Protocol Learnings & Looking Forward

Protocol History

The PoolTogether protocol has gone through several major versions over the last 4 years. Each version of the protocol has built on learnings of the last to better create a protocol. The ultimate goal is a protocol that can be the foundation of global prize savings.

Throughout the versions, the core invariants of the protocol have remained the same – no loss, the ability to withdraw your deposit at any time, and the chance to win prizes. However, while keeping these things the same, the protocol has still changed in meaningful ways.

This document outlines PoolTogether Inc’s views on what has been learned in the recent major versions and what is needed to create a final version that can last as a true hyperstructure for prize savings.

PoolTogether Version 3:

PoolTogether V3 was the first version of the protocol to gain widespread adoption. The V3 launched in October of 2020 and took off in February of 2021 when the POOL token was introduced and automatically distributed to all depositors.

The V3 launched with a single asset supported – Dai. Shortly after being launched USDC, UNI, and COMP were added. The V3 initially only supported one yield source, Compound. But additional yield sources and tokens were added.

At its peak, the V3 protocol hit $235 million in total deposits and was distributing $180,000 in weekly prizes. In many ways, this was the realization of a long time goal – the protocol had life changing no loss prizes! However, there were a couple of issues that arose and made the growth unsustainable and prevented the growth flywheel from really going strong.

The first issue was simply gas fees. At the time, it could easily cost $100+ to deposit into PoolTogether on Ethereum. This precluded mass adoption from happening where all the largest prizes were. As an alternative with lower gas fees, the PoolTogether protocol was deployed to Polygon. However, there was no connection between the networks or assets so the prizes on Polygon had to start from scratch. They were far smaller averaging $6,000 per week at the peak. Users had to decide between large prizes and high gas fees or low gas fees and much smaller prizes and TVL on Polygon never grew past $10 million.

The second issue is the prize distribution on Ethereum was severely limited. There were only 5 winners per week and the prizes were evenly split among the winners. This made it almost impossible for small depositors to win anything and meant the average winners had millions of dollars in deposits. This created a huge amount of user frustration. Because of the way the protocol was built, the prize distribution could not be adjusted so short term fixes that didn’t solve the problem were pushed while the ideas for V4 started.

The third issue was the distribution of the POOL token. Some depositors were there solely or primarily to get the POOL distribution. POOL was distributed in real time meaning it was especially susceptible to yield farming because it could be accrued and sold in real time. This ran counter to the designed purpose of the POOL token distribution which was intended to get tokens into the hands of long term users.

In summary, the V3 finally had large no loss prizes, however, they were not accessible to normal people due to high gas fees, the inflexible prize distribution meant the prizes couldn’t scale and the over-reliance on POOL to drive deposits meant it was not sustainable and the growth flywheel couldn’t start.

PoolTogether Version 4:

The biggest improvement was moving to an unbounded prize distribution. This meant that there could be an unlimited number of winners and variable prize sizes. Instead of a $100,000 prize being split equally between 5 winners the protocol could distribute those prizes in any way! This introduced a far higher level of regular engagement and excitement. Instead of having 5 winners per week, the protocol now had thousands.

The V4 also fixed the Ethereum gas issue by creating a cross-chain prize distribution. That meant you could deposit on chains like Polygon and Optimism with low gas fees and still have the same chances to win prizes as Ethereum. The chains no longer fragmented prizes but combined prizes reinforcing growth.

The V4 didn’t solve the POOL issue, instead POOL token distribution was simply stopped with the introduction of V4.

The V4 introduced other great upgrades like “Chance Delegation” and the ability to distribute tokens to users based on their time weighted average balance. Like V3, fixed token distribution has proven to be a major driver for deposits with TVL going to new all time highs driven by $OP token distribution.

The V4 protocol though also had some new issues. The biggest is that large prizes disappeared, the largest prize ever awarded by the V4 protocol was $2,500. This was due to a limitation in the prize distribution that required prizes to all be awarded on the same frequency.

