How Virtual Pools Work

A Virtual Pool can be composed from two native AMM pools if they share one common asset, and also if at least one of the pools accepts the non-common asset from the other pool as a reserve.

For example, if native pairs between A-B and B-C exist, and the A-B pool can accept C as a reserve token, a virtual pool A-C can be created, capable of performing C-A trades.

To reflect the VirtuSwap Whitepaper nomenclature, the first of the two pools composing the Virtual Pool is called jkPool or Active Pool and the second pool is called ikPool or Reference Pool.

When a swap occurs, the Active Pool (A-B in the example above) performs the actual exchange of the tokens by accepting a certain amount of token C to its reserves, and paying out a corresponding amount of token A. The Reference Pool (B-C in the example) is used for constructing a virtual AMM curve between A and C, which is used to calculate the amount of A that should be swapped for a given amount of C.

The liquidity in the Virtual Pool is calculated as follows:

$$LiqA_V = LiqA_{AB}*\frac{MIN(LiqB_{AB}, LiqB_{BC})}{LiqB_{AB}}$$

$$LiqC_V = LiqC_{BC}*\frac{MIN(LiqB_{AB}, LiqB_{BC})}{LiqB_{BC}}$$

As can seen from the formulas above, the total value of the two balances of the Virtual Pool equals the TVL of the smaller one of the two pools.

After a swap is completed, the Active Pool A-B ends up with a lower amount of Token A and a corresponding amount of C in reserves. Since token B is not participating in the swap, the pool's imbalance (and thus arbitrage opportunity) is much lower than in the case of a regular trade. The reserve token C is to be exchanged at a later point in time for A or B with other pools in the system.