We need to adjust tier1 and tier2 definition

Since the introductions the multi-tier liquidity in JL’s Evolution of liquidity operation thread, many things changed.

The Tier1 vs Tier2 definition is now ambiguos and we ended up with at least three interpretation (as attested by @willy and @woolly_sammoth recently on gitter).

The scope of this thread is to survey the community to create consensus and finally vote a new motion where a clear definition of T1 and T2 is provided.

The origin of the ambiguity is that the software to provide pooled and solo liquidity evolved, and we now realise we need a solid base to build upon.

Early in the days we used to call T1 all the “liquidity currently on the orderbooks of an exchange, in the form of buy or sell orders” … We did not specified anything about the price, so we used to call T1 a buy order at 0.99$ or 0.1$ .

As opposed to T1, we used to call T2 all the “liquidity sitting in the balance of liquidity providers on the exchange, not on order”

Then after the introduction of parametricorderbooks in nubot and ALP servers with tolerance for payouts, we started referring at T2 as “all the liquidity that is either on order at premium prices* or sitting on the balance of exchange not on order”. We also agreed that it doesn’t make much sense to risk funds by having it on balance and not on order, better place them at high/low prices on limit orders.

Then in motion 0ec0be7f113a0bf6ff603545a974cd6410458e00 we said something about tolerances by suggesting a minimum bid/ask spread of 0.007 $ and a max spread 0.052 $ .

It’s time we review T1 and T2 definition .

From a liquidity provision perspective we clearly have three different kinds of liquidity :

  • Liquidity Immediately available at 1$ (± interval) on limit orders. This is the riskiest yet best liquidity, the one our providers should be compensated for. I don´t know if the level of compensation should be the same for the whole interval or should vary with it.
  • Liquidity Immediately available on limit orders, but at prices outside of our confidence interval (price > 1$+interval )
  • Liquidity that for some reason is sitting on the exchange of liquidity providers, not on order.

If you agree on this taxonomy (if you don’t please say it now), then we need to find consensus around :

  1. how do we restructure the new liquidity tier categories (aka, how do we call the groups in the bullet list above) . I am in favour of finding a name scheme which allow us to keep the other Tiers unchanged.
    A possible naming scheme would be using for the categories in the bullet list above the names : T1, T1*(star) and T2 . What do you think?

  2. We need to clearly define the line that allow our software to distinguish between T1* and T1, since it determines good liquidity from bad liquidity somehow, we should be reflecting on this … Here we have different options, the simplest being using a fixed value, predetermined, similar to what we [voted in the past](([Passed] Motion to regulate spread values for liquidity operations). Otherwise we could use a dynamic yet predetermined formula to compute it. Finally we could read the value from a datafeed or another trusted source, on or off chain.
    No matter what we chose but ideally the definition has to be simple yet unambiguous : is the tolerance simmetric buy/sell side? How we compute the 1$ equivalent? …

Would love to hear your feedback so we can draft a motion together.

5 Likes

Definitely anything outside the 5%(+, I guess) tolerance specified in the spread regulation motion is T2.
If we open T1 up as anything within 5% of the price, what does Alix look like?
I think we need to start a talk about how to clean up our custodian-submitted liquidity data.

So I’m actually not a fan of T1* as a network parameter. You can try to carve up the order books that way, but unless you can track prices like a banshee you’re gonna have problems. During the BTC pump, the CNY pool felt this pretty solidly. I have begun mentioning operations with LP reward curves rather than single valued and implemented the shift parameter which is literally applying a local shift in the price. Then there’s the T4 sell side operations recently, where the spread was 0.1%. I just think there’s too many contingencies and we need a hard definition and walk away.

The motion gave the definition. It’s complicated, but we can probably simplify it as 5% for most cases.

Yep, I am a fan of that and also proposed the same thing somewhere else at some point.

So how would you forge the definition to make it forward-compliant with hypothetical LP reward curves?
How would you call the “good liquidity” which is within the x% boundary from 1$ mark?

I wouldn’t define what is ‘good liquidity’. I’d just say T1 is <5% and T2 is >5% or off the books.

This is how I interpret it currently and also what we wrote on https://docs.nubits.com/nubits-liquidity/

I wanted to change it because :

:arrow_up: is too wide. I would like to have a way to refer to those kind of liquidity separately…

I’d do 1.5% or 2% for T*

1 Like

Bump.

Throwing some numbers into the ring:

T1.Best - up to 2% away from the defined USD price
T1.Parametric - up to 5% away from the defined USD price

2 Likes

I don’t have much input except that we should move forward in the proposed direction laid down by the OP.