A quick update/highlight on the on-going research carried on by @vasilysumanov and TE team.
(please see here Proposal 11: Issue a grant to co-fund research on Balancer AMM with Balancer Labs)
While the job the guys are working on is pretty extensive and fundamental, here is one of the practical implications of it, which will not only improve the design of the existing and upcoming indexes, but will likely allow creating a brand new type of index mechanics.
So using the Balancer model @vasilysumanov will provide, we are analyzing different types of index mechanics including the most obvious - an index with dynamic weight changing.
Below I am speaking only about the abovementioned sub-task as this is something we need pretty urgently (while @vasilysumanov’s input will bring us a set of instruments to carry on a way more extensive research)
Main goal
Produce an optimal design of an index with dynamically changing weights - a product for ETF investors, which would (i) automatically re-balance based on selected market metrics (ii) be save from attacks, (iii) be cost efficient while outperform other possible configurations of itself (i.e has the optional re-balance period), (iiii) fastest/cheapest launch, (v) preferably to use PP oracle
Task: Determine and data-proof the optimal configuration of the index (i-v above) with the best performance/yields/risk metrics
To Do:
- estimate the index performance for an extensive number of simulation scenarios
- estimate the gas costs for each case
- estimate the oracle costs for each case
- offset the key risks of attack
Methodology:
- We are using the 15 second price changes quotes for 365 day period
- For simplicity reasons initial weighting mechanics is based only on MCaps
- For the scenarios - we use both real price data and dummy prices to run a pretty large number of simulations of totally different market dynamics (e.g. all tokens are growing/dumping, part of tokens are growing / part of tokens stay flat or dump etc.)
- Conservative gas expenses (i.e gas is expensive) for each scenario
→ the results are to be benchmarked to select the optimal design/mechanics for the index