Deep dive into game theory of Synthetix
Synthetix is a decentralized platform and exchange for synthetic assets on Ethereum. Read the litepaper and research if you are not familiar with the project. This article is aimed at advanced readers.
What’s going on with Synthetix? We published our research in November and soon after, the project became a DeFi success story of 2019. The platform grew in volume, price of SNX appreciated and stakers were earning very motivating rewards.
Yet, in the face of recent (2020) bullish price action across crypto markets SNX loses steam and looks extremely undervalued comparatively to other projects.
If you want to read more on how this metric can apply to tokens, I recommend this Bankless article, from which these charts are stolen.
Why is that? Is market just stupidly overlooking it or is something else going on?
First, let’s not kid ourselves about efficiency of crypto market. Hopium still rules over fundamental analysis. Plus SNX had a strong run in 2019 and now looks exhausted despite everything else booming.
But I think even this market would price SNX better if all was shiny. And when you look closer you will realize that indeed all is not shiny in the Land of Synths. In fact a battle is raging. But who is battling whom?
First front-running war
There are ongoing issues with front-running on Synthetix. The first case of front-running on the platform was pretty dramatic. But it’s nothing unparalleled in crypto space. Many DEXes suffer from it. Front-runners see order transactions propagated on-chain and try to over-pay for gas in order to jump their orders in the queue.
With Synthetix it’s not so simple though. While the project already solved the issue of on-chain front-running by introducing gas limit cap, now it faces a different problem.
Front-running war II — doom of oracles
Trading synths doesn’t happen via peer-to-peer order matching like in majority of other DEX. It’s peer-to-contract = traders do not exchange anything, they just swap one price oracle for other oracle and switch assets against the debt pool. The debt pool of synths is backed by SNX tokens staked by minters.
However, oracle based price feeds are still on-chain. It’s not feasible to update oracle prices fast enough, so a latency occurs between price of synths sourced from oracles and spot prices on other exchanges.
Front-runners track or sometimes manipulate the price of an asset on spot and then trade it on Synthetix before the oracle updates the price. Essentially creating an arbitrage. It is particularly problematic with less liquid assets and during higher volatility periods.
Oracle updating price feed is not always fast enough. Synthetix is in the process of outsourcing it’s price feeds to Chainlink oracles.
If front-runners are quicker than oracles and make a profitable arbitrage, who pays the bill? SNX minters! Because of the way Synthetix liquidity is structured minters are collectively on the other side of all trades done by synth traders. So in a way, every synth trader’s win is minters loss.
Once again SNX community rallied to action against front-runners. To tackle oracle front-running, team decided first to increase trading fee from 0.3% to 0.5%. This made some edge cases unprofitable but many opportunities remained profitable, especially on volatile assets.
Increasing fees is just a temporary solution too. And potentially harmful as it doesn’t only affect front-runners. It affects all traders negatively, making the platform more expensive and uncompetitive.
Since Ethereum blockchain is public, we can see addresses of major frontrunners. Punishing or blacklisting those actors specifically would, however, lead to a bad precedent. And lets keep in mind that we aim to build open public protocols that don’t need gatekeepers and legal protection, but rely only on balanced design of crypto-economic incentives.
Here is the main culprit of front-running. Source
But sometimes ideologies are not enough. Sometimes we need to get pragmatic. SIP 12 proposed by a community member pondered on options to hot-fix the oracle front running, before the real solution is implemented. The best option, disabling the most exploited synths for XTZ, LTC and BNB, was implemented in short order.
“Delisting” synths overnight without further notice is far from ideal. It also creates a bad press. Some honest traders positions were closed, leading to missing out on profit from further rally of assets like recently booming XTZ.
Furthermore, the mere mentioning of slashing individual addresses in proposal (despite being rejected) is kinda alarming. It hints that with large enough consensus (or single minded decision from the core team) it could happen.
But there is a light at the end of the tunnel — with each attack the protocol evolves. Synthetix aims to provide a solution to oracle front-running with a waiting period (proposed in a SIP 37). This should, after some tweaks, be enough to put front-runners in check, while enabling honest traders to use the exchange without major drag.
Front-running traders are not the only one that can hurt the protocol. Even SNX stakers may act maliciously to exploit the system.
Dishonest SNX stakers could perform a “snapshotting” exploit, where they earned rewards without adding value to the network.
Current reward claiming model allows SNX stakers to essentially become a free lunchers by staking SNX just before the claiming day, reap rewards and then unstake and sell their SNX, not providing any value to the network. This leaves honest stakers with much less rewards and more risk.
The solution is under way, tweaking the reward claim rules as outlined in SIP 9. Same as with front-runners, the ultimate goal is not to identify and punish misbehaving actors but to change the game theoretical rules of the system in a way that makes the honest participation more profitable than exploiting.
The aim is to fix the future, not redeem the past. But speaking of past, can we at least estimate how big was the damage?
Volume speaks volumes
Probably the most important metric is volume, since fees are calculated from it. Bigger volume = more fees = more profits. At least in theory. But remember front-runners and how fees were raised to deter them? They are in profit only if the price arbitrage opportunity is bigger than the fee they need to pay. So volume generated by front-runners means net loss for SNX stakers, despite the fees.
By analyzing trades on chain it should be possible to gauge how much of the overall Synthetix DEX volume is “organic” (meaning honest traders taking synth positions), how much is simple price arbitrage and how much is oracle fron-trunning. But I leave that to someone else. Let’s just do a rough visual estimate:
It’s good to note here that not every arbitrage is front-running and thus malicious. But to be safe we may assume that at least 80% of overall volume is front-running, which means that SNX stakers are not profitting from 80% of volume generated, but incurring a loss! (I edited this part — originally I was more generous with 50%, but looking at volume by main culprit, it seems it is much more in last couple of weeks).
This loss reflects itself indirectly. SNX stakers still earn the fees. But their collateral ratio falls below 750% and they are forced to burn their synths or add capital to fix it.
Synthetic Cold War
We can observe two distinct classes of Synthetix users — SNX stakers and synth traders. While any entity can do both — mint synths and trade them, in general there should be much less SNX stakers than it is traders, since staking and minting is pretty complex and carries risks, that not everyone wants to take.
We have seen that both groups contain those, who are trying to exploit the system to maximum, but even in an exploit-less system, there is an incentive gap between these two groups.
It is a bit similar to a certain kind of CFD and FX brokers called Market Maker brokers. Instead of pairing the orders via liquidity providers, these MM broker is taking the opposite positions of the trader. The Market Maker then should hedge the positions separately. But naturally MM brokers have a conflict of interest against traders. They are incentivized to provide suboptimal price feeds, to be “liberal” with stop-loss triggers etc.
Luckily shenanigans like these are not quite possible in an open connected world of DeFi where Synthetix operates. But the conflict of interest remains. SNX stakers do not want traders to be winning in aggregate. But traders are their clients so SNX stakers shouldn’t want them to lose too quickly either. Synthetix CEO Kain addressed this in his blogpost, calling it trader/staker dichotomy.
SNX stakers should be incentivized to care for long term sustainability of the business. But for that we need to align staking incentives for long term AND prevent traders from exploiting the system.
What can we do?
Remember how we were pondering why SNX seems to be so undervalued? I humbly suggest these are the main reasons, not the blindness of the market.
To recap: When you stake SNX tokens, you are not simply earning trading fees. You are taking the other side of traders synth positions and you are paying the profits of front-runners and other arbitrageurs. As an honest staker, you also share your rewards with hit and run stakers who are exploiting the current reward system. This is not priced in the Mcap/Earnings ratio nor does it reflect in aggregate numbers like total volume or fees generated.
All being considered, I would propose to multiply current Mcap/Earnings ratio by factor of 6, getting to around 36. Which is still great for such a quickly growing project, compared to others. It’s just important to realize that rough aggregate numbers are deceiving. (I edited this part — originally I was more generous with factor of 3, but looking at volume by main culprit, it seems it is much more in last couple of weeks).
While it may look grim, it is actually beautiful to watch the incentives in real world action. In legacy world, user exploiting system could easily get his account frozen and blacklisted. In some cases legal action would be considered (e.g. for market manipulation).
But in crypto we are working with a permissionless premise. We do not want to kill Thanos. We want to take away his glove. To fix the incentives of all actors and align them with the long term benefit of the protocol. Synthetix, despite aiming for decentralization and immutability, is still flexible enough to allow for tinkering until we get it right.
Edit — one day later: I am humbled by so many reactions to this article. It really stirred some drama, but also showed that many people care about Synthetix and are interested in it succeeding. The team prepared a number of measures to reduce the impact of front-running and hit and run staking. Also honest stakers will be partially reimbursed. Read more in recent announcement.
Short term outlook
Will it get better after we fix these exploits? Hopefully. It will certainly be harder to exploit the system in ways described above, but we may see other clever exploits happening.
However, SNX stakers will still be on the other end of each synth position, this part is an inherent property of a debt pool system.
It’s not an issue in a theoretical balanced scenario where the platform has the same amount of shorts as longs for every asset. In this case, SNX stakers would have a market neutral debt pool and simply earn profits from fees.
But the practice is very different. Looking at current positions, there are hardly any i-synths (shorts).
Debt pool increases with increasing price of ETH and vice versa. SNX stakers are thus indirectly shorting ETH, which is very painful in a current bull market.
This creates an interesting dynamic where SNX staking is essentially a counter-trend trading to ETH (and crypto market in general). This may also mean that SNX can be seen as anti-cyclic asset within the crypto ecosystem, at least in short term. Anti-cyclic assets are good for balanced portfolio, but can severely underperform in a bull market. It’s up to every SNX staker to either bite the bullet or hedge by ETH longs.
Long term outlook
In the future, Synthetix aims to become an all round liquidity platform for synthetic assets within AND outside of crypto. This means traders will have access to synthetic long and short positions on commodities, stocks and indices, with or without margin. With wide enough usage, this could lead to more balanced debt pool, where stakers are not extremely exposed against any particular synth position.
However, sophisticated SNX stakers will still need to hedge and often this hedging will need to happen outside of crypto, on the legacy financial markets. This leads me to believe SNX staking will get more complex and institutionalized. Essentially the same thing that happened to Bitcoin mining — we went from hobbyists mining at home to large professional mining operations. This is inevitable for any crypto-economic system that will become big and important enough.
You may object, that this may concentrate the SNX ownership from the core community to institutions. I think, that this may not necessarily be the case. Delegation services could emerge where SNX holders will be able to stake their tokens with the most efficient staking and hedging entities for a fee, while retaining their ownership and community voice within the protocol.
I would be surprised if none from the core SNX community members did not already work on such delegation service.
SNX staking is not simple, there are many caveats and risks associated with it. These are not reflected in aggregate metrics and ratios such as Mcap/Earnings, that’s why it may look extremely undervalued. There are many gaping holes in the system, but team and community is actively tackling all of them.
We still think that SNX is one of the best investment opportunities in DeFi. The future of synthetic assets is bright and exciting. That being said, it’s also carries unique risks that should be accounted for. As such, only experienced and well researched investors should consider becoming SNX stakers.
Author: Fiskantes is a CIO of Sigil Fund – Experienced Investor Fund investing in digital assets, crypto money and decentralized networks, relying on deep fundamental research. Fiskantes also co-founded Crypkit – digital assets tracking and accounting toolkit for crypto companies.