Thread by @SlawTeh

1/ This long thread is an exploration on the notion that when you LP into an AMM, you are selling convexity. At the end of a longish abstract thread, I point out some of the ways that Thorchain and its native token $RUNE attenuation this risk.

2/ In finance, convexity is a notion of acceleration. Think of like the arc of ball thrown off the roof of a building. It drops at an accelerating rate. Imagine that ball was an investment, dropping at an accelerating rate. Putting assets into an AMM LP can kinda be like that.

3/ How so?

Imagine a bitconnect:ETH AMM LP. Hah! You can’t unthink that thought. As bitconnect ponzied ever higher, LP investors would have been furious about the way that token was sold out for more ETH by the AMM. This outcome is one version of convexity risk.

4/ It can be even worse.

At a certain point, cease and desist notices were sent to the bitconnect project. Its token plunged immediately to near zero. And the LP went to near zero. And since double near-zero is still near zero, being in the LP mattered little.

5/ This outcome is the worse version of convexity risk from LPing.

When you enter funds into an AMM LP, you are agreeing to continually selling the better performing asset for the worse performing asset. In other words, you are selling your winners to buy your losers.

6/ Such transactions are generally against some of the first tenets of investing: cut losers quickly and let your winners run.

Why do this? Two reasons.

The first is that as an AMM LP holder, you get fees that you wouldn’t get if you hodled the two LP tokens separately.

7/ These fees alone can provide double digit yields on the AMM LP capital. Looked at by themselves, the fees are a money-for-nothing trade. But are the chicks for free?

8/ The second reason to sell-your-winners to buy-you-losers is mean reversion, the presumption that any significant outperformance of one asset is temporary, and soon the pair return to a more typical relative value.

9/ It’s only significant persistent outperformance of one asset that generates convexity loss, and if we simply assume mean-reversion, we can look past this risk.

10/ As an investor in an AMM LP, you receive fees in exchanging for the geometric mean return of the two paired assets. And like with a ball arcing down from a roof, geometry can be a bitch. So LP holders deserve compensation. How much? Are fees alone enough?

11/ It’s at this point, this thread gets more exploratory. The place I start is with the FTX’s BTC MOVE index: help.ftx.com/hc/en-us/artic…. The index is simply the absolute difference between BTC’s period high and period low. If BTC runs from $50k to $60k, the index value is $10k.

12/ If BTC falls from $60k to $50k, the index value is the same - $10k.

Right now, you can buy or sell the 3Q BTC MOVE contract for $22k (and BTC is $59k, so a 40% premium - Nice!).

13/ At contract’s expiry, the seller pays the buy the index value - the difference between the quarter’s high and low BTC prices. This contract is means of buying or selling convexity risk.

14/ If you sell it and BTC goes down, you end of feeling like the bitconnect holder who LPed during ponzi-time. Maybe BTC falls 40% in a quarter and you pay as much as you receive, but the contract probably never puts you too much in the hole.

15/ But if BTC triples, it’s rektage - you receive $22k and owe almost $100k. Rekted perhaps worse than the bitconnnect holders who lost everything because you probably posted additional capital as this trade went against you.

16/ At least that’s what your AMM would do with your LP capital.

Well how much is BTC likely to move during a quarter? It would be kind of nice to know this value to have an expected value / return from the trade.

17/ If the year’s high on BTC is expected to be 2X the year’s low, then the quarter’s high should be expected to be 19% above the quarter’s low. Sell 40% and pay 19%. Nice!

Selling convexity pays upfront. And every day you LP, you are paid those days’ fees.

18/ But it’s still the continuous sale of convexity.

Here is where my understanding and thinking slows and my brain smooths. I know currently I can expect a return of roughly 20% a quarter for selling BTC MOVE contracts (which is selling convexity).

19/ Given this return opportunity, what return opportunity would make staking into an BTC:USD AMM LP equivalently attractive?

I’ll pay 20 RUNE to anyone who can write an analytically precise answer that makes sense to the typical RUNE-twitter reader.

20/ My friends at @Qi_Capital - a great follow - will decide if someone has earned this payout.

My instinct is that this return requirement is probably higher than provided for in many of the AMM LP opportunities out there.

21/ Interestingly enough, this problem is one that three factors of the Thorchain / $RUNE project addresses.

The first is downside Inpermanent Loss protection - gitlab.com/thorchain/thor… which is a form of downside convexity risk reinsurance.

22/ But the value of this guarantee may not persist over time.

The size of the LPs will grow and grow and hopefully become massive - a $10B BTC AMM would be awesome! Also, Thorchain slowly emits its stash of RUNE into the ecosystem.

23/ LP capital is what gets insured (and grows rapidly). Thorchain reserves decline. There will be a point where the system cannot reasonable cover any bitconnect like event.

And huge market correction could be problematic too.

24/ There may be a day where it make sense for the ecosystem to either charge for IL protection or reduce the coverage.

Those days are likely far away because of current the size of the LPs and the reserves.

25/ So the presence of IL protection does minimize the convexity risk taken by Thorchain’s AMM LP investors.

The second factor: Thorchain is governed by an incentive pendulum which directs fees and block rewards to LPs and node operators.

26/ For the past many months of single-chain ChaosNet operations, this pendulum has decidedly favored LP holders - to the surprise of many. LP holders have earned substantially boosted returns.

27/ I believe because of the amount of convexity risk that is being sold, this was a market clearing condition. The risk was that RUNE would moon. RUNE mooned. And this risk persists.

28/ Even as this risk abates on RUNE’s ascent to $100, I believe the incentive pendulum will persistently favor the LPs over nodes.

29/ The third factor: Absent the idiosyncratic risk of a RUNE melt-up, the fact that Thorchain is composed of tokens for the chains with the most significant economic value in the crypto-verse means that RUNE takes on the qualities of being a crypto index fund (with network…

30/ …effects!).

If we were to design a base token to build AMM LPs around, we would want it to have exactly this quality. LPs are provided returns for letting investors trade into and out of specific tokens against a backdrop of the overall crypto market place.

31/ I hope for a more analytical understanding of the returns deserved to those that sell convexity through AMM LPs in general, and the Thorchain AMM LPs in specific. Until then, I’ll probably low IQ ape in a bunch of pools.