Another problem with V4 was that adding deposit tokens or yield sources to the network was no longer permissionless – they needed to be approved by POOL token holders. As mentioned, a problem with V3 was that no network existed, every new asset and yield source fragmented prizes. This meant people on Polygon could not access the prizes available on Ethereum, people depositing Dai could not access the USDC prizes, etc. The V4 fixed this by introducing a network to share prizes but there was no permissionless way to add more pools to the network. This became a big problem as yield rates for USDC went to historic lows.

Finally, the V4 did not have an autonomous and reflexive prize distribution. In the V3, the prize distribution was reflexive to yield generated. It automatically adjusted to match the generated yield. In V4 prizes were not directly tied to yield. This made the protocol less reliable to build against because the prize distributions could change in unpredictable ways. Ideally we want a prize distribution that adjusts according to immutable rules.

In summary, the V4 improved upon the V3 by unlocking the prize distribution design space and creating a shared prize network but growth was limited because no large prizes ever happened and there was no way to add new pools and asset types to the network. Additionally, the absence of the POOL token distribution removed a large growth lever.

What’s next?

It’s PoolTogether Inc’s belief that after 4 versions of the protocol a final version is in sight. A version that combines all the learnings thus far and can act as the long term, immutable layer for prize savings globally. A version that can be built around and extended but with a solid core that does not change.

The PoolTogether Inc team has been ideating, researching, and prototyping what we think this could look like over the past ~6 months. In the coming weeks, we will share more details around what we have built. Ultimately, as we are just contributors to the protocol it will be up to POOL token holders to decide to accept our contributions.

At a high level, we believe a new version would improve on the V4 in a few key ways:

  • Enable the permissionless addition of new asset types and yield sources to a prize network
  • Enable the prize distribution to be autonomous and automatically adjust based on yield
  • Combine life changing large prizes with small prizes
  • Integrate the POOL token into the protocol

We believe these changes can unlock an organic growth flywheel and ultimately provide a better way for the world to save money.


Great overview and I agree with a lot of your takes!

What are your thoughts on the reserve rate between versions and going forward? This was one of my favorite parts of v3 and I don’t feel like it had enough time to snowball and create gravity.

Another thing I miss about v3 was the Grand Prize - where you could win other tokens and NFTs. I think there is a lot of design space and creative ways to incorporate NFTs into PT. In future versions, do you think NFTs will be part of the core design, or more adjunctive?


oh, these are great questions!

What are your thoughts on the reserve rate between versions and going forward?

I too was a big fan of the perpetual growth mechanism. I think the narrative is really strong and it’s a sound mechanic. It is not an area of research we have been focusing on. I think the main issue is that 1) in the short term, it inhibits large prizes (because the yield for prizes is being diminished) and 2) it takes quite awhile for the reserves to get meaningful enough to contribute to the prize.

With that said, it’s a simple mechanic and so it could still be implemented.

Another thing I miss about v3 was the Grand Prize

The grand prize will be back in the sense that there will be one single prize much larger than the rest. However, what I think you are referring to is the “lootbox” where anyone could permissionlessly add to the prize. This was. very cool feature and I do hope to bring it back (although slightly differently).

We already have the “TWAB Rewards” contract that enables anyone to permissionlessly distribute any token to PoolTogether depositors. I would like to see in addition to this, a “TWAB prize” contract that similarly lets anyone add prizes. This is something that is both possible today and also would be possible in future versions of the protocol being considered.

I think there is a lot of design space and creative ways to incorporate NFTs into PT.

Agreed, this is a great thing to point out! There has been a ton of traction with PoolTogether + NFTs. It’s something I would like to continue to push into and even now the PT Inc team has been talking to David about a new NFT collection he has made that he wants to integrate directly with the protocol.

With that said, similar to the last question, I see the main goal right now to be making it easy to integrate NFTs into the protocol – not building a definitive NFT integration.

