EXECUTION of crypto arbitrage?

[bojoboy]

Hi guys,


It's a BEGINNER speaking here.

Very few words about myself, I spent the last 5y trading fixed income for a BB algo desk in London (currently in the same position). I'm recently approaching the crypto currencies world on my own. What's really attractive to me is to see traded volumes increasing but liquidity non really improving (i.e. lack of institutional approach in market making the currencies and in running the exchanges).

That said, long story short, I went through weeks of research and data analysis, including all major exchanges and crypto currencies. I'm convinced there are solid arbitrage opportunities on the market may be executed (happy to further discuss this point). However, I learnt that the INFRASTRUCTURE lies between algo and p&l is 100% of the game. Now... I have a few ideas for simple structures, including potential barriers, but let me leave the board white for the moment.

For those interested, this should begin as a very conceptual exercise, discussing all together the potential profit of those arbitrages and your trading experience if any. I'm interested in the technical development of the infrastructure more than anything else. I keep this very high level on purpose.

What do you think? Anyone has experience?    

Many thanks

[1715573029]

1715573029

[1715573029]

1715573029

[1715573029]

1715573029

[jackg]

Just watching prices on different exchanges over the past couple of months I've seen divergences that seem large to me (not sure if they'd be enough to cover the fees) but a difference of ~0.3-1% is able to last a while.

I'm not sure what you mean by infrastructure? Is this as in a bot of your own or on an exchange site as exchanges seem to be quite slow at responding (although this is probably the same all round as stocks can be slow to execute). APIs are normally considerably faster for exchanges too.

Also it'll depend on how much you want to deposit and how safe you want it to be. The main trusted exchanges atm seem to be binance and coinbase... I'd regard bitfinex, kucoin and a few others which would be the next most secure ones to be quite a bit further behind in security - but may be good enough...

[bhadz]

The institutional markets are finally coming in crypto and it's just a matter of time but it's been said that they're coming. And for arbitrage opportunities? I don't think that you'll be happy to try it. Your chosen exchanges might put you into delay with your transfers.
You know the market is very volatile and could make you even lose that quick with split seconds. But if you successfully done it once or twice, I don't think that you can do it again perfectly but I'm not saying that it would be closed and no more opportunities left.

[seleme]

Happy to see the interest by traders in institutional finance organizations. Regarding the arbitrage opportunities, the system is not the same as in the spot forex market, there are negative sides of using algo on fast crypto markets. Think it like trading the news 2 minutes before the event time and you should avoid trading during the rush hours. The slippage, delay, illiquidity, wash trading will be there and the flash market movements will combine together on the bad trading day.

[TheUltraElite]

I appreciate your CV here, but honestly the chance of doing arb anymore has reduced. Previously traders were getting more than 2% +/- on separate exchanges and doing arb to get a good profit off the price differences. I am not saying that it is completely gone but has reduced a lot now.

By tech development I believe you are looking to develop some arb bots for this matter? Or looking for funding as to get a patron to try this arb bots?

Either way you would be better off building your bot and using it yourself, because majority of these bots that you guys "develop" are not worth using anymore on these markets. But good luck making some profit on the same.

[bojoboy]

As institutional only angle atm is to develop execution strategies for the buy-side. Nobody will bother with AML/compliance for market making peanuts for a while.

I'm not sure about your math guys, but in my view if you execute a daily net average of +10bps for 253 trading days a year...

For infrastructure I mean tech solutions may be implemented for this purpose. For sure scanner>API>bot is one of those.

So should we start discussing about scanners (fee structures, order book depth, time lag, restrictions of data providers, computation time)?

Again, please anyone with experience in the field?