COMP Distribution Speeds

Each block, a total of 0.1760 :comp: are distributed to suppliers of assets, and 0.1760 :comp: to borrowers; this is the COMP Speed parameter. COMP speeds are allocated by market, e.g. DAI is 0.0977, and ETH is 0.0016.

With the passage of Proposal 033, COMP Speeds are now set through governance (by calling setCOMPSpeed for a given market), and are no longer automatically/dynamically adjusted based on borrowing demand. This is an important change – the community now has the power and responsibility to allocate COMP by market.

The current allocation is a relic of the moment that the proposal went into effect; the COMP Speeds will stay at the current rates, until they are set through governance.

The prospect of governance proposals to modify the allocation is exciting–before the community acts, here are a few guiding questions to shape how this tool is used:

  1. Should the sum of COMP Speeds match the current total, 0.1760? If markets are adjusted individually, should an increase in one market be offset by a decrease in another?
  2. How should important collateral markets like ETH and WBTC, that historically have little borrowing demand, be allocated :comp:? How would this change user behavior?
  3. Should assets that behave similarly be given similar COMP Speeds? Can assets like BAT, ZRX, and UNI, or stablecoins like DAI, USDC, USDT be grouped?
  4. How can the COMP Speeds be used to increase the distribution to more users, and increase decentralization?
  5. Should the community set COMP Speeds infrequently, or actively calibrate them?

At this time, the COMP Speed applies equally to suppliers and borrowers–modifying that relationship would require a significant protocol change.


Would it be possible to dynamically adjust comp speeds? For example: borrow rate on USDC is 20% and no new supply is coming in, maybe increase comp rewards for a set # of blocks, possibly scaling every few blocks to sweeten the pot to incentivize people to supply more USDC. This would work for all the markets and would incentivize people to keep reserve ratios at healthy levels.


My starting point on this would be “if it isn’t broken don’t fix it”. To me the current distribution speed and market allocation seems pretty good. Intuitively it makes sense – markets with higher interest rates receive larger amounts of COMP. I would like to see how the vesting effort plays out before making major changes.

I don’t really want COMP to distort normal market behaviors. So that would mean you always would want the distribution APY to be lower than the market APY The one exception might be the COMP market itself. Distributing more COMP there could be a nice way to reward holders… or it may just get abused by farmers.

Those are my first thoughts :slight_smile:


Hash: SHA256

Optimizing COMP Rewards


As Compound pioneered liquidity mining, it is only natural to ask if the distribution of rewards can be optimized. Why might the community decide that this makes sense? Firstly, it is likely — as often happens with market maker and liquidity incentive programs at centralized venues — that an issuer ‘overshoots’ and overincentivizes the either the supply or demand side of a two-sided market. This problem isn’t unique to financial asset incentives, either — even ride-sharing markets like Uber and Lyft are notorious for losing money due to excess payments to both the demand side (riders) or supply side (drivers). In Compound’s case, COMP emissions are used for a multitude of reasons, including but not limited to:

Given the fixed cap, finite supply of COMP, overspending on one of these components reduces the protocol’s ability to spend on other components. As such, it is prudent for the protocol to optimize how much it spends on liquidity to ensure that other uses of COMP are not drowned out. Moreover, liquidity providers already earn yield from borrowing activities and it is not clear how much of that liquidity would leave without active experiments. However, it is clear that liquidity incentives are one of the best ways to decentralize a protocol and maximize the number of active token holders — which should be taken into account when adjusting rewards.

Let’s do some back of the envelope math to compute how much the protocol is currently spending on liquidity

      • 7 day moving average of the number of blocks produced by Ethereum per day: ~6514.29
      • Price of COMP: ~$150 (USD)
      • Total COMP emitted per block (both supply and borrow sides): 0.352 COMP

This means that the protocol is spending ~$343,954.29 per day or $125,543,314.28 per year on liquidity incentives. Spending ~$125M to attract users is quite expensive! If we approximate the number of users of Compound by the an optimistic estimate for number of active addresses (~300K) that means that the protocol is spending $416.67 per user per year on customer acquisition costs (CaC). These are numbers that would scare any investor outside of cryptocurrency!