Our hope is that going into the future, becomes much more robust as the core protocol becomes easier to integrate with and extend.

1 Like

Thanks for starting this discussion @leighton. I have spent a lot of time watching both versions of the protocol. V3 and V4 both share A LOT of great fundamental ideas.

I could spend days assembling and organizing these ideas but am putting them out there a bit raw so please bear with me. I’ve included some opinions that could be viewed as contentious but genuinely think these topics need to be thought on and discussed. Please know that my intention is only to share my perspectives in hopes of agreeing together on how to move forward on this path to build the protocol that can be the foundation of global prize savings.


User interface / experience (I know, not protocol specific)

  • From what I can remember, issues for users were practically a constant in V3. The tech generally was not nearly as evolved - The Graph, for example, gave us endless failed fetches and delays that prevented the app from rendering at all.

  • The permissionless pool builder was awesome. The V3 community UI was a great concept, but lacked the most basic of features - like showing if someone won, or who won. Permissionless pools can be more supported.

Whales are the problem - “Whales win all the prizes” “Yearn farm and dumps the token”

  • I think this is mostly a narrative issue and one that we went too far to battle with V4. I’d be curious to see analysis of depositor percentiles in V3 vs V4.

“$180k in prizes per week” with V3

  • Prizes were overshadowed by the dollar value of POOL being distributed at the time. POOL incentives drove the growth bus, not prizes.

  • In the middle of April 2021, when we had $180k per week in prizes, POOL price was over $25

  • 2295 POOL per day were distributed to both USDC and DAI pools. Not counting other pools that was > $800k per week in POOL for the $180k in prizes.

  • Later in May of 2021 this proposal reduced POOL to 2700 per day (18,900 POOL per week and still incentives overshadowed prize value) PoolTogether Governance

The prize distribution was limited to equally divided prizes as mentioned but had other nice features

  • Like @uncle said, the prize distribution of V3 had a great feature to award any erc-20 or NFT. A very powerful idea for prizes and one that should be explored more.

Gas generally was an issue to adoption and a major pain point

  • We had success with Polygon pools and V3, but again it was MATIC incentives that drove deposits, not prizes.

  • Gas costs have been significantly less on Mainnet since the days of V3. L2’s really change the landscape of transaction costs and V3 on L2s would be totally different than Mainnet Ethereum.

The early withdraw fee to prevent gaming the protocol was a pain point of V3 but not as bad as gas.


Open sourcing the whole stack was an amazing improvement realized in V4 (not protocol specific)

  • The community showed up to build around the protocol in ways not seen before. This can be further fostered in a new version. Bringing in new assets will bring in new communities and new builders.

TWAB technology was created and offers a unique way to see historical balances onchain.

  • This is an awesome primitive but has it proven to be a feature that grows the protocol? What are the costs? Gas? I think this should be explored more before assuming it’s beneficial to carry over into the next version. The ability to build on top of the protocol using onchain TWAB data likely makes it so. I can see the potential but not sure we have seen the proof, yet.

Computing prizes and prize history offchain

  • I think this should be onchain data and recent discussion has leaned towards this being the case. Prizes should be at the core of the protocol.

We launched with daily prizes thinking we could later change to weekly but discovered that not to be the case practically speaking.

  • We had two $2500 prizes per day for the first 100 days of V4 that could have been awarded as one $35k prize per week. A big prize value was there but it was split into daily prizes that were not headlines. Again, prizes did not drive the growth we hoped.

We awarded $100k in prizes per week for the first 100 days of V4 with only $15mm - $35mm in TVL![|624x197]

Source PoolTogether V4

  • It’s important to ask why the subsidy/incentives for prizes did not bring the growth we wanted. Again incentives (OP), starting around the beginning of August, brought more value into the protocol than prizes. Less value spent with incentives, but more growth.

Statistical prizes are confusing.

