Reducing hedging risk by increasing offset/spread for liquidity operations - to what level?

masterOfDisaster 2016-01-21 02:41:01 UTC #1

I want to prevent losing this train of thought in that verbose post of mine.
It might be good to discuss this before market awareness has been programmes into NuBot, because depending on the results of this discusssion, it might have an impact on the implementation.

Increasing the offset in times of high volatility (and high volume traded by NuBots!), prevents funds from getting drained, sides from running dry, reduces the hedging risk and increases the endurance LPs might have (nupool was empty for an extended period of time).
As future ALP will be based on NuBot, it’s working for that as well (as it worked in the recent days with the Nu funded NuBots on Poloniex).

With the experience I recently gathered, I dare say that an offset of 0.01 (1%) is already enough to slow down trade while still keeping the peg quite closely in times of Bitcoin rollercoaster (like in the past days).
The parametric order books have at least 0.01 offset for the best price orders and typically increments of 0.002.
Adjusting that dynamically would be great.
As long as traders are still buying in the walls in times of extended volatility, the uplift they need to pay for the offset is still too low!
Panic buyings are expensive - that should be the case at NBT/BTC as well!
But what is an acceptable level and based on what basic parameters?

Continuing the discussion from Current Liquidity:

Current Liquidity

What I’m going to talk here now is a paradigm shift that can technically be achieved by market aware offset adjustments (which is on the road map of NuBot development).
With less volatility, the offset can be lower. With huge voaltility, it needs to be increased.
It would be neat to have a feedback loop between the offset and the volume that is traded by the bots.
If you still sell at an offset of 2%, you should do it and increase the offset. If you don’t increase the offset further, you will get the side drained, before anybody can refill it and the peg is lost anyway.
You have sold the product cheaper than you could have. And increasig the offset buys time, slows down the trades if the traders are aware of the offset. If they don’t, their bad.
That doesn’t perfectly match the vision of Nu with high quality products, which keep the peg.
The question is: do you really need to offer a product, that is worth exactly the same at any time? Or might it be appropriate to have an uplift for panic buyings?
As long as you can ensure to buy back 1 US-NBT for 1 USD in normal times, the perception of that product should be good.
This is conceptually heading in the direction of the motion to create a new Tier 6 (Restricted Network Access) liquidity, but from a slightly different angle.
If traders know that during very turbulent times the peg isn’t necessarily as close as during normal times, they can’t complain about an increased offset.

[Passed] Single Sided NBT Gateway on HitBTC - Updated 18FEB16

Nagalim 2016-01-21 02:45:11 UTC #2

2% SAF, which is 1.2% offset on poloniex. Anyway, that’s my two cents (heh, get it?).

masterOfDisaster 2016-01-21 02:45:58 UTC #3

“Per cents”?

Got it?

Bear with me - I should sleep first and ask later…
Don’t the fees add up to the spread?
With an offset of 1.2% per side and 0.2% fees you have a total of 1.4% deviation from the price, resulting in a SAF of 2.8%, no?
In that case to have 2% SAF, the offset needs to be 0.8%

mhps 2016-01-21 03:27:07 UTC #4

Sounds ok to me. looking at the prices when big nbt/btc trades happened, 1-2% spread is not an issue. traders are looking at 10+% profit.

Cybnate 2016-01-21 06:44:48 UTC #5

I’m not a trader but to me it is simple. The lower the spread the more trades happen, the higher the less. When the aim is to ‘advertise’ NuBits and increase trading than we need to be prepared to pay for it. If it is not that important we can increase spreads to 2% on NBT/BTC pair and keep tight peg on the fiat pairs.

Not sure how long we should be advertising and pleasing traders. I agree that in periods of high volatility we shouldn’t be given away money, so increase spreads temporary on BTC pair appears prudent to me. Will leave it to others to find the right balance/formulae between spreads and volatility in BTC pairs.

dysconnect 2016-01-21 06:58:28 UTC #6

When we have multiple NuSafe custodians and also some volume on the fiat pairs, we can definitely loosen the spreads for BTC buyside. However for the time being I think new NBT sales will still occur around NBT/BTC.

masterOfDisaster 2016-01-21 06:59:08 UTC #7

I don’t see an inherent conflict between the two.
Having a (market aware) offset that makes people perceive US-NBT as pegged while reducing the costs for maintainaing an always super tight peg can do the trick.

I don’t the seem unhappy, being able to panic buy/sell US-NBT at an uplift during times in which the daily volatility has double figure percentages.

That will need some time to test, learn, improve, especially if the market awareness is about to be used in the first line and not only in what I called “line of defense”

zoro 2016-01-21 09:42:26 UTC #8

I think it is more simple than it looks.
For example, by reading the news and “reading” the market we can predict a BTC extreme
volatility and go higher in spreads.
The opposite when BTC is practically “stable”.
Nubot users can do this quickly (go from saf;1% to saf:2%).
For ALPs is more complicated and don’t know if it is useful long term.

masterOfDisaster 2016-01-21 11:12:53 UTC #9

You might be able to try adjusting a “minimum offset” manually.
The actual (dynamic!) offset needs to be automatic.

A derivative of BTCUSD price (“price velocity”) helps predicting what side will be hit how hard.
Including the “trade velocity” (volume per time) is another indicator that should be included.

The derivative of BTCUSD price has a “global” significance and can be taken to shift offsets of all liquidity operations (ALP as well as MLP).
The “trade velocity” has only local sigificance (at a particular exchange), but is useful to take local deviations from the registered price feed into consideration.

I recognized more than once a “local NBTBTC” rate at Poloniex, that was off of the registered “BTCUSD price feed”.
It could be read from the asymmetry of orders on the book; symmetric offsets of NuBot caused the orders on one side to be close to the front line and the ones on the other side far away from it.
Including “trade velocity” into the dynamic offset calculation moves walls further from the front line, the faster they get emptied.

This very much reminds me of the proposal to mitigate blockchain spam attacks

The formula to calculate a locally significant offset would be:
NewOffset = OldOffset + (x+y) * ( TradeVolume / TargetTradeVolume - 1)

Where “x” would be the “minimum offset” (decided by shareholder motion).
Where “y” would be the “global offset” (derived from BTCUSD price velocity; to correct the minimum offset).
TargetTradeVolume the volume you consider normal (the total amount or a percentage of the order(s) that gets bought per time frame)
TradeVolume the volume that gets actually bought from the order(s) (again total amount or percentage)

edit:
with the future integration of NuBot into the ALP scheme

the (automatic; once developed) dynamic adjustments are available for all LP!

Current Liquidity