How can we optimize incentives?

Numerous other industries that manage two-sided marketplaces, such as ride-sharing, centralized trading exchanges, and gaming, deal with incentive optimization using data science. Given historical data and models for how users behave, market designers can effectively construct A/B tests. These tests could operate in the following manner (stylized and simplified for readability):

  • At time t, on-chain data and centralized market data is used to make a prediction of how much liquidity will change as a function of incentive (= USD price of COMP * COMP emitted)
  • This leads to a hypothesis test:
    • Null Hypothesis: Reducing the COMP emitted to the Dai market by 10% will reduce liquidity by 15%
    • Hypothesis: Reducing the COMP emitted to the Dai market by 10% will reduce liquidity by much less than 15%
  • A governance proposal is proposed (akin to Proposal 021) to reduce emissions by 10% to the Dai market
  • If passed, the data aggregated from monitoring emissions for some time interval T (say, 30 days) will provide us a way to estimate whether the null hypothesis can be refuted or not
  • New data from the test will be reincorporated into the model (first step) and used to construct another hypothesis test

The literature on designing such trials for online, two-sided marketplaces is vast, but none of this has ever been done using markets with full data transparency, like a DeFi protocol.

How to answer @rleshner’s questions

Part of what we are working on at Gauntlet is something called Automated Governance. This is system that continuously runs simulations (improved versions of those used in the Compound market risk report and in our DeFi Pulse score for Compound) and estimates ‘safe’ and ‘unsafe’ parameter values. These parameter values can be for per collateral protocol parameters (e.g. collateral factors, reserve factors) as well as for COMP incentives.

We propose that all of @rleshner’s questions can be analyzed in a scientific and experimental manner using a system like this for proposing hypotheses. In my opinion, it is impossible to answer these questions for a novel, unique market like Compound without performing some experiments via adjusting COMP speeds. These experiments will naturally be able to tell the community if spending $125M per year is expensive or if it is necessary in the competitive DeFi environment. Moreover, experiments can easily tell us whether these speeds need to be adjusted infrequently or not. For instance, if an experiment to reduce emissions by 10% leads to the null hypothesis staying true, then we have accrued information that suggests that user behavior is insensitive to changes and speeds should not be adjusted too frequently.


The beauty of cryptocurrencies is that their data is completely transparent and public. This means that anyone who has a clever model can test it out without having to spend money on data (like in traditional finance) or harvest user data and make it nonpublic (like a tech company). However, someone has to actually produce this data and analyze it. The only way to do that is via carefully designed experiments whose results can be clearly interpreted, much like a clinical trial or A/B tests in centralized online platforms. Compound led the way by showing that yield farming was a viable methodology for protocol decentralization and it can lead the way in providing scientific rationale for parameter choices in the years to come.



Strongly agree with this. It’s not wise to anchor on the status quo, especially since this distribution was initiated without any foreknowledge of COMP’s price or its effects on lending/borrowing rates. Compound needs a more rigorous way of determining: what is the best COMP emission rate that maximizes the long-term value of Compound. If it’s higher than the current rate, increase it; if it’s lower than the current rate, decrease it.

Sticking with the status quo makes no sense if it was arrived at basically randomly.


I like Tarun’s idea of experiments to optimize the rate.

And as Tarun also points out, we could reallocate some/much of this COMP to other (perhaps more productive) uses- the forefront for me being development work on the protocol. For example, using COMP to fund initiatives into new lending products could expand the user base and assets on platform in way that’s more effective and enduring than a COMP distribution to protocol users. At the very least, it seems like an attractive area to explore.

Personally, I could do without such a large COMP subsidy- it distorts the natural market that forms between borrowers and lenders (ie this protocol is already very useful without the subsidy) and also leads to a lot of unproductive actions within the protocol, such as yield farming. The distribution is quite expensive and has unclear utility to protocol growth. Running A/B tests and other such experiments will help quantify that utility to some extent, but I already have a strong hunch that some of that ~$125mm a year in COMP distributions could be better spent on funding/studying new initiatives/products.

