[Proposal] Algorithmic Treasury


The algorithmic treasury is an addition to the Dollar Protocol. in order to avoid the death spiral during negative rebase, especially when the protocol is young and not widely used.

It is an evolution of the current Treasury contract.

The Algorithmic Treasury should work alone and no governance work is needed.

The Treasury

The Treasury is a smart contract that operate on a set of reserves

Operations that should be possible in the Treasury are (can be called by anyone):

  • Sell USD to buy the reserve currency If USD price is above the sellLimit, the Treasury sell some USD tokens to buy reserve up to the sellLimit price. Nothing is done if there if no USD tokens are in the contract.
  • Sell reserve currency to buy USD If USD price is below the buyLimit, the Treasury sell some reserve tokens to buy USD up to the buyLimit price. Nothing is done if there is no reserve tokens
  • Sell reserve currency to buy&burn SHARE If USD price is above $1 and the reserve market cap is above the USD markey cap * reserveLimit then SHARE tokens are bought with the reserve tokens.

Positive Rebase

On a positive rebase, all newly minted USD will be deposited to the Treasury that will sell them to decrease the price of USD and accumulate reserves. Notice that if too much reserve are accumulated, it will be used to burn SHARE tokens (increasing the price of SHARE).

While SHARE holders will no longer have the pleasure to find new USD tokens in their wallet, this operation create the same wealth for SHARE holders and is more effective. For instance it avoid USD token to be trapped in smart contracts (for instance the ETH-SHARE Uniswap LP, but almost anytime you stake SHARE). This remove the aquaponic mining potential of SHARE, reducing its value. removing USD distribution increase therefore the SHARE value (most likely at some marketing cost).

One might say that if USD is conserved in the reserve and not used to burn SHARE tokens, it is at the expense of SHARE holders. But the reserve is owned indirectly by SHARE holders so there is not a huge difference. Moreover, those reserves reduce highly SHARE dilution on negative rebase.

Most valuation methods use business value + treasury value to value a company, it would be the same here. The stock market is a very good example (the stock price decrease by the amount of the dividend when it’s given to shareholder).

Negative Rebase

In the current system, USD tokens are burnt and new SHARE token are minted, diluting SHARE holders. If the treasury has reserve it will avoid this process because USD buying will take place at a higher price point than USD burning.

When the treasury has no more reserve, the USD burning and SHARE minting resume as USD price fall below the process threshold.

The treasury act as a first level USD price protection to avoid SHARE dilution.

Those bought USD will be converted into reserves when USD price gets above sellLimit.

Those USD tokens might be considered to be outside of the circulating supply. It would be the same if we just burn them (maybe we should?).


I suggest the reserve token to be ETH. It is censorship resistant and doesn’t have any protocol risk. It make sense while the Dollar Protocol is only on Ethereum. The issue is that the price can fall and reduce our negative rebase insurance.

Other parameters:

  • sellLimit: Price when USD is sold on the market for reserve tokens. I suggest to start with $1.05 then reduce it progressively to $1.01.
  • buyLimit: Price when USD is bought on the market using reserve tokens. I suggest to start with $0.95 then increase it progressively to $0.99 when USD become mainstream.
  • reserveLimit: The maximum ratio of the reserve versus the USD market cap. Above that, reserve is used to burn SHARE. I suggest to start with 100% to avoid any death spiral at the beginning of the protocol (except if the reserve currency depreciate). Then reduce it progressively to 10% when there is wider acceptance.

Open questions

Managing multiple currencies

In order to keep thing simple and stay antifragile, an algorithmic treasury will be instantiated per currency.

It would always be possible to transfer reserve token between treasuries if needed/wanted by governance.

Incentivizing actions

We might want to incentivize call to the the treasury fonctions. Alpha Homora gives 3% of the profit to the caller. It would work best is the rebase on the usd contract is also something that can be called anytime.

Limit USD burning to the Algorithmic Treasury

As an addition, we might want to allow only the Algorithmic Treasury to be able to burn USD for discounted SHARE (and buyback USD).

The Treasury can always mint USD so it is always possible. The ending USD position can be kept or burned (same impact)

It would have those following pros:

  • The SHARE dilution will be the same for every SHARE holders. No one can exploit a great discount to increase its position at the expense of other SHARE holders. Basically, we no longer care if the discount is too big for any reason. The arbitrager no longer keep a huge profit.
  • It is more efficient as the arbitrage profit is not sold by the Treasury for ETH (or whatever outside money some arbitrageurs can use). Therefore, for the same SHARE dilution we have a better USD price increase.

Wouldn’t the ability for the Treasury to mint SHARE and sell them for USD be the same but easier to understand (and no discount issue)?

PS: Overall, it seems that the whole process can be simplified by merging the Treasury and the rebase (if reserveLimit = 0 it would work the same).


Thanks for taking the time to write all this out. It’s very well written and structured.


"On a positive rebase, all newly minted USD will be deposited to the Treasury that will sell them to decrease the price of USD and accumulate reserves.

While SHARE holders will no longer have the pleasure to find new USD tokens in their wallet, this operation create the same wealth for SHARE holders and is more effective. "

This is a major turn off. It literally removes pleasure of experiencing the positive rebase. I doubt you will attract many new people to join Dollars/Shares program.

Again, this also does not prevent front running and fancy pump and dump schemes. The treasury will always get the worst price.

If implemented, I promise to front run the treasury until I’m satisfied w/my profits and leave the ecosystem. I will continue to exploit this.

You may as well call this Communism Protocol.

From the whole project point of view, this algrithmic treasurey is good idea to make dollar protocol towards to be stable coin as its vision.
But from new comers or investors point of view, what incentives for them to join this system? Holding shares to gain value looks not easy and takes (long )time.
Better consider seriously the condition of sell reserve to buy & burn share, it is USD>$1 and reserve mcap more than USD mcap, or else?

1 Like

Good point. I’m bullish on ETH so I think there will be some burn even with 100% reserve limit. Selling usd at $1.05 provide some burn potential (0.05$ of SHARE every USD sold).

Personally, with a dashboard showing the net asset value of a SHARE token (increasing at each positive rebase and each SHARE burn) and SHARE being burn, I would be quite happy. We can have something like https://makerburn.com/#/.

When there will be $10 of ETH inside each SHARE and a projection that 10% of SHARE will be burn in the next 12 months, I will be buyer if the SHARE price is $10.

You will also get new SHARE if you stake ETH-SHARE LP.

I fully understand that it’s less fun. But it’s even less fun to be in a death spiral because we haven’t put aside for bad times (okay it would not have work due to the bug anyway).

Ultimately, the point I have consistently been arguing is the protocol works. The mechanics work. It just needs some refinement here.

I quote Yunus:

"what happened to discount rate today? it was 13% with 39 usd left to burn. today its 5% again. i was hoping it to be 18%. "

The incentives work. It’s just a matter of refining the execution and speeding up the system so it recovers faster.

There is no need to overcomplicate things and add more tools to a system that IS working.

Interesting. How does this differ from https://reserve.org/protocol ?

Reserve works like Meta. A bunch of stablecoin with an insurance. This is not censorship resistant and neither profitable (they made a small spread, but unsure what parameters are).

The proposition here to use something strictly superior as reserve (ETH which value should increase versus stablecoins) and not having 100% reserves as a guarantee (which is needed as ETH can fluctuate a lot). This is more an insurance policy for SHARE holders not to be too much diluted.

RSR holders are diluted as soon as the collateral ratio is below 100%. It’s not the idea here. That allow to have more powerful assets in the reserve.

I also think SHARE/USD has way less governance than Reserve.


Looks very interesting.