Fund Rebalancing

Mechanism

Avareum Fund Rebalancing

Following our principle, fund managers are allowed to manage funds using a set of restricted actions to prevent malicious actions. Rebalancing portfolios is one of the possible actions. The fund manager rebalances the portfolio by submitting a rebalancing request to the protocol. The request contains only the quantity of strategy-aware assets to buy and sell, including its corresponding collateral adjustment actions. After the fund manager has confirmed the rebalancing request, the smart contract begins the following process:

1. Verify basic qualifications, including but not limited to:

   1. Authentication and authorization

   2. Strategic actions e.g. to enter or exit strategy-aware assets

   3. Collateral adjustment actions, if applicable

2. Populate a list of executable actions from the strategic actions.

3. Execute the sequence of actions according to the sorted list. This should increase and decrease the exposure of each asset component to meet the target portfolio weight. If applicable, the collateral will also be simultaneously adjusted to meet the target level.

Note that the rebalancing operation is neither scheduled nor triggered by an on-chain event, but only responds to off-chain requests. The off-chain process, however, can be manual requests, schedulers, conditions, or trading robots and should be left to the decision of the fund manager. See Action Sequencing Guideline for more details.

Action Sequencing Guideline

The action sequencing guideline is a mere suggestion - not intended to be mandatory. Fund managers may reserve the rights to follow their own strategies.

1. Adjust the weight vector as follows. Let $w=[w_1, w_2,...,w_k, w_{k+1}, ..., w_n]$ be the target portfolio weight. We have $w_i > 0$ since long and short positions on the same underlying asset should differ in their indices thanks to the strategy-aware definition. Assume that $w_1, w_2, ..., w_{k}$ represent the weights of investible assets and $w_k, ..., w_{n}$ the weights of non-investible assets. Let $\kappa_i$ be the target collateral ratio for position $i$. Let $\kappa_i = 1$ for long positions and $\kappa_i > 1$ for short positions (over-collateral). The adjusted weight is given by: $w = \dfrac{1}{\sum_{i=1}^k \kappa_i w_i} [w_1, w_2,...,w_k, 0, ..., 0]$ Note that the notation for the adjusted weight above is not entirely formally correct. The variable $w$ is overloaded to avoid the overuse of symbols and shorten the subsequent expressions.

2. Populate a list of rebalancing actions and their associated parameters. Let $w'=[w'_1, w'_2,...,w'_k, w'_{k+1}, ..., w'_n]$ be the current portfolio weight. Let $\kappa'_i$ be the current collateral ratio for position $i$. Let $\kappa'_i = 1$ for long positions and $\kappa'_i > 1$ for short positions (over-collateral). Also, let $\text{nav}$ be the net asset value of the portfolio in the denomination unit.

   Calculate the following quantities:

   $\delta^{\mathtt{exposure}}_i = \mathtt{nav} \cdot (w_i - w'_i)$, representing the change in the exposure of asset $i$. $\delta^{\mathtt{collateral}}_i = \mathtt{nav} \cdot (\kappa_i w_i - \kappa'_iw'_i)$, representing the change in the collateral of asset $i$.

   $\Delta_i = \begin{cases} \delta^{\mathtt{exposure}}_i & \text{for long positions} \\ \delta^{\mathtt{collateral}}_i - \delta^{\mathtt{exposure}}_i & \text{for short positions} \end{cases}$

   Or alternatively and conservatively,

   $\Delta_i = \begin{cases} \delta^{\mathtt{exposure}}_i & \text{for long positions} \\ \delta^{\mathtt{collateral}}_i - \min\{ \delta^{\mathtt{exposure}}_i, 0\} & \text{for short positions} \end{cases}$

   Then, to avoid friction cost due to marginal adjustment, add the strategy action to the list only if any of the following conditions hold for asset $i$:

   • $|\delta^{\mathtt{exposure}}_i| > \epsilon^{\mathtt{exposure}}$

   • $|\delta^{\mathtt{collateral}}_i| > \epsilon^{\mathtt{collateral}}$

   • $|\Delta_i| > \epsilon$

   where $\epsilon^{\mathtt{exposure}}, \epsilon^{\mathtt{collateral}}, \epsilon > 0$ are tolerance thresholds for each quantity.

3. Sort the actions in the list as follows. Unless otherwise stated, the order within the same group does not matter.

   1. The group of exit actions for long positions. These actions liquidate assets into the denomination asset to be used to in the subsequent actions. It is, therefore, essential to get executed first.

   2. The group of actions with $\Delta_i < 0$. If no additional constraints, e.g. locked-up periods, are applied, these actions add to the portfolio the nomination asset, e.g. cash, to be used in the subsequent actions. These actions include but not limit to short selling assets or reducing collateral. It is, therefore, essential to get executed second.

      Among these actions, they are sorted by their change in exposure $\delta^{\mathtt{exposure}}_i$ in a descending order i.e. from positive to negative. The higher $\delta^{\mathtt{exposure}}_i$, the earlier it gets executed.

   3. The group of actions with $\Delta_i \ge 0$. These actions convert the nomination asset into strategy-aware assets and their collateral if applicable. These actions include but not limit to buying assets or increasing collateral. It is, therefore, essential to get executed last.

      Among these actions, they are sorted by their absolute change in exposure $|\delta^{\mathtt{exposure}}_i|$ in a descending order i.e. from positive to zero. The higher $|\delta^{\mathtt{exposure}}_i|$, the earlier it gets executed.