At the very least, I’d like to see the COMP distributed adjusted such that Net Borrow Rate > Net Supply Rate for each token (by Net, I mean the rate net of COMP distribution and reserve). It seems crazy to have positive carry in the same token on the same protocol.


Very well written, by @tarun but completely based on Hypothesis that COMP is about 150$ and that protocol is spending $ on liquidity incentives.

Reality is that COMP is about 150$ with current market free float. Do you really believe if we, for example, double market free float of COMP right now, price will hold? I sincerely doubt it. In future it might though, if market grows. However from observing current conditions we can also suggest that current release of COMP to market doesn’t move price much, and demand keeps up with selling pressure, which is likely less than daily emission, as some are probably accumulating.

Protocol isn’t really spending any on liquidity incentives, as it's merely distributing tokens, created from thin air and not really backed with any sort of value and which are 150 only as long as there is a demand at that price.

However, even if we look at COMP from perspective of zero-value token, and merely an utility token for governance, do we think that all tokens, reserved for distribution, should be given out to liquidity miners? I believe not. And that is reason why i believe current structure should be adjusted. And that is also a reason why i believe airdrop to early users should be done as well. Not because i see it as giving out money, but because it’s by far better idea, than just give everything to liquidity miners. They are going to get major share anyway.

So from that perspective i think it’s very reasonable suggestion to decrease emission for dai market by 10% and observe results.

What i would Suggest personally?

  1. decrease daily emission by 10%-20% (That could be taken either exclusevly from dai market, or dai and usdc market)

  2. Use that 10-20% of emission for continious market sell for stable coin/eth/wbtc. Use the proceeds to create treasury. Hypothesis here is: Liquidity farmers are selling good part of daily emission to market and that doesn’t collapse the price. Protocol could as well sell that tokens personally, rather than giving them out to farmers so they could sell and have profits. What protocol should do with treasury? First it should direct that money to reserves of the biggest utilised markets on borrow side until reserves reach at least 1% of supply for every market. It would take quite a time, what to do later after goal is achieved could be decided at a later time. That will greatly derisk markets by seeding a reserves to be used in case of unexpected events, like Coinbase DAI price event. As natural reserves creation is going take forever with current reserve factors. And COMP reserves, are not really reserves as it’s not money until it’s sold, and you don’t now how much you could possibly get for it untill you actually sell. It’s like paper profits. And there’s no difference if supply is limited or not, as certanly it’s very much possible to create COMPV2 with different supply, some conversion rate COMP to COMPV2 and transition governance to new token. So it’s limited only as long as governance agree and want it to be limited supply token. Doesn’t mean it’s set in stone forever and impossible to change. Always nice pitch though, of course :slight_smile:

  3. Redistribute remaining emission, rerouting part of emission for stable coin markets towards ETH and WBTC markets.

Minor markets could be adressed later, but in general i think good approach would be take some percentage of total daily emission, let’s say 10%, and distribute it in equal portions to minor markets and observing the results. (Interesting here is would it boost utilisation or not) I believe there should be minimum portion of total daily emission every market should recieve regardless of utilisation.


A main reason of having large COMP rewards was to decentralize ownership of the protocol amongst protocol users. This is a major difference for DeFi protocols vs. the traditional comparisons above: customer acquisition cost is not necessarily what should be optimized. The $125m/year is about spreading ownership, not just attracting users. Decreasing the rate of rewards (w/o increasing it elsewhere) slows that down. There may be good reasons to slow down the decentralization, but I’m not really seeing that discussed.

To me, the larger question is: how decentralized is the protocol becoming through the distribution (and how can that be optimized)? There are many ideas floating around about this (like vesting). Ideally, this would also happen to go hand-in-hand with attracting many new users from the distribution and fixing some of the market distortions arising from the current distribution pattern.


