[Discussion] about the size of buy and sell side / T1-3 (T4)

NuBits
#1

Continuing the discussion from FLOT Operations - buy side / BTC related:

TL;DR
The buy side needs to be on a minimum level (in percent) of the NBT in circulation to increase resiliency over temporary decline of demand for NBT
/TL;DR

Long read:

I fully agree with these concerns.
T1 to T3 need to be bolstered up if more NBT are in circulation.
That can be achieved like @mhps already lined out:

It was necessary to adjust T4 buy side to NBT in circulation to have a sufficient buffer for temporary decline in NBT demand. I’m glad that the motion to do that passed.
But it’s it necessary to adjust T1-3 buy side to NBT in circulation as well. Otherwise the impact of a decreasing fraction of the NBT in circulation can mess the sides up and endanger the peg (I’ll explain later why I don’t think it’s necessary to adjust sell side dynamically as well).
With a static T1-3 buy side the peg can only be kept if static amount of NBT hit it. That is an unrealistic expectation.

This thinking (adjust buy side to NBT in circulation, but not sell side) moves Nu away from the scheme to have balanced walls, but spends money (for compensation) only where it’s required more urgently.
The sell side size can remain uncanged, because there’s no direct relation between the size of the sell side and the NBT in circulation. More NBT in circulation don’t necessarily increase demand for NBT, right?
But there is a direct relation between size of the buy side and NBT in circulation. More NBT in circulation means that more NBT can be sold for BTC.

In practice it could look like this:

  • total value of T1-3 sell side (for which is paid a compensation): 40,000 NBT
  • total value of T1-3 buy side (for which is paid a compensation): 4% - 7% of NBT in circulation

How do I get to 4%?
Approximately 746,710 NBT are in circulation at the moment.
The current total T1-3 volume is approximately 80,000 NBT. Each side (obviously valid for buy side) shall be between 40% (32,000 NBT) and 60% (48,000 NBT).
32,000 / 746,710 = 4.x%
48,000 / 746,710 = 6.x%

And why the 7% as upper limit?
There shouldn’t be a requirement to adjust compensation for liquidity providers with each change of NBT in circulation. A corridor is useful if the NSR holders don’t want to be kept busy with continuous adjustment of compensation to keep the T1-T3 volume on a desired level.
Plus a change of compensation might have only effect beginning next term, if there are no amendments to contracts which allow a dynamic adjustment without direct NSR holder involvement (this will be quite easy with the upcoming fixed compensation).
It needs to be taken into consideration that this compensation not only depends on the NSR holders’ desire to in-/decrease T1-3 to adjust to a changed amount of NBT in circulation, but to changed requirements of liquidity providers as well.
The situation of a particular exchange (compensating for liquidity providing on cryptsy should be quite expensive at the moment) or the market in general (other ways to make more money than bein an LP) play a role.
I recommend to focus on the the adjustment of compensation for providing liquidity to a changed amount of NBT in circulation here.
After all the 5% and 7% T1-3 buy side size are only a recommendation to initiate a discussion.
NSR holders decide over the amount of T1-3 buy and sell side with the compensation they pay for providing liquidity.

Why should we even start thinking about having unbalanced sides?
Should the sell side need to stay as big as the buy side (even if the buy side gets dynamically adjusted)?

My view is:
Sell side funds can be injected in the market through gateways in minutes. My NBT entry gateway for Poloniex is currently being up for voting. We need more of those gateways!
Buy side gateways (NBT exit or BTC entry, respectively) are possible as well (I have one of those up for voting as well), but transferring BTC is much slower and for that reason I think the buy side should offer more buffer.

The drawback of adjusting the T1-3 buy side dynamically to the NBT in circulation is an increased cost for compensating liquidity providers.
The benefit of adjusting the T1-3 buy side dynamically is an increased resiliency against temporary decline in demand for NBT.
Adjusting it to e.g. 5% of NBT in circulation is no defense against people who want to attack the peg by first buying a large amount of NBT and dumping it on the buy side son after, breaking the peg this way.
If an attacker buys 50,000 NBT a dynamical adjustment of the buy side like outlined above increases the buy side by 2,500 NBT value.
But it’s still better than no increase at all!

