[Passed] NuPond Term 5 Grant

what this means and what is the purpose?

The smaller deviation makes cancelling and placing orders more agile - if I understood that part right.

Iā€™m abusing the parameters to squeeze out 0.1% more offset. These parameters are for the cny bot not the btc bot.

ā€¦compared to the standard settings for the bter cny operation.

I wouldnā€™t call it abusing - itā€™s still within the limits of the operation, right?
It makes adjusting orders to changed price more agile and the increased offset reduces the loss for liquidity providers - or even allows them to make money with this offset!
This is necessary to make the LPs keep up the walls.

edit:
To make it clear:

With an offset of 0.004 and a fee of 0.2% (0.002) the deviation of 0.0015 is small enough to keep the orders within the spread after fees: 0.004-0.002+0.0015=0.0035.
With an offset of 0.005, a fee of 0.002 and a deviation of 0.0005 the result is the same: 0.005-0.002+0.0005=0.0035.

I only wonder about one thing:
isnā€™t the ā€œspread after feesā€ twice the [offset - fee + deviation]?
In that case both settings would be outside the 0.4%

SAF is purely 2*(offset-fees). Deviation will not ever decrease your spread, but only allow your price to drift some from the server price. This is an abuse of the parameters because it will most likely cause more bugs in the long term but weā€™re getting killed by the arbitrage on Bter (i.e. the pools are getting out of balance). The CNY bitcoin exchanges have been very pumpy these last few days.

Ok, new update: https://github.com/Nagalim/alp-collection

Iā€™ll change the tolerance later, and maybe do some other things, but hereā€™s how you use this update. Basically, strategically you have nothing to do from the default parameters but decrease the spread, so this method will decrease SAF. Try parameters like these (SAF = 0.2%):

Determines window of allowed price deviation

deviation = 0.0005

Determines market spread

offset = 0.003
shift = 0.002

NBT are getting eaten on the CNY orderbook at high prices. I plan to increase the tolerance to 3%, possibly even 5%. I will either suggest or automate a reasonable shift parameter for use with that high a tolerance.

1 Like