  • With V4 the 1 grand prize per day can happen 4x in one day or not for 4 days. This is hard for users to understand and it’s hard to build hype on. We now have a randomly occurring prize targetted for once every 30 days that could happen 3x in a month or once in 3 months, but the amount is the same. I think if prizes are statistical they should be progressive, growing in value as they are not awarded.

Statistical prizes come with a prize cap

  • This is also confusing, leads to gaming, and arguably downright misleading. I don’t like it. It’s a way to punish whales, and further distribute prizes, which sounds good but the operations of it feel slimy.

The reserve / perpetual growth

  • V4 has the ability to grow a reserve (give out less prizes than yield generated) but the narrative is not the same because it’s not programmatic like V3 and has yet to be used in V4 as we have always subsidized. Again, let’s make it autonomous.

  • There is confusion at the social layer if the protocol should capture any yield. These issues need more discussion.

Deposit and Withdraw costs in V4 were not reduced from V3. Gas costs increased for users with Poolers having to claim prizes.

A lot of NFT campaigns drove new wallets to deposit (not protocol specific)

  • I would like to see analysis that this grew the protocol. Deposit transactions =/= growth in users or TVL most of the time from what I’ve seen. Lots of deposit transactions is good for L2 gas usage though.

Proposed to be resolved

  • Prize distributions became controlled by multi-sigs - We have learned a lot by being this close to the prizes but we should move back to being autonomous with prizes.

  • Prize disconnected from yield -Let’s reconnect it, I think everyone agrees. Autonomous.

  • V4 is USDC only - We all want to see more assets supported. This was a major limitation thus far in V4 but now that the pools are isolated we could add pools/assets/yield sources. I don’t think V5 is necessary to solve this particular problem.


  • 24 hour wait to claim - Confusing UX - has been resolved by DPR upgrade

  • TVL was communicated across chains - The design of networking pools to aggregate bigger prizes added centralization and the hope of bigger prizes never proved to grow the protocol. The cross-chain networking of pools was removed with the DPR upgrade. Future networking of pools should be more closely considered for cost/benefit.


I brought up Ethereum a few times because I think it’s important that we consider it in future development. V4 has not worked on Ethereum. Why? Narrative and gas costs? Perhaps most notably - a lack of incentives.


Ethereum still has 40x the TVL of Optimism. I don’t think it’s time to go all in on L2s yet. If we look at Aave on L2s we can see that any substantial prize pool will quickly dilute lending rates. In fact we are currently roughly ⅓ of the entire lending market on Aave Optimism. How do we revive PoolTogether on Ethereum?

Headline prizes

I think it’s a mistake to think headline prizes will be the biggest catalyst for growth in the near-term. Incentives have always been the driver and until there is broader adoption in the space I assume the trend will continue. We should be thinking about how to leverage incentives in a prize protocol way that makes sense to Poolers (I just thought of an idea but will share elsewhere). This is all not to say big prizes cannot drive growth. Big prizes are great and should always be a core component of the prize distribution design. Outsized returns are the PT value proposition and in my opinion we got it wrong with V4 subsidy allocations and daily prizes.

The most memorable proof of big prizes was the viral growth with OHM in V3 that included no incentives. This brings up a big question that we need more discussion on - assets/yield sources. We all want the ability to add assets and yield sources. Having USDC only in V4 has been a crux. How do we balance the risk of different yield sources with the no loss ethos of PoolTogether? I think permissionless pools are the wei. Now that we have disconnected the V4 network with the DPR upgrade we could support the addition of new prize pools / assets / yield sources. Something people have been wanting to do for over a year. Why wait?

Addressing Conclusions

  • Enable the permissionless addition of new asset types and yield sources to a prize network
  • Enable the prize distribution to be autonomous and automatically adjust based on yield
  • Combine life changing large prizes with small prizes


  • Integrate the POOL token into the protocol

This would be an experiment and one I remain skeptical on. Building the new version to be a foundation that can support these types of experiments makes sense. However, I think it’s imperative we get the core parts of a new version right before building an experiment on top.

What to build?