Thank you for this helpful reminder on how the COMP distribution aids in decentralizing the protocol. Perhaps we could then just at minimum adjust the velocity of COMP distributed in each market such that Net Borrow Rate > Net Supply Rate for each token (by Net, I mean the rate net of COMP distribution and reserve). By righting these incentives you would be more likely to distribute COMP to folks actually using the protocol for its intended function vs sending it to folks who are yield farming or trying to get free carry. Note that the COMP distribution economics in this adjusted case would still help decentralize the protocol, make the yield for lending better, and also subsidize the cost of borrowing- just not so much it borders on a handout for those who lever up on both borrowing & lending simultaneously.

Great set of comments & opinions–

@johndoh this is how the distribution originally functioned; it scaled the COMP allocation as a function of the interest rate in each market. It was very easily gamed, as users/farmers attempted to use whichever market already had the highest interest rate; first, USDT, then BAT, then DAI. It’s possible that this model combined with vesting/cooldown (Gauntlet is doing fantastic work there) could be viable.

@Sirokko and @aklamun bring up an extremely important point, which is that the distribution was originally (and still is?) intended to distribute COMP to users that want to be long-term stakeholders in the protocol; to participate in governance; not to “farm”. Changes to the protocol or market allocations that decrease the distribution to “farmers” and increase them to “users” (also to @lay2000lbs point) fit this goal.

Increasing the allocation to the large collateral assets (ETH, WBTC) and COMP (and in part, away from stablecoins that could lead to recursive leverage) could begin to achieve this.

And to @tarun and @haseebq’s point–the community should prepare to experiment, improve, and adjust the COMP distribution to best achieve it’s goals.

Taking all of this together, and speaking with a number of you in Discord, I’ve tried to synthesize a “baseline” set of allocations by market, which we can then run experiments against.

This baseline:

  • Maintains the current total distribution (0.1760)
  • Creates a modest allocation to the primary collateral assets (which didn’t exist previously due to a lack of borrowing demand)
  • Standardizes (but doesn’t materially increase) the allocation to secondary collateral assets
  • Maintains large allocations to stablecoins, while setting DAI and USDC to be equal to eachother as a point of reference for future experiments
  • Increases the allocation to COMP

Does this jive with everybody’s thinking?


@rleshner I do not quite understand how you got the numbers in Supply and borrow, can’t be COMP per day, as if speed is same than it couldn’t be more than current about 2300 total COMP. But regardless, i’m a huge proponent of simplicity, predictability and ease for understanding, so take a look at my suggestion and below i’ll explain how i arrived to that numbers and some reasoning behind.

I group all existing markets into 3 Categories:

  1. Important Collateral Markets
  2. Important Stablecoin markets
  3. Minor markets

First i took 10% of distribution (0.0176) and distribute it in equal portions to each active market. Which makes 0.00195 per market. That portion just goes there regardless of current utilization of that market.
Next 10% of distribution goes to ETH market
Another 10% goes to WBTC market.
Both of theese markets are important Collateral markets.
Remaining portion is split equally between 2 stable coin markets. USDC and DAI. USDT market is positioned in Minor market because it’s not usable as collateral and thus can’t possibly compete with USDC and DAI at Compound. It’s not going to be big regardless of demand for that token as long as it’s in that state.

There might be arguments that Collateral markets might need bigger share, but we should keep in mind that we already taking away some distribution from stablecoins, we should observe impact and don’t introduce fast radical changes.

As for 50/50 and 75/25 distribution reasoning is. I believe it makes no sense to micromanage exact numbers, as liquidity will always be faster than governance, 50/50 split existed for quite a time, people used to it, liquidity will be easy to adopt and adjust. Could be changed later if needed, but i think it’s a good basis point. At introduction there might be some earning opportunities in minor markets, but i expect money quickly to adjust and profitability will balance itself.
I believe we should start with 75/25 for collateral markets, as borrow side of that markets was always on the low side. There might be some arguments that 25% might be too big for borrow side, but that’s for a reason. Here we not just follow market, we creating some initiative to bring more utilisation to that side. Not too much, that’s why it’s not 50/50, but big enough that we could see some interest there. I believe in long run, it beneficial to have somewhat more on borrow side of ETH and WBTC markets.

I believe it’s good simple starting point. Can be easily adjusted from there if need arise.

