PoolTogether V5 (relaunch) Optimism Param Recommendations

PoolTogether V5 (relaunch) Optimism Param Recommendations

What is this?

This is not a final draft and is subject to change. Your feedback is welcome and appreciated!

This post outlines the launch parameters for PoolTogether V5 as recommended by Generation Software. The prize pool params are immutable and final for each deployment and cannot be modified later. Other parameters for the prize vault, claimer, and liquidation contracts have been excluded from this list since they are adaptive and can be permissionlessly experimented with by vault deployers.

Many of these parameters are intertwined with others such that a change in one param can affect the reasoning behind another. These dependencies must be considered when suggesting changes. As a result, when the final params go to a snapshot vote, it should be treated as an all or nothing vote rather than a vote on each individual param. As such, this post aims to start the conversation on the desired prize pool behaviour within the community.

This document also serves as a reference for future deployment params on different networks. By looking back at the reasoning behind the choices made for the Optimism deployment, future deployments can tweak the numbers to fit their target environment.


Param Recommendation TLDR
prizeToken WETH as abundant as air
drawPeriodSeconds 86400 (1 day) L2s are cheap and so is RNG
drawTimeout 60 draws (2 months) a long time, but not too long
shares (tier/canary/reserve) 100 / 4 / 10 no more penny prizes!
tierLiquidityUtilizationRate 50% ensures people get their prizes
grandPrizePeriodDraws 91 draws (3 months) GP needs to mature now

Supportive Reasoning

prizeToken | WETH

WETH is a very liquid asset, which makes liquidations and prize compounding cheaper to execute. The switch from POOL to WETH also reduces the need for robust POOL liquidity before deploying to a new L2.

drawPeriodSeconds | 86400 (1 day)

There’s very little downside to daily draws on L2s now that the RNG cost will be negligible. This becomes a question of effeciency: “Can the prize pool support daily draws without spending more than x% of daily yield?”. In the case of Optimism, it is very likely that RNG through Witnet will be less than 1% of expected yield even at low TVL.

drawTimeout | 60 draws (~2 months)

The draw timeout is a new addition to the prize pool that gracefully shuts down the pool if the RNG stops functioning. After this period passes without any draws being awarded, the remaining prize liquidity in the pool will be split between depositors based on TWAB and can be manually claimed.

The draw timeout must be long enough such that it cannot feasibly happen by accident or network outage. 2 months should give depositors and UI providers enough time to coordinate and provide users with easy paths to claim their share of the pool.

shares (tier/canary/reserve) | 100 / 4 / 10

For every draw, the new prize pool liquidity is split between each tier, the canary tiers, and the prize pool reserve. These params define the ratios of this split and affect the prize sizes and RNG payment.

The defined ratios have the following implications:

Canary Tiers

The canary tiers are now purely for gas discovery and are not awarded as prizes so the system can avoid frequent and inefficient penny prizes. There are now two canary tiers; if the first canary prizes are not claimed, the system will lower the number of tiers. Inversely, if the second canary prizes are claimed, the system will raise the number of tiers. This makes the tiers more predictable and easier to understand.

By setting the canary shares to 4, we ensure that the gas claimer bots spend on the canary tiers is kept below 1% of the daily yield in expected operating conditions. The canary shares also help determine the daily prize size. With the recommended ratios, we can expect to see daily prizes on Optimism that range from $1-$15 (based on current gas prices).

Reserve Shares

The purpose of the prize pool reserve has shifted in the new deployment as it is no longer required for prize backstops (see tierLiquidityUtilizationRate). The reserve is still used to pay for RNG, but since this will now be done through Witnet on Optimism, the RNG fees are expected to be very low and the reserve will have an excess balance at the end of each draw. The current plan for this excess is to use it to buy POOL from the market every draw and then burn the POOL, creating a deflationary pressure on the total supply.

The reserve shares must be set high enough to pay for the RNG every draw. The higher it is set, the more excess will be used for POOL buybacks at the end of each draw. With the recommended ratios, we can expect to see around 2-3% of daily yield used for POOL buybacks on Optimism.

tierLiquidityUtilizationRate | 50%

The tier liquidity utilization rate is a new addition to the prize pool and is intended to be a more effective strategy to backstop prizes that are over-awarded. Since prizes are based off a statistical model, there is some variance in how many prizes are awarded each draw. This is unpredictable and if more prizes are awarded than expected, there won’t be enough liquidity to cover them which causes a race condition in which the first prizes to be claimed are the ones that receive the liquidity. This is an undesirable event that the utilization rate helps protect against.

