Guardrails & Impermanent Loss Mitigation

1. Guardrails & LP Safeguards

Mechanics Goals
The goals can be generally defined as follows:
1. Make the LPs whole (QTY): Hold sufficient quantities of the deployed assets in the system reserve
2. Maintain or increase the POA (system reserve): Pull operational surpluses and system revenue into the POA

1.1 The Asset Stack & Mitigation Waterfall

For all examples in this document a mitigation for a relative change in the exchange rate of up to 100% between the two paired assets will be demonstrated.
If a certain threshold (as defined below) of net-withdrawals by the LPs of a particular reactor is reached, mitigation mechanics become necessary in order to make the LPs whole while simultaneously maintaining or increasing the POA:
  1. 1.
    Make LPs whole
    By drawing the asset in deficit from assets in reserve
  2. 2.
    Assets in Surplus / System Revenue
    Draw system-wide asset surpluses and system revenue into the POA. This will cover a large portion of the negative asset flow. The amount of surpluses that can be pulled into the reserve also depends on the net-withdrawal requests.
  3. 3.
    TOKE staked
    TOKE that was staked to direct the asset (whether Pair Reactor or Token Reactor staked TOKE) are used to cure and pulled into the POA. There is a difference in how staked TOKE are being used depending on which side of the pool the deficit occurred (asset vs base asset).
  4. 4.
    Protocol Owned Assets
    In a last step, should the above steps not be sufficient to make users whole, the system will resort to:
    • Using ETH and/or stable coins from the POA to make the LPs whole.
    • Should not enough ETH or stable coins be available, highly liquid reserve assets are sold for ETH or stables on external venues.
    This last step would be performed without regard to a net-negative effect on the POA. As previously mentioned, this should not occur under circumstances within the range defined by the guardrails. It will be initiated by the DAO multi-sig and executed using best practices in order to mitigate front running and mitigate other manipulations. This is highly unlikely and we don't anticipate this stage will ever be reached.

1.2 Deployment Guardrails

General Definition:
Deployment guardrails are restrictions imposed on the protocol limiting the maximum amount of assets deployed per individual Token Reactor (asset pairs) and deployment cycle. The guardrails allow for managing market risk by ensuring that the mitigation mechanics are effective within predetermined market conditions.
Defining and Setting Guardrail Parameters
In order to define and set appropriate parameters for the deployment guardrails the range of market conditions (exchange rates) under which the changes in quantity of the deployed assets are to be mitigated has to be defined and potential parameters have to be identified.
More refined metrics (such as volatility) will be implemented soon for asset pairs, initially the goal is to set parameters conservatively such that relative changes in exchange rates of up to 100% can be covered by the reserve. Below is an example of a fictitious asset paired with ETH and deployed to a 50:50 AMM pool under such conditions:
Table 1. Status of the pooled assets at the beginning
After a 100% price increase of ABC the status of the pool is as follows:
Table 2. Status of the pooled assets upon withdrawal
Examining the changes in QTY and associated notional values produces the following results:
Table 3. Changes in quantity and resulting loss in notional value
The following can be observed in a scenario of a 100% (doubling or halving of value of one of the assets) relative change in price:
  1. 1.
    A QTY-deficit of 29.3% in ABC will have to be covered by assets from the ABC reserve to make the LPs whole.
  2. 2.
    A net withdrawal of over 70.7% of ABC would be required before mitigation mechanics would become necessary.
In conclusion it can be inferred that the protocol will be able to make LPs whole after a relative change in price of 100% as long as no more than 3x (as 29.3% of the initial qty will have to be covered) the reserve QTY held in the POA of an asset are deployed.
It should also be noted that even stricter, gradually lowering guardrails will be implemented in the beginning stages of the protocol, which will provide additional mitigation against other potential issues.
Guardrails to be Implemented
In accordance with the above definitions the guardrails will be controlled via parameters that can be altered by changing variables, which as the protocol matures will be adapted to match asset pair-specific risks, market conditions and other factors such as overall POA value and DeFi-landscape (e.g. new protocols that can be leveraged).
For XXX / YYY pool, deploy only the liquidity that obeys the following rules (use most restrictive):
  1. 1.
    Deploy only V-multiple of the quantity (leverage) of XXX available in the POA Deploy only Z-multiple of the quantity (leverage) of YYY available in POA
  2. 2.
    Deploy only M% of XXX contributed to the reactor by LPs Deploy only N% of YYY contributed to the reactor by LPs (this guardrail which is not based on any above calculate percentages will be loosened over time, as its primary function is to ensure safety during the first deployments of liquidity.)

1.3 TOKE Backstop

In order to further safeguard the reactor and its LPs, TOKE staked to reactor can be used to cure the deficit if the price moves outside of the aforementioned assumptions.

1.4 Asset Stack and Reactor Health

The below graphic illustrates the asset stack at the beginning of the cycle, from which the reactor health (RH) and the health of Pair Reactors can be calculated using the quantities:
RH=AssetPOA+AssetLPtAssetDeployedRH = \frac {Asset_{POA}+Asset_{LP}}{tAsset_{Deployed}}
According to a 3x POA Multiplier the RH is 1.33 at beginning of cycles and can increase and decrease depending on the market price of the asset.
Additionally, the total collateral (TC) of the reactor can be calculated using the notional values, which includes the TOKE staked to the reactor:
TC=AssetPOA+AssetLP+TOKEtAssetDeployedTC = \frac {Asset_{POA}+Asset_{LP} + TOKE}{tAsset_{Deployed}}

1.5 Surplus / Deficit Balancing

At the end of every cycle the reactors will rebalance. In the case that there are more LP ABC than tABC, the surplus ABC is flowed into the POA. In the case there are less LP ABC than tABC, ABC from the POA is flowed out into the LP.
Using these mechanics, each cycle Tokemak is flowing out the same amount of LP ABC as there are tABC deposited into the system (while adhering to the above guardrails).