The next big drop of BTC/USD will very likely cause a lot of NBT being sold for BTC. Nu should be prepared as good as possible.
Adjusting the value of T1-3 buy side to NBT in circulation might be as important as adjusting T4 buy side to NBT in circulation.
Especially if BTC/USD crashes, keeping the T1-3 and T4 value above a minimum level is important.
One of the reasons for this motion:

1 Like
#2

I think dynamic buy side determined by circulating NBT is an improvement to a static buy wall set to be equal to the sell side.

#3

Do you think adjusting sell side dynamically as well is necessary or do you think my reasoning makes sense?
I was thinking about the implications of not adjusting buy side dynamically.

Adjusting it comes at a cost (compensation for LP).
But not adjusting it comes at a cost as well. If the FLOT is continuously involved in balancing the walls, it’s likely that FLOT members ask for more compensation or quit their membership.

Increasing the buy side (and optionally the sell side as well) decreases involvement of FLOT.
That is good to keep FLOT members at it.

And it’s good to have more buy side buffer, because involving FLOT to balance walls takes time, more time than involving a single person.
NBT can be released with 3 (of 5) signatures.
BTC requires 5 (of 8) signatures.
Filling buy side takes more time than filling the sell side, because of the slower BTC confirmation and because more signers are required.

#4

What’s the technical mechanism by which we increase T1 based on circulating nbt? For fixed reward it makes a little more sense: have the target be a % of the circulating nbt. But what rate should it pay? Should it just raise the rates until the target fills? For fixed cost, even that strange notion isn’t possible. Basically, adjusting the payment scheme to attain a specific wall target is a fixed reward strategy.

I’m a bigger fan of the fixed cost strategy. Here, we talk about the LP budget as a function of circulating nbt, thus making it an inflation rate. Then, the shareholders strategize about the most effective places to put that budget. If that budget isn’t getting what we need, increase it as a %/month of the total circulating nbt. Eventually, we will find something resembling the rate required to get the liquidity we’re looking for.

#5

The function yet needs to be determined.
Do you agree that more NBT circulating require an increased (buy side) liquidity to keep the peg or does that make no sense to you?

#6

I’m just saying I wouldn’t phrase it like that. I think more circulating nbt means we need to increase our LP budget, which often has the effect you desire, yes.

Currently, our total LP budget is something like 1.5%/month.

1 Like
#7

You arw right.
And I should have added “as long as it seems that NBT are mainly used by daytraders to hedge BTC volatility”.
Once NBT is used for payments and kept in circulation it will be different.

#8

It seemed that the sell side was not enough a few weeks back. I think if the liquidity of A side is increased, B side will be automatically increased when the increased liquidity of A side is used. So there is a natural rebalancing effect. What makes T4 necessary is when the natural ebb and flow exceeds a threshold (40% now), a sign that the ebb and flow will crash the dam.

So the process seems to be

  • when we sell new NBT, we increase T1 target of pools by some factor, and increase T2-T4 holdings by some related factor.
  • when the sold NBT is sold back and the abovementioned theshold is not crossed, we do nothing and let the T1 sell side grow (the target does need to be increased. A pool should be then compensated by its combined liquidity at any moment).
  • If the threshold is crossed mobilize T4 and T5
  • when we get too much NBT than the t1-5 can handle, we burn NBT, and decrease pool target and T2-T4 holdings by some factor

Does this sound right?

#9

[quote=“masterOfDisaster, post:1, topic:3135, full:true”]
In practice it could look like this:

  • total value of T1-3 sell side (for which is paid a compensation): 40,000 NBT
  • total value of T1-3 buy side (for which is paid a compensation): 4% - 7% of NBT in circulation[/quote]

[quote]The drawback of adjusting the T1-3 buy side dynamically to the NBT in circulation is an increased cost for compensating liquidity providers.
The benefit of adjusting the T1-3 buy side dynamically is an increased resiliency against temporary decline in demand for NBT.[/quote]

So how about a third option, I agree that it would be wise to increase our buy side support to a dynamic one that grows with the amount of outstanding Nubits away from a static support which we have currently. However I offer a slightly different solution. Now the reason why we want to increase buy side support is not because of continuous sell pressure but to prepare for possible large dumps of Nubits in short timeframes. Hence we could state that our current buy side support if sufficient until such a huge dump occurs. As I’ve stated in a few different topics as well I’m against borrowing money if you have the money yourself. Now I won’t make this topic completely about the borrowing of funds to provide liquidity.

