The usual concept about algorithmic stablecoins states that it must sustain the market peg everytime. If the stablecoin depegs on market, then it is not working fast enough. However, this concept is wrong: in extreme situations it not only yield to smart contract collateral syphoning but, as I will demonstrate, it is not even necessary.
What we must understand before anything else, is that market depeg reflects panic of many market agents. Many times this panic is not rational. And if the swaps acts very fast, as we have seen in practice, then it consumes BR, lowers waves price, feeds more panic, which in turn feeds more arbitrage in a vicious circle. So not only don’t solve the market depeg but also yields to a spiral of smart contract syphoning and collateral price. And it is very vulnerable to market manipulation.
Important question here is: should a stablecoin algorithm try to solve market depeg as fast as it can? Answer is no. That approach not only has the effect described above, but also it is not necessary at all. What a stablecoing algorithm must do is to ensure than sooner or later, depeg will be restablished, so those who panicked loose money because of selling cheap, and those who standed without panic don’t. Evenmore, those who bought cheap from weak hands will have a profit at expenses of the weak hand.
In this way, the smart contract don’t work by attacking depeg fast, but instead, by punishing the irrational actions and rewarding the rational one. In the long term, as more people understand that peg is always restablished, the panic will be less and stablecoin will be even more stable on market.
Many proposals has been advanced in order to increase collateralization, and some ideas are good and probably necessary, others are bad. But both have in common that they don’t attack syphoning, so the final result will be that, no matter how much new collateral we add by other means, syphoning will always be faster, thus only delaying the final result at the futile expense of rewards from other parts of the system.
So, how can we improve the neutrino algorithm? By adding increasing delay to burn swaps as burn swap volume increases. It will not act immediately on market depeg, it may yield to bigger temporal market depegs, but in the long term will be much more solid and safer. Remember key point here: the algorithm must not try to attack fast and directly the depeg, as this frequently has the opposite effect, but instead punish those who sell cheap USDN.
How much delay must be added? Probably, enough that in extreme situations, raise to 1 week for instance (no randomization, as we want only to slow down). Also, if one swap is in process, the user cannot swap again until completion of current one, regardless it may have passed 24 hrs.
An additional complexity can be added for further tunning: swaps may have an option to perform swap either at the price in the initiation of the swap, or in the completion of the swap, and the user will choose one of them. If at the initiation of the swap, it is more suitable for acquiring waves to hold. If at the completion, it is more suitable for acquiring waves for selling on market and arbing. In the first case, the swap time need to be equal or even bigger than in the second case. The purpose of this additional feature is to avoid in stress situations, to discourage swaps for waves accumulation, because if swap is performed always at the price at completion time, the risks for waves potential holders increase. Notice we don’t want to punish arbing, because it is necessary for restablishing peg. We only want to slow down it in order to avoid syphoning vicious circle.