I’ll come out and say it - I think V3 had more going for it than V4. The core tenants of decentralization, autonomous, and permissionless are the power of DeFi. V4 missed that. The attempt to network pools across chains to aggregate prizes for growth never materialized and was abandoned. The resulting cost of only supporting USDC continues to stifle our growth. Effectively deserting Ethereum was a mistake.

The two major drivers of growth in V4 have been prize subsidy and OP incentives. Had we dripped $50k-$100k USDC per week to depositors with V3 instead of the V4 prize subsidy I’m confident we would have seen more growth for the protocol. I’m also sure that we would have the same success if not better on Optimism with OP incentives using V3 vs V4. People like streaming rewards over confusing TWAB epochs. We could even put some OP in the prize pool as prizes, whereas with V4 we have no setup deployed to add other tokens as prizes. The NFTs could also be prizes.

Building off of V3, which I believe was closer to the foundation of global prize savings, the three important things that I believe could be improved are - the prize distribution, gas costs and the early exit fee.

  • The autonomous prize distribution for a hyperstructure deserves it’s own thread. Flexibility for prize assets and distributions sounds promising in the form of TWAB modules. Thinking about how to make prizes as attractive as incentives is important. I think this is the core area to focus on and get right.
  • Are we considering gas costs and building to include Poolers on Ethereum?
  • TWAB solves the early exit fee and adds a lot of foundation for building on top.

Reflecting on successes and failures in an honest way is the first step to building the future we all want. I believe together our experience will find the best path forward.


This is a lot @underthesea! I’m going to focus specifically on your conclusions.

Building off of V3, which I believe was closer to the foundation of global prize savings, the three important things that I believe could be improved are - the prize distribution, gas costs and the early exit fee

Yes! You could certainly say that what we are talking about in the next version is much closer aligned with V3 than V4. If you’d like to think of it as an iteration over V3, this is how I would view it…

  • V3 had fundamentally constrained prize distribution. It had hard caps on how much it could grow. The next version should not have this. The prize distribution should be reflexive to yield and autonomous but not constrained.

  • V3 had a fundamental fairness issue, the TWAB fixes that

  • V3 had no ability to connect yield from pools, we believe prize pool should be able to be networked

  • V3 had no integration of the POOL token, we believe the token should play a role in the protocol

So we have two of your three main points (the one we don’t have is gas reduction, I’m unsure on how that one lands). I think we also have two additional points which you may not agree on their importance.

But generally speaking, I do think we are aligned on the direction and I think it is fair to point out that what we are talking about share more similarities with V3 than V4.

I want to make sure though I’m not missing anything important you are trying to say? Is there anything else you would point out in terms of divergence or agreement?

A lot to think about!

Re: prizes specifically, as I think that’s the most important aspect of the protocol, a lot of my opinions are above. Theres a lot to discuss there. The crucial question I feel is - How does the protocol deliver value in the most engaging and rewarding way to poolers? I think at this stage we should be paying more attention to incentives than yield, because they have been the primary driver of growth.

Where we plan to grow the protocol is important and I think we should have a vision for that. One chain? Only L2s? Any EVM chain? Deploying to all chains was a premise of V4 but was not the case in practice as we managed the protocol over the past 16 months.

I think we do agree on the core tenants- decentralized, autonomous, permissionless. To what extent we will go to enshrine these in code is tbd. It should be highest priority imo.

I think we disagree on networking prize pools and POOL.

I disagree. V3 pools are owned by POOL holders via governance. POOL holders can change V3 pools parameters with onchain votes. V4 is not run as such, it is managed by multi-sigs. This is a big distinction imo.

We networked pools in V4 and the tradeoffs cost us. I can’t really understate this. No networked prize benefit was seen and eventually we abandoned the network for isolated pools with the DPR upgrade. Networking pools failed in V4, do we agree on that? Why should we double down on this idea when it adds costs, complexities, and risks? Those don’t seem like good pillars of a hyperstructure.

1 Like