard against validators abusing the system by letting their bonds run to depletion. The validators who do this will now be charged the full cost of the re-delegation of their stake elsewhere which will protect stakers’ rewards and boost Marinade APY.
-
Abandon MNDE Enhanced Stake
Current State: MNDE Enhanced Stake gives real power to MNDE holders by allowing them to influence which validators can receive more stake. Locked MNDE can be used to increase a validator’s maximum stake cap. This feature as of today does not change any caps as it is largely unused by MNDE holders.
Removing this feature will simplify the infrastructure paving way for other improvements.
-
Increase the maximum Stake a single Validator can obtain up to 15% of TVL
Current State: Validators are capped at 4% of Marinade’s TVL (MIP-10). Prior to MIP-10, the cap was 2% of Marinade’s TVL.
This change is in line with allowing the best validators take up the lead in securing the Solana network.
Rationale
1. Allow validators to win stake priority in the auction
At the core of SAM, there is a fundamental tradeoff between re-delegating stake to the highest-performing, top-bidding validators and keeping the stake distribution as-is. This tradeoff exists because re-delegating stake incurs a cost: for one epoch, the stake must remain undelegated and unproductive.
There has been a tension emerging lately, as the slowing market has dramatically reduced opportunities for Marinade to redelegate stake to higher-bidding validators profitably. To protect stakers’ APY, Marinade has been re-delegating less and less stake, since in the current market re-delegations often cost more than they are able to deliver in value. This behavior, albeit optimal, undermines SAM’s value proposition for validators. The open access to Solana stake that SAM exists to provide has disappeared as validators now wait many epochs to receive meaningful stake delegations from SAM.
To fix this situation, this proposal seeks to allow validators to bid to win stake priority in the SAM auction. Once implemented, stake priority will be assigned solely based on validators’ bids. A validator whose bid—after accounting for their commission—is higher than another validator’s bid will receive their stake allocation from Marinade strictly sooner than the lower-bidding validator. To keep the competition healthy, validators will start paying for the activating portion of their Marinade stake. The bid they will pay for the activating portion will be the difference between their actual bid, and the bid they pay currently for active stake. Payments for the non-activating portion, which includes all of their already active stake, will remain unchanged.
2. Introduce a Bond Risk Reduction Mechanism
Over the past two quarters, Marinade Labs has carefully monitored SAM’s efficiency. This analysis identified that the main source of inefficiency is validators SAM must undelegate due to insufficient bond coverage. If left staked, these validators would expose stakers to losses from downtime or commission rugs.
The proposal of the Bond Risk Reduction Mechanism seeks to reduce re-delegation costs by incentivizing validators to top up their bonds promptly to remain staked or exit the auction properly by withdrawing their bonds completely—in which case Marinade will redelegate their stake elsewhere through the standard process.
This proposal may seem incomplete, as validators might exit the auction to avoid the penalty, only to re-enter later. Such behavior would still require Marinade to pay for the re-delegation, rendering this change useless.
This dodging behavior, however, would not be economical. If a validator chooses to leave and re-enter later, it could take considerable time for them to reach their previous stake levels. Moreover, with the introduction of payments for activating stake, stake acquisition from the auction no longer comes for free. All of these would hurt their bottom line making this behavior unprofitable.
Once the proposal is implemented, only validators who truly want to exit long-term will opt to do so. The rest will just make sure their bonds are topped up properly and in a timely manner, saving re-delegation costs, improving APY for Marinade stakers as a result.
3. Abandon MNDE Enhanced Stake
While MNDE Enhanced Stake was designed to give MNDE holders meaningful influence over stake distribution, in practice it is now seeing little to no usage (as can be seen from the public auction data) and has not materially affected validator stake caps for a while. Maintaining this mechanism therefore adds complexity to the delegation system without delivering proportional benefits to stakers, validators, or MNDE holders.
Therefore this proposal seeks to abandon MNDE Enhanced Stake feature and simplify the infrastructure and governance surface of SAM, reducing cognitive and technical overhead while allowing Marinade to focus on improvements that directly enhance capital efficiency, stakers’ APY.
4. Increase maximum Stake a single Validator can obtain to 15% of TVL
The main strength of SAM has since its inception been that the validators offering the highest yields receive stake and are in turn rewarded with higher rewards flowing from commissions. This ensures that stakers get the best APY on the market and the validators get their fair share as a payment for their excellence. To further deepen this main strength it is proposed to increase the per validator cap on stake a single validator can get to up to 15% of TVL. The current—4% cap—artificially fragments stake among validators who may offer inferior yields, diluting stakers’ returns. Increasing the cap to 15% of TVL will allow proven top performers to be rewarded and, above all, boost APY for those who stake with Marinade.
This proposal authorizes Marinade Labs to increase the caps gradually in at least two rounds at their discretion.
The ASO and Country concentration limits remain unchanged.
Technical Details
Winning stake priority in the auction
Once implemented, Marinade will allocate stake strictly by the formula used currently in the auction
totalPmpe = bidPmpe + inflationPmpe + mevPmpe,
where
inflationPmpe = expectedAverageInflationPmpe (1 - inflationCommission), mevPmpe = expectedAverageMevPmpe (1 - mevCommission),
where expectedAverageInflationPmpe and expectedAverageMevPmpe are the currently used estimates.
All of this is subject to existing caps and constraints, in descending order.