         Current            New            Supply         Borrow
ETH       0.0016        0.01955           75%            25%
WBTC      0.0011        0.01955           75%            25%
BAT       0.0000        0.00195           50%            50%
ZRX       0.0013        0.00195           50%            50%
UNI       0.0006        0.00195           50%            50%
USDC      0.0675        0.063575          50%            50%
USDT      0.0057        0.00195           50%            50%
DAI       0.0977        0.063575          50%            50%
COMP      0.0004        0.00195           50%            50%

Total     0.1760        0.176`
1 Like

FWIW, I’m coming around to the idea that it would be fine for current Net Borrow Rate < Net Supply Rate, provided COMP vesting will happen in such a way that the user has to take a non-trivial price drift risk in the COMP component of the yield. If that’s the case, then probably okay to just optimize for depth of collateral markets per unit COMP distributed with the aforementioned approaches.

1 Like

I would like to throw out an idea on the point “allocations that decrease the distribution to “farmers” and increase them to “users” (also to @lay2000lbs point) fit this goal.”

If we define “farmer” as a user that frequently claims and sells COMP to compound farming returns, cash out, etc., would some sort of distribution model that “punishes” frequent claiming (rather than conventional vesting, more opt-in), be worth considering? Has this been brought up?

I certainly lack the sophistication to model it formally myself, but intuitively, it seems that allowing some level of farming activity to subsidize liquidity (to some degree), while at the same time rewarding users who accumulate COMP (and perhaps especially those who vote with it) fits the goal of shifting distribution from farmers -> users.

The “punishment” (/pentalty) for the “farmers” (however that makes sense to codify, certainly lots of ways to do that… e.g. carry traders, frequent claimers, etc.) could even be distributed to treasury.

1 Like

@Sirokko re: the original post:

  • The Supply and Borrow numbers are the market sizes (at the time of post), in $ million, as a point of reference.
  • The even split between suppliers and borrowers is how the protocol functions; adjusting the ratio would require changing the code of the protocol. I completely agree with you that being able to fine-tune the split by market (e.g. 75/25 for collateral) would be very useful and the community should bookmark this as a future protocol improvement!

@mrhen It’s indeed important to properly define things, so we could clearly understand each other. In case of COMP farmers, as i define it it’s an activity of supply/borrow assets to Compound in that way, that it generates positive net flow achieved by selling distributed COMP tokens, measured in USD while minimising the volatility risks at same time.

Good example of that strategy will be, for example supplying DAI and borrowing DAI at same time. Normally that operation makes no economical sense, as borrow rate you will be paying will be higher than supply rate that you recieve. However, if you account for COMP distribution, you have positive cash flow. For small users it makes not much sense, as returns aren’t that huge, but if you throw bunch of millions, you arrive to quite a steady profits.

Of course, farmers are indeed users of Compound too, problem is that they not really bringing much value to protocol aside of high numbers, which could be called a sort of expensive advertisement. As if i bring for example 10 millions and borrow back 6-7 millions from that, i’ll draw high numbers on both sides, but actual liquidity i provide doing that is much smaller.

So, it’s not really about selling COMP or holding COMP, it’s about using Compound specifically for the purpose of capturing that distribution by activity, which isn’t profitable without that distribution.

Sold COMP to market isn’t really a problem for protocol. For every seller there is a buyer. COMP is just being redistributed from those who don’t want it, to those who want. There’s nothing wrong with that process. It doesn’t look like being huge impact on price valuation either.

I don’t see the need to “fight” or penalise farming, as well as i believe vesting, isn’t really that much needed aside of vesting for initial venture capital… And that’s not to discourage development of vesting. Any development is great, as others might see utility where i don’t see one. And certanly the more instruments protocol could utilise, the better.

I personally see vesting (and by that i mean basically delaying a COMP distribution by a certain period) as a proved inferior solution. I think it’s quite observable from other projects in DeFi, that “staking” works much better. In quotes is because while term is widely used in DeFi, it’s nothing to have with “proof of stake”. Basically usually it’s simple timelock of token in smart contract to collect more rewards.
That could be done in similar manner for COMP and it could recieve some portion of new distributed COMP if there is a need to provide incentive for holding COMP. That’s quite common workable solution in DeFi. Question here is what sort of benefits it can bring to protocol?

