Editor's Note: This article comes fromSerum Information Platform, published with permission.

Automated Market Maker (AMM) technology has played an important role in this DeFi explosion. Different from the market maker mechanism of traditional exchanges, AMM encourages users to put their own tokens into the trading pair liquidity pool to form a small market maker. Are there any disadvantages of AMM? How can it be improved? (The following article is translated from FTX and Alameda Research founder SBF’s tweet on improving AMM on October 15)

secondary title

"Can the AMM protocol be improved?"

Disclaimer in advance, the following views are not investment advice, nor market maker advice.

The discussion of this question is a bit strange: AMM has allowed liquidity providers to earn millions of dollars this summer, how can there be problems with AMM?

OK, let's sort this out slowly.

The first question is, why did AMMs exist in the first place?

That's because most of the blockchain transaction throughput (Throughtput) is not high, and the delay of the blockchain makes the decentralized exchange (DEX) unable to support the transaction order book, so DEX is forced to use the AMM protocol.

So how are AMMs doing? AMM wasn't doing well until this summer.

Although AMM seems to be operating very efficiently now, this efficiency is not completely "naturally formed".

Most of the DeFi transaction volume and lock-up volume (TVL) are due to Yield Farming. In any case, this form is equivalent to the project airdropping its own tokens to users. No matter how you think about liquidity mining, these statistical data are actually due to the strong incentives of the project party.

Now everyone is using the AMM protocol, and the pledged tokens are in the liquidity pool because someone paid them to do so.

But this approach is not a patent of the AMM protocol. Everyone can use the same method to transfer profits to market makers, users who mortgage tokens, or anyone else who wants to reward.

And once these rewards are withdrawn, how much do you think the trading volume and lockup volume will be left? Not sure, and no one knows.

Having said that, most people have discovered that there is a big problem with AMM, which is "Impermanent Loss".

What is impermanent loss? The impermanent loss means that if you provide liquidity in the trading pair pool and the token price changes, you will also lose accordingly. The reason why it is called impermanence is that whether the liquidity provider loses or not depends on the price of the token. If you continue to provide liquidity, as long as the token price recovers, the value you lost will naturally come back.

Now many projects are trying to improve the impermanence loss of AMM. Some have improved the slippage curve, some use insurance, futures, and hedging as solutions to impermanent losses, and more projects choose airdrop rewards.

Do they really solve the problem of impermanence loss? No.

secondary title

Why is it not resolved?

That's because impermanence loss is not a "wrong parameter", just adjust it back. The impermanent loss is just a euphemism, the vernacular is: "bad trade selection".

For example, suppose you staked 1 ETH and 400 USDC in the AMM, and the current ETH price is 400 USD. Assuming that the price of AMM’s taker and pending orders is 30 basis points (30 bps, about 0.3%), then if you mortgage your assets at $400, it is tantamount to announcing to the world:

"Hey! Anyone want to buy at $401.20 or sell at $398.8? If yes, come on!"

So what will happen? Well, you just continue to sit and wait, continue to wait until someone sells it to you at a price of 398.8 when ETH drops by 60 basis points, because this is the market price.

You who provide liquidity are like a duck sitting there, facing a total of 60 basis points of two-way risk, and you must withdraw liquidity (assets) if you want to cancel the order, but the delay of the blockchain makes this Not only is the action slow to execute, but the handling fee is also high. So you can only wait for the market to trade with you when the market moves more than 30bp.

At this time, the market price is not good for you, and then someone will sell it to you at a price you don't want. This is "impermanent loss".

In the normal trading order book, everyone’s order price is the price they really want to buy or sell; but in the AMM agreement, everyone is in the liquidity pool, so both sides of the token are forced to buy or sell at the market price , regardless of the price, these prices do not represent any opinion, so it looks like a group of ducks sitting in the trading pool.

Hedging risk solutions such as futures and options cannot help. Because the core problem of AMM is not "risk", but a bad deal with a negative return expectation.

Rewarding transaction fees can't improve this bad deal, it can only balance the loss and make the loss lower.

Changing the curve for slippage doesn't help either.

This means that any "sane" order maker, that is, a trader who just wants to buy tokens at a reasonable price, must pay a fee of 30 basis points, which is a very inefficient way of trading.

secondary title

What if you use an algorithm to set the price?

There are three possible situations:

1. The data of this algorithm is taken from the oracle machine of the centralized exchange. Then traders might as well go directly to the centralized exchange to trade.

2. Algorithm data is taken from on-chain data 1> Oracle data comes from decentralized exchanges, and its price algorithm is the same as DEX, which belongs to recursive logic

3. It is also possible to use other algorithms than the above two, usually these are garbage algorithms

Therefore, the AMM provider will inevitably have the following situations:

A. The liquidity provider is conducting a bad transaction, constantly suffering impermanent losses without knowing it

B. The volatility of the token is very low, the impermanence loss is close to zero, but the reward fee is also close to zero, just like Curve

C. Too many random and irrational order takers enter the market, resulting in profit > impermanent loss, but the efficiency is still lower than that of general exchanges with order books.

or

D. Someone subsidizes you to use AMM, such as liquidity mining

So as long as the liquidity mining is over, AMM will be returned to its original shape.

secondary title

So what if you just want liquidity?

Yes, this is perfectly reasonable and the clearest way for AMMs to do it.

You have some tokens, you want to trade it, you don’t want to be a market maker (MM), so stake your tokens into an AMM and forget about it.

Sure, you lose some money, but that doesn't matter because that's the price of providing liquidity.

But what if you want to write a "smart" AMM protocol, such as pending order transactions?

But it still has to be said that AMM can demonstrate its value in some applications, such as the exchange of stable coins with low volatility, or new projects that require token liquidity.

AMM forces you to take bilateral risks at the same time, which is obviously an ineffective trading method, even if you improve the algorithm of slippage, hedge risks at the same time, or other methods are useless.

Now just because of the rise of liquidity mining, we suddenly forget that AMM was actually a very inefficient trading method before, and how unpopular it was with the market.

But it still has to be said that AMM can demonstrate its value in some applications, such as the exchange of stable coins with low volatility, or new projects that require token liquidity.

You can't solve the AMM problem at all, you can only make them less difficult to use.

You can't solve the AMM problem at all, you can only make them less difficult to use.