NBT’s primary use right now is arbitrage. An exchange has a local bitcoin price in BTC/USD which is related to the global BTC/USD price through arbitrage channels, which are essentially means by which BTC or USD can leave the exchange. If we ignore market manipulation, the BTC/USD price is theoretically related to the cost it takes to supply BTC as opposed to the cost to supply USD. This for sure includes global relative price, but it also includes a concept of transfer fees and local availability. The global relative price is unknown to a certain extent, and that our transfer fees are much lower than our competitor (NBT vrs USD, big difference in transfer fee), so we use the local availability as our parameter. This comes out in our offset from the price feed (theoretically asymmetric, as shown by my attempts at a shift parameter)
The argument is that it is in the interest of both Nu and Nu’s custodians to offer liquidity at an offset just smaller than the cost to arbitrage.
Right, so we could add together all the costs and measure every possible pathway dynamically and choose an offset correspondingly. Is there an easier way to discover this value experimentally in real time?
If we are talking about btc/usd, then the local btc/usd price itself is the most comprehensive arbitrage cost indicator. But I guess we are more interested in btc/nbt arbitraging. Nubit is so easy to move around I think btc/nbt price should be almost the same everywhere, leaving big local nbt/usd price differences across the exchanges, reflecting the difficulty to move fiats at diff exhanges.
In general the cost of arbitrage is deposit/withdraw fee plus expected price fluctuation in a time frame that equals to a round trip fund transfer.
This is when you don’t mind btc or usd price risks when holding both currencies. If you want to avoid price risks when you don’t arbitrage, you have to do the round trip fund transfer thing (e.g. if usd is your base currrency you have to do a usd transfer and a btc transfer during arbitraging.)
Right, so there is some risk associated with arbitrage. We can imagine instantaneous pathways between exchanges, calculate the price difference and fees, ignore the risk, and come out with a number that is reasonable for use as an offset. Might actually not be unreasonable to do that with NBT/BTC pairs in real time with something like this.
While this seems intuitive, what are the clues that point in that direction in your case?
About the costs:
Not always arbitrage involves moving funds from one exchange to the other (cc for confirmation to expert traders like @Chronos or @pennybreaker). You could start by having BTC and USD parked at two exchanges and increase your net worth without transferring funds.
If you are referring my post above, I said a round trip transfer is needed if you mind price risks. If you constantly hold BTC at the exchanges, you are exposed to its price fluctuation, which could overwhelm your arbitrage gain over night.
Yes, I think most arbitrageurs keep funds strategically placed at exchanges in which they operate. They are also willing to hold BTC for extended periods of time, since this is necessary for liquidity. The exposure to BTC fluctuations can be mitigated by taking an equivalent short position, if desired.