Hi, I'm am interested in getting into FX high frequency trading. I was wondering if anyone has any recommendations for good prime brokers in this space? I have experience in equity HFT. Also, I am unsure of the typical latency requirements in this space. Is sub millisecond execution typical? Just trying to guage expectations. Thank you!
You'll need femtosecond latency for what you want to do. Millions of dollars in bandwidth and equipment. You have to travel at twice the speed of light, which information cannot currently do on human mediums. Thank you for wanting to contribute to the mess that is HFT. Your service to the trading industry is duly noted.
Few points of clarification: 1. Latency is certainly no where near what you mention. My best guess is close to 1 micro for fastest players to do internal work, and then they're at the mercy of the matching engine which costs several hundreds of micros. 2. From first hand experience in equity HFT, not all strategies are malicious. In fact, most players simply market make or do some type of stat arb trade. My guess is most players with the exception of large institutions actually benefit from this behavior.
To answer that I would need to know what type of strategies you look to employ. That will determine everything
Thank you for the response. In interested mostly in stat arb. I would also look to the triangle arb and some straight market making, but I would imagine the former has very thin margins/ultra low latency required.
Arbing what in fx space? It's virtually impossible today. Triangular arb is dead already for years. It's an incredibly low hanging fruit that anyone would immediate exploit if it still existed. If you answered market making I could direct you limited to my own experience in this field. But arbing fx pairs... I don't think so
Not sure I understand your response. For straight market making, what is the typical threshold for latency? I have achieved micros in the equity space with relatively limited infrastructure. As for stat arb, what do you see as the limitation? Speed? The triangle arb was a side thought and by no means my main interest.
I don't see speed issues. If you can be as fast as you claimed then you can employ some serious market making algorithms, though fx space raises a whole different number of issues. For example fx space is a fragmented market, much more fragmented even than US equities. There are no exchanges but segregated liquidity pools that are balanced out via 4 or 5 very large banks and they in turn hedge themselves in the interdealer broker market and/or cb players. Very different ballgame and flows... Back to Stat Arb. Again I do not see what