Frankly, i’m much more concerned by lack of reserves in pools in comparison with TVL. Preparations for rainy days are usually done far before rainy days actually arrive. But that’s another topic of discussion. :slight_smile: Here we talk about Comp distribution adjustment, which is quite a priority since auto-adjustment is removed already.

I agree with this reasoning generally. However, in the context of yield farming COMP to sell on exchange, Compound is punting on letting the protocol benefit economically from this sort of decentralization. For example, if the yield farmer is just going to flip the farmed COMP, why not instead have the protocol just sell COMP directly via DEXes or something of the like? That way the protocol could diversify its COMP to other tokens (through the COMP sale for another asset/token) while also decentralizing COMP governance to whoever purchases the sold COMP on DEX.

And so, if there was a policy we could enact to reduce COMP distributions to yield farmers while keeping reasonable COMP incentives for actual “real” users of the protocol, we could then route the saved COMP to exchange to still diversify ownership but also diversify the Compound protocol treasury.

1 Like

I’m with you on that also. I indeed think that it would be more beneficial for protocol to reduce emission of COMP by 10-20% and use it for direct market selling of COMP instead to seed a reserves for pools. And when reserves reach at least 1% of supply either decide on increasing reserves more, or direct funds to create treasury consisting of eth/wbtc/dai/usdc basket.

I already suggested that on several occasions, but to have discussion more focused i’d suggest deciding on one thing at a time. Yes, reserves are important, treasury is important. Yet, shaping initial frame for new COMP distribution is important too. It’s not very productive and more difficult for everybody to understand if we will try to fit everything in one single proposal.

I presented my suggestion about new distribution. Robert also suggested one. They are quite similar in numbers, but i believe mine is more clear about how numbers are delivered. Maybe you have some comments on suggestions on that? Might be you see a better model? Better distribution?

I’m sure if people would give more participation and supply their reasoning, something better can be shaped, rather than just waiting untill somebody create a proposal with what they think is the best, just because there is literally no other suggestions.

You can hardly blame VC that they make proposals with what they believe is better if nobody else really bothers.

1 Like

I hear you on that and apologize for not translating how ideas would percolate down to a distribution speed or a proposal that closely ties to the topic thread. That is good feedback. I’ll try to run some calculations in the coming days/weeks on how this overlay policy of Net Borrow Rate > Net Supply Rate would affect yours & other speed suggestions. I have multiple kids in diapers and no childcare right now, so haven’t had much free time to do the level of work I’d like to do on this. May take a bit longer to get back with firm numbers.

I should also re-note, I’m coming around to the idea that it would be fine for current Net Borrow Rate < Net Supply Rate, provided COMP vesting will happen in such a way that the user has to take a non-trivial price drift risk in the COMP component of the yield. It is likely a superior way to reduce farming because it avoids adding another rule to the COMP speed system that may require relatively frequent adjustments/votes (since COMP price and other protocol variables would likely affect the Rates too often).

1 Like

Correct me if I am wrong, but here exists opportunity cost for Ethereum (example). Eth has a low-interest rate and that can disincentivize Ethereum holdings on the protocol (because of staking yield). I think a high LTV ratio is not enough incentivizing for holding because of volatility and aggressive liquidation mechanism.
It would be more logical to reduce the incentive to stablecoins on the deposit and lending side because they have a high-interest rate on holding and borrowing anyway. Perhaps this would partially solve the problem with - “stablecoin leverage farming” in which I see no point other than fictitious TVL pumping.

I think this is interesting idea

In this industry (early stage) this is important use case for COMP token, but voting results on proposal show otherwise.

If you compare Aave vs Compound user adoption rate the data is worrying. Is that because airdrop on Coinbase Earn program? Active addresses with 3$ USDC?
Similar marketing tricks in the area of traditional business are less effective on decentralized protocols.