However in this case essentially we are talking about emergency buy support, I feel it would be a huge waste of money to contentiously pay interest over additional liquidity that is only there for an unexpected emergency (huge dump of NBT). It would be very costly to pay continuous interest over an amount of money used for liquidity that is probably only going to get used a few times a year at most. So how about the following.

Total value of T1-3 buy side stays the same as it is now. However part of T4 funds are used as a backup to insure Nu against sudden huge sell offs of Nubits. This could be easily done be having some of the T4 funds be on an exchange with an liquidity bot but at a slightly greater spread then the Liquidity pools are providing thus insuring that the funds never get used unless someone dumps all over the walls of the liquidity pools.

Advantages:

  •     Guaranteed additional buy side support for emergency situations.

  •     No additional LP costs

  •     Full control over emergency buy side backup support for Nushareholders

Disadvantages:

  •     Funds would have to be placed at an exchange account causing them to be centralized (could be offset party by having multiple shareholders do this and perhaps collateral) and vulnerable to counterparty risks

  •     Funds would be vulnerable to exchange default
#10

If we boost to 2%/month, that would make our daily budget ~$15k/month. We end up with ~$500/day. NuLagoon limit can be changed to 0.7% (giving currently >$5000/month). Nupool can mirror this. Nupond only needs 0.2% (~50/day, more than i’m doing now). This leaves 0.4% for the others, most likely something like 0.25% liquidbits, 0.1% NuRiver, 0.05% modPuddle (325 nbt/month, again a higher rate)

#11

@Dhume this is a T2 solution, and an interesting proposal. I still think we need 15% in T4 multisig, but we could do another 1% on poloniex at least.

#12

We could have 15% in multisig X % in emergency backup buy walls and the rest to buybacks.

#13

As @Nagalim said this is not T4 but T2. T4 is designed to be decentralized, low maintanence, with trade off being slower-than-t3 and still exposed to btc price risk.

I agree paying for emergency all the time seems to be a waste. However if we sell more NBT it is inevitable that the walls will have to face more volume in normal times. So I still think the walls need to be bolstered, although the factor might not be all emergency-oriented.

1 Like
#14

That sounds great!

That is an even bigger step away from the decentralized liquidity model than the gateways that are on standby. But I agree it’s a viable way to create another line of defense for the peg.
Maybe the complete decentralization of T1-3 liquidity is not the optimum and a clever combination of decentralized and centralized liquidity providing in the end is cheaper for Nu.

The more I think about this the more I miss B&C Exchange :wink:

Until BCE is ready, there might be another option, which is faster than a 5 of 8 multisig, less prone to exchange defaults than operating a bot on an exchange with Nu funds, and able to inject (BTC) funds where they are required (speaking of a low buy wall at a single exchange, while overall buy side is not emptied).

  • we need BTC entry gateways (NuBots, buy side only) on major exchanges (basically for all at which Nu maintains liquidity operations)
  • small amounts of BTC (a few thousand USD value) on e.g. 2 of 3 multisig addresses managed by FLOT (with 8 members of the FLOT, two of these 2-of-3 multisg addresses could be created)

Advantages
An additional layer is introduced, which increases the peg reliability
A shorter lead time than having only 5-of-8 multisig BTC address(es)
The exchange default risk is minimized, because no funds are permanently kept there

Disadvantages
An additional layer is introduced, which enhances complexity
It still takes some lead time to bring BTC to buy side compared to active liquidity providing with Nu funds

As you see I compare with different solutions/proposals when I list the (dis)advantages. This is intentional, because this proposal is somewhere in between what we currently have and what providing liquidity with Nu funds would be.

Interpretation of the liquidity tiers, a waterfall model, triggers, metrics and actions
T3 custodians
#15

FLOT just needs to make two levels of addresses, level one is big fund addresses that need many to sign, level two need 2 of 8 to sign but only have several k usd. Level two is replenished from level one.

1 Like