[Passed] 1% Maximum Spread in Shareholder Funded Liquidity Operations

Nagalim 2016-05-20 18:23:46 UTC #21

We need our orderbook to look like the orderbooks of trusted markets. Image is everything, when people look at an NBT order book, it should look like this:

JordanLee 2016-05-20 18:26:56 UTC #22

You have obscured that these are questions for desrever, not assertions, by removing “Would there be” from the beginning of the first sentence. In any case, it applies to illiquid pairs, which are the only pairs for which parametric order books are well suited.

Nagalim 2016-05-20 18:28:18 UTC #23

Ok, so here, I’ll give you it unobscured.

There is a useful role in automatically changing the curve values to encourage balancing the available liquidity. In essence, if the sell side has less liquidity, steepen it while while making the buy side more flat.

Illiquidity is a price feed issue, not a spread issue.

JordanLee 2016-05-20 18:59:46 UTC #24

It seems like it would be useful to review the reason for parametric order books.

Imagine we decide to maintain liquidity on an NSR/NBT pair. With very low liquidity, there is a real risk that NuBot will attempt to peg NSR. This isn’t possible to maintain, and will result in a loss for anyone attempting to peg NSR. The important question is whether the supported market is creating the price or reacting to it. If most volume in NSR is on a supported NBT pair, then we are making the NSR price, not following the market. That is the problem that parametric order books can address.

With liquid pairs, our own NBT/BTC pair at Poloniex isn’t attempting to peg the BTC price, because trading on the pair is dwarfed by Bitcoin trading on other pairs, which determine the price NuBot offers trades at.

Essentially, parametric order books are needed if there is a risk that our supported pairs will have any where near the majority of trading volume for a particular asset. That isn’t a risk for BTC, ETH and USD markets, so they should not employ parametric order books.

Nagalim 2016-05-20 19:05:14 UTC #25

Parametric orderbooks cannot address this price-feed issue. If you attempt to use parametric orderbooks as a solution for the price feed issue on illiquid pairs, your system will abused by people who are manipulating the illiquid price feed. Parametric order books are a solution for better bulk liquidity provision on liquid pairs like BTC. A solution for illiquid pairs will require a lot more thought and attention to the price feed rather than the spread.

The solution to this is a recursive price feed, not parametric order books.

JordanLee 2016-05-20 19:09:28 UTC #26

This is what an order book for an asset that is variable in price looks like. A price pegged asset order book ought to be two vertical walls ideally.

Nagalim 2016-05-20 19:10:59 UTC #27

Good luck telling that to traders and bots that don’t care what our product is. They’re just looking at the order books and see big red flags that say ‘this is being manipulated, stay away!’

JordanLee 2016-05-20 19:12:47 UTC #28

Of course it is. People who don’t understand we regulate the value of NuBits shouldn’t be trading them.

Nagalim 2016-05-20 19:13:38 UTC #29

Everyone should be trading in NBT. To say otherwise is suicidal.

You really want to turn away all the bots and traders who don’t know the intimate details of Nu? Seriously? How does that make any sense?

JordanLee 2016-05-20 19:19:38 UTC #30

I wouldn’t call knowing that we peg USNBT to the US dollar an intimate detail. And yes, every NuBit user needs to know this.

Nagalim 2016-05-20 19:23:55 UTC #31

The only way a trader can know whether the manipulation is from Nu or from some nefarious 3rd party is to know the intimate details. What if someone starts pegging NBT at $0.99 on HitBTC? The volume is so small there we wouldn’t respond for maybe weeks. So then a trader goes and just assumes that the price is correct because they don’t know our details, and they get manipulated. With the parametric order book, a takeover like that would require making a whole lot more waves that shareholders will respond to.

A trader doesn’t need to know anything to buy an altcoin. How are you going to take their NBT away from them?

If you mean to say every NBT user should know this, well if ‘shoulds’ were bitcoins we’d all be rich.

You say you want traders to use bots to arbitrage exchanges. Bots are not aware of what is pegged and what isn’t. If you want bots, we need to cater to bots. People aren’t going to redesign their entire bot just to take advantage of our measly little market. We need to bring them to us, not force them to accommodate our completely new economics.

A lot of bots react based on the order book. If our order book reacts to the LP balance, we can get bots to react to our LP balance. This would be a huuuuuge step forward for us.

JordanLee 2016-05-20 19:39:56 UTC #32

Bot operators do typically specify what pairs are to be traded, just like NuBot does. An arbitrageur would set up the BTC/NBT pair as though it were a BTC/USD pair. It is quite simple.

Nagalim 2016-05-20 19:41:21 UTC #33

Right, and their bot would proceed to not trade at all because the order book looks screwy. Making a bot that understands the fact that our orderbook looks different from other pairs would require much more than a simple modification.

JordanLee 2016-05-20 19:46:25 UTC #34

Arbitrage bots don’t examine the order book. They only care about the top of the book, the best price on each side.

Nagalim 2016-05-20 19:48:08 UTC #35

So you are 100% on not wanting any bots that look at the order book? Meaning you think a vast majority of the trading bots out there are undesirable for our network? In that case, it’s no wonder traders don’t use NBT.

Cybnate 2016-05-21 03:04:42 UTC #36

I have another view and submitted my draft motion here: [Withdrawn] 0.6% Maximum Spread for ALP/MLP, 1.2% for Nu Funded Liquidity Operations

Key differences are:

Lower spread (0.6%) for ALPs/MLPs to improve liquidity quality (with exemption when high fees are incurred) and with that likely the volumes
Medium spreads (0.7%-1.4%) for Nu-funded liquidity but limited to maximum of 30% of total liquidity when ALP and MLP fails e.g. in high periods of volatility as a first line of defence.
High spread (1.4%-2.5%) for Nu funded liquidity but limited to a maximum of 10% of total liquidity as a last line of defence
Adding a role for FLOT to respond when High Spread funds are being touched.

Here is the link to my motion:
https://daology.org/proposals/5f9ce9e6fabd149ce8c321cfd28491827602a955

Happy to have the discussion here though.

Nagalim 2016-05-21 06:29:43 UTC #37

minimum practical tolerance is 0.475%, or 0.95% spread.
https://daology.org/u/nagalim/proposals/a8bb5e317723403d8e84bdacb3279b3c2f877bb3

crypto_coiner 2016-05-21 12:38:38 UTC #38

I must admit I agree, in principle in everything that @JordanLee has said in that motion, thus in the provisions of that motion. Great proposal.
However, practice has indeed showed us the role played by gateways, as last lines of defence.

Cybnate 2016-05-22 06:02:24 UTC #39

I don’t understand. When fee is 0.2% on both bid and ask side as spread of 0.4% is technically possible.
Are you referring to the risk of different bots trading into each other? Or strong volatility on the pair?
For fiat pairs this is less likely, for highly volatile Bitcoin this can be a problem. That’s why I took a 0.2% margin in my proposal.

dysconnect 2016-05-22 06:10:36 UTC #40

Allowing 1% spread on the order book means 1.25% tolerance, forcing 1% spread means 1% tolerance where LPs would sometimes only be compensated at 0.75% spread.

Asset liquidity should be defined by volume, where there is at least one easily accessible exchange with considerable volume. Litecoin doesn’t cut that nowadays.