After some statistical analysis on the chances of these undesirable race conditions occurring, we’ve determined that if we use a utilization rate of 50%, we can expect there to be less than one race condition in every 10 years of prize pool operation.

This analysis assumes we will be operating at 5 tiers or more. Simulations on Optimism show that this is achievable with a minimum of $1m TVL @ 5% APR.

Prize Maturity

This feature also comes with an additional side effect in which prizes keep a higher value even after their expected liquidity has been used. For example, if the GP is awarded with a utilization rate of 50%, half of the accrued liquidity will still remain in the GP tier for the following draws (without this, all the liquidity would be used and the GP would reset at 0 the following draw).

This will lower the prize size for the first few GPs, but over the long term the liquidity will stabilize and the GP size will be roughly the same regardless of this param. For example, let’s assume that the GP is awarded exactly at the end of the GP period every time and that $100 in liquidity is added to the GP over each period. With a utilization rate of 50% the liquidity remaining in the GP tier can be represented by the following chart:

X-axis is # of GP periods passed, Y-axis is liquidity remaining in the GP tier.

We can see in the example above that over time, the liquidity stabilizes and the prize value is equal to the liquidity coming in over that period. It will take 4 GP periods for the GP to pass the 90% maturity threshold. As such, a lower GP period has been recommended to reach this maturity faster.

grandPrizePeriodDraws | 91 draws (~3 months)

With the addition of the tierLiquidityUtilizationRate, the GP now has a “maturation period” in which it will grow to it’s full potential. With the recommended utilization rate, this will take about 4 GP periods before the liquidity stabilizes and the GP has fully matured. As such, a lower GP period has been recommended so that the GP can reach maturity within 1 year.

The first GP will be worth half of what a fully mature GP will be. Check out the spreadsheet here.


As described above, these are the standing recommendations, but are subject to change. If you agree with these recommendations or would like to see any changes, now is the time to discuss! There will be a snapshot vote that ratifies the final parameters before deployment.

  • I agree with these params
  • I think some changes should be made and I will explain why
  • This is too complex for me, I just want prizes
0 voters

Daily $WETH prizes?
Regular grand prizes that grow bigger each time?
A backstop that distributes the yield in case the protocol ever stops?

You got me! I’m very excited about the tweaks that were made. Great work.


Update on Reserve Shares

Today the G9 team dug into what the value of the reserve shares should be, so I’m going to provide our updated recommendation and rationale.

Reserve shares: 30

The reserve shares determine the portion of yield that is used to cover RNG costs, with the remainder going to POOL utility. We estimate that the minimum reserve shares is 30, given tier shares of 100 and canary shares of 4. This is the lowest amount of reserve in order to cover RNG costs.

We calculated the minimum reserve shares such that net reserve = 0 for a given minimum Prize Pool TVL and APR.

That is to say, the minimum TVL and APR we believe that is needed in order for the prize pool to award prizes.

Net reserve = gross reserve - RNG costs = 0 for minimum prize pool size


The minimum number was computed with these assumptions:

  • Daily RNG costs of $1.20 (based on testnet)
  • Smallest functioning Prize Pool has $100k USD TVL and yields 3.5% APR.

At the minimum TVL and APR the yearly gross reserve would be $441, and the yearly RNG cost would be $438.

Net reserve = $441 - $438 = $3.

The reserve will net (nearly) zero; there is only enough reserve to cover the RNG costs.

Net Reserve

The net reserve as a percentage of yield varies due to the way the algorithm is designed; the net reserve depends on the number of prize tiers.

The prize pool increases the number of prize tiers if the prizes are too large. If prizes are the right size, then the number of tiers remains the same. We’ve made the assumption that the number of tiers is stable when the daily prize is approximately $1 in size.

With reserve shares set to 30, the net reserve percentage of yield would be on average 4.3% under stable conditions, with a range from 0.1% - 6.7%.

During unstable conditions; i.e. when the number of tiers is about to increase; we’d see the net reserve as a percentage of yield range from 0.1% to 10.4%. However, 10% would only be a temporary outlier before the number of tiers increases and the percentage would jump down again. For more detail see the spreadsheet below.


If you’d like to see the calculations for yourselves, check out the PoolTogether V5 Reserve Google Sheet.

Next Steps

This finalizes the G9 recommended parameters. If you believe more reserve should be captured, please speak up. Otherwise, we will shortly move to Snapshot vote on these parameters.


Snapshot signal is now live until April 12