On top of the current payment, validators will start paying for activating stake as well, this means that the total payment for stake each epoch will be
totalPaymentForStake = auctionEffectiveBidPmpe (activeStake / 1000) + max(0, bidPmpe - auctionEffectiveBidPmpe) (activatingStake / 1000), where auctionEffectiveBidPmpe = auctionWinningPmpe - inflationPmpe - mevPmpe.
Stake activates during a single epoch after delegation, so this payment for newly received stake is only transitory and occurs exactly once for each unit of new stake.
Example
Epoch 100: stake is won
Epoch 101: stake activates
To keep things simple, we assume there is a single commission, the inflation commission. MEV commission, if any, is treated in a similar manner.
Effective Bid is based on the Bid while taking inflation and MEV commissions and the expected size of the respective rewards into account.
Assume the auction winning price in all cases is: 0.072 SOL / 1k SOL
New stake available to delegate in Epoch 100: 15,000 SOL
Epoch 100 — Auction outcome (stake won now, activates next epoch)
| Bid | Inflation Comm. | Effective Bid | Already Active (start of Epoch 100) | Stake Won in Epoch 100 (activates in 101) |
|---|---|---|---|---|
| 0.090 | 10% | 0.081 | 20,000 | 15,000 |
The validator’s Effective Bid is higher than the winning price and hence they receive an allocation of new stake.
Epoch 100
Nothing is activating yet.
As the commission is 10%, the validator pays is 0.079 SOL / 1k SOL which is derived from his commissions, the expected rewards and the auction winning price.
| Activating in Epoch 100 | Pays on Activating | Already Active in Epoch 100 | Pays on Already Active | Total Paid (Epoch 100) |
|---|---|---|---|---|
| 0 | 0 | 20,000 | 0.079×20 = 1.58 | 1.58 |
Epoch 101
The stake won in 100 activates.
| Activating in Epoch 101 (from Epoch 100 win) | Pays on Activating | Already Active (non-activating) | Pays on Already Active | Total Paid (Epoch 101) |
|---|---|---|---|---|
| 15,000 | (0.090-0.079)×15 = 0.012×15 | |||
| = 0.180 | 20,000 | 1.58 | 1.76 |
Active stake at start of Epoch 102
| Validator | Active at start of Epoch 100 | Stake won in Epoch 100, activated in 101 | Active at start of Epoch 102 |
|---|---|---|---|
| Alpha | 20,000 | 15,000 | 35,000 |
Bond Risk Reduction Mechanism
Once implemented, the validator will pay a penalty for not topping up their bond which would cover the re-delegation costs Marinade would incur by taking their stake away.
A number called minBondRequired will be introduced. This number will reflect the possibility with which Marinade will have to remove stake from the validator due to their bond not covering the risk associated. Once below a threshold, Marinade will redelegate a portion of their stake so that their bond after the charge will be good for their new stake, or until their bond is depleted.
Let
totalPmpe represent the total pmpe for the validator, which combines their bid amount with inflation and MEV rewards pmpe
effPmpe represent the auction effective pmpe for the validator, which combines the auction winning bid corrected for their commissions with the expected inflation and MEV rewards pmpe
maxEffParticipatingBidPmpe represents the estimated upper bound on the amount the validator will pay out of their bond for 1,000 SOL of stake in the upcoming epochs. The specification of how this number is calculated is left to the discretion of Marinade Labs.
minBondEpochs be a positive parameter specifying for how many epoch payments should the bond last
bondBalanceSol be the balance of the bond excluding pending withdrawals
marinadeActivatedStakeSol be the amount of active delegation on the validator in question
At each epoch, where the bond is not enough to secure at least minBondEpochs worth of bids on top of the PSR protection, we will undelegate forcedUndelegation of stake and charge a bond risk penalty bondRiskPenalty so that the bond would be good enough for the next idealBondEpochs epochs.
That is if
bondBalanceSol < minBondRequired
where
minBondCoef = (totalPmpe + minBondEpochs * maxEffParticipatingBidPmpe) / 1000
minBondRequired = marinadeActivatedStakeSol * minBondCoef
Marinade will take forcedUndelegation of stake and charge bondRiskPenalty that satisfy
(bondBalanceSol - bondRiskPenalty) / idealBondCoef
= marinadeActivatedStakeSol - forcedUndelegation
where
idealBondCoef = (totalPmpe + idealBondEpochs * maxEffParticipatingBidPmpe) / 1000
bondRiskPenalty = forcedUndelegation * effPmpe / 1000
This ensures that the redelegated amount is paid for by the bondRiskPenalty on top of which we will not charge anything in the next idealBondEpochs - minBondEpochs - 1 epochs if the auction winning prices, bids and maxStakeWanted won’t change substantially.
It can be calculated that
forcedUndelegation = min(marinadeActivatedStakeSol, forcedUndelegationBase / forcedUndelegationCoef)
where
forcedUndelegationBase = marinadeActivatedStakeSol - bondBalanceSol / idealBondCoef
forcedUndelegationCoef = 1 - (effPmpe / 1000) / idealBondCoef
Since idealBondEpochs > minBondEpochs the quantity forcedUndelegationBase will always be positive, and so will forcedUndelegation.
Conclusion
SAM was built to help Solana stake flow to validators that deliver the best returns, ensuring stakers consistently earn the best APY on the market. As market conditions have evolved, MIP-19 refreshes the system so it continues to perform as intended—making stake allocation leaner and reducing wasted redelegation costs, with the result of a better SAM and higher APY for stakers.
If you support these updates, please vote YES; otherwise vote NO and leave a feedback in the forum thread.