First, what's a vAMM?

A Virtual Automated Market Maker (vAMM) is just as customizable as a spot AMM, except the underlying is a "virtual" contract cleared between parties (perps!), not a fungible token. So there is nothing exchanged upfront, just future settlement.

Why is this an attractive mechanism?

Because it allows a DeFi protocol to have a built-in designated market maker. Designated market makers provide guaranteed liquidity, which helps to bootstrap markets and makes trading / liquidations safer.

So what's the FUD?

The problem is that some basic vAMM designs offer too much leverage. In fact, a basic pricing rule could be increasing leverage. The long-short imbalance is implicitly owned by the protocol itself. You can calculate vAMM leverage = imbalance OI/funds.

What designs make a good vAMM?

Limiting vAMM liquidity at attractive prices should be a function of collateralization. Better pricing rule can be built to include when user longs > user shorts, longs are paying funding, and the best bid/ask price vs oracle skew higher. Lots in this design space.

so what is the REAL problem from the FUD?

So this is less a vAMM problem and more a leverage/accounting problem. Exchanges with orderbook matching must account for collateral imbalance of longs and shorts. You must solve for what happens to long PnL if prices spike higher vs levered shorts? They can't collect gains from bankrupt users.

How can you solve the p&l accounting problem?

One user's settled profit should always be another user's settled losses. This should be true for both orderbooks and vAMMs. P&L accounting is completely unrelated to the matching engine. Orderbooks can suffer from levered losses as well.

Any good vAMMs out there?

Drift v2 vAMM has been running for almost 7 months. using constant product curve still BUT updating its reserve PRIOR to trades based on the oracle price (and other MM signals). Using x * y = k can still work great as a price impact function.

What is Drift's vAMM thesis?

The thesis is "why pay an external designated market maker who would leave tomorrow when you could build an advanced MM strategy into the protocol itself". Designated makers typically get special treatment/privilege by an exchange for consistently providing liquidity.

Designated makers earn this privilege by risking toxic order flow. There are numerous ways to prevent latency arb, compensate for toxicity, and even "unload" that have been built into Drift protocol as well.

Final thoughts?

TLDR - don't conflate good vAMM with bad vAMM. I think the design space is HUGE and very under-appreciated given the historic failures. I'm always happy to discuss / share thoughts with other vAMM designers (and anyone looking to understand deeper!)