An analysis of the collapse of IRON on Polygon

For those of you following the collapse of IRON / TITAN on Polygon, here’s my take on what happened.

IRON basics — an algorithmic stablecoin?

IRON, like FRAX, from which it was forked, uses a two token, partial collateralisation system to stabilise peg. FRAX has been proven effective for well over a year. IRON on Polygon, however, collapsed in days, despite IRON on BSC holding peg since its launch several months ago.

The IRON / FRAX model revolves around the concept of a collateralisation ratio (CR), which represents the percentage of the native stablecoin that is backed by another stablecoin, such as USDC. The CR starts at 100%, and is gradually lowered by the system over time, with the theoretical goal of reaching 0% and becoming a full “algorithmic” stablecoin.

What many people don’t realise is that this is in fact a 100% collateralised stablecoin model, and not strictly an algorithmic stablecoin. Before the crash, IRON had a CR of about 70%, meaning it was 70% collateralised with USDC. The other 30%, however, was not uncollateralised. It was collateralised by Titan.

The only part of FRAX / IRON that can be said to be “algorithmic” is the algorithm for adjusting the CR based on market demand. With IRON, the rate at which the CR can be adjusted was much faster than FRAX. USDf anyone?

The problem

There is in fact an inherent weakness in the FRAX peg mechanism, which FRAX mitigates through careful control of FXS inflation and FRAX vs FXS market cap. IRON inherited the system and therefore this weakness.

In times of panic, people sell and redeem IRON en masse. the peg mechanism works like this. People sell IRON -> IRON drops below peg -> Arbitrageurs buy and redeem IRON for part USDC and part TITAN -> Arbitrageurs sell TITAN in order to profit from the arbitrage -> IRON price creeps back up to peg, TITAN price drops from the redemption sell pressure.

Theoretically, at some point people should stop panicking when they realise that IRON will not lose peg. However, there are two factors affecting this outcome. First, there needs to be enough TITAN liquidity to last through the panic. Second, the IRON redemption oracle needs to keep up with fast movements in TITAN price, otherwise you wouldn’t actually be able to redeem IRON for 1.

Neither of these conditions were met by IRON. IRON supply far exceeded TITAN supply, and the oracle was set to update TWAP on a 10 minute cycle, meaning that during times of fast dump, you could not actually redeem IRON for 1, compromising the very arbitrage system that is supposed to keep IRON to peg.

Remember when I said that FRAX / IRON actually uses a 100% collateralised model? At 70% CR, IRON is 30% collateralised by TITAN. This means that, theoretically speaking, TITAN’s market cap (or more accurately, liquidity) would need to be at least 30% of the market cap of IRON in order to hold the peg.

The problem, however, is that TITAN does not have a stable price. Here we run into the same kinds of problem that the likes of DAI and Synthetix have. There is a reason why their stablecoins are overcollateralised.

So, IRON was a 1:1 backed stablecoin with 30% of the collateral being a new, unstable token. What’s more, the TITAN collateral was not actually adjusted based on the market cap or liquidity of TITAN — instead it was adjusted based on the Target Collateral Rate (TCR) set by the protocol. There may only be enough TITAN to collateralise 10% of the IRON supply, but the TCR can still be set at 70%, or even lower.

There may only be enough TITAN to collateralise 10% of the IRON supply, but the TCR can still be set at 70%, or even lower. Add to this the fact that TITAN was launched with a very ambitious rate of inflation (what some would call the “degen” farming model), and we have the perfect storm.

The aggressive lowering of CR (30% over a matter of days) drives demand for TITAN since TITAN is needed to mint IRON, so while the demand for IRON is high, TITAN price is also being inflated, in turn pushing up IRON and TITAN LP staking yields, and everything looks great.

However, at the rate that IRON / TITAN grew, a lot of people were making a lot of money. A dump at some point is inevitable. We have seen this again and again in basically every degen farm.

The difference with IRON is this. Some degen farms that were building products with a use case survive the initial dump (and the later dump cycles that naturally occur as waves of investors exit and enter the pool). With IRON, because of the inherent weakness in the collateral model, a mass initial dump of TITAN caused a significant discrepancy in the value TITAN available as collateral and the % of IRON actually collateralised by TITAN.

As people redeemed IRON and sold TITAN, eventually the available TITAN liquidity pools were drained. The result is what some are calling a bank run. However, I don’t really see it as a bank run. Or, if it is a bank run, it is a bank run of TITAN collateral, not so much of IRON.

The Effective Collateral Ratio (ECR) of around 70+% of the IRON peg that was redeemable for USDC was never at risk, remaining rock solid through to the redemption of the very last IRON in circulation. It was TITAN that was insufficient to cover its ~30% of the IRON peg value.

Comparison with FRAX and UST

Compare this with FRAX. There is 116m of FRAX in circulation and ~30m of FXS. The current CR for FRAX is currently ~85%, meaning 15% of FRAX value, or 17.4m, is backed by FXS. Roughly speaking, that’s 1.72x the amount of FXS required to back FRAX, but of course the actual amount of FXS in liquidity pools will be less.

This is in fact a “soft” collateral, in the sense that people redeeming FXS need to sell the FXS on the market, the price of which is volatile. To mitigate this, FRAX adjusts its CR down very slowly and carefully controls the inflation of FXS, yielding more “boring” APRs and implementing long token lockup mechanisms in exchange for protocol stability. It took over a year for FRAX to be at its current CR of ~85%.

In short, the FRAX model is simply not compatible with degen yields. For IRON, on top of a potential profit sell-off of TITAN like with any degen farm, there is an additional layer of sell pressure coming from redemptions as people exit their IRON positions. So in this case, fast growth + fast lowering of CR = perfect conditions for the triggering of a death spiral.

Comparing with LUNA / UST — this is an interesting one. It can be said that LUNA / UST is already where FRAX wants to be — an “algorithmic” stablecoin that is 100% backed by another token in its own ecosystem. UST is effectively at a CR of 0%, in FRAX terminology. How has it remained stable?

Terra built an ecosystem that gave LUNA a use case outside of being just the mechanism tied to a stablecoin’s peg (although this is one of its core functions). This gives the token more price stability since its supply and demand is no longer mainly driven by yield farmers chasing the most lucrative yields.

On the back of more sustainable token value, Terra then created stablecoins backed solely by LUNA. At the time of writing there is nearly 2b UST in circulation and 2.4b of LUNA. However, this is a rough comparison because ultimately what matters is what’s in liquidity pools / available for trading, since it is the liquidity that is serving as effective collateral.

As with all collateralised as opposed to real algorithmic stablecoins, depegging will occur if the collateral value falls significantly short of the stablecoin value. UST is always redeemable for 1 of LUNA, meaning that if LUNA prices fell by 50% today, UST would be significantly undercollateralised. This has yet to happen, but will always remain a risk for any collateralised stablecoin that has a significant proportion of its collateral as non-stable coins, without enforcing significant over-collateralisation. There is no magic, only maths.

The Mythical Beast

True algorithmic tokens are fundamentally different from the likes of FRAX, IRON or UST. They can be partially collateralised in an effective sense, like MALT, but the uncollateralised parts are uncollateralised in truth — they are backed by nothing but the protocol’s algorithms designed to influence supply and demand. They are the holy grail of decentralised stablecoins, but whether they are more than creatures of myth remains to be seen. I will continue watching the space with keen interest.

Dark Ninja + Crypto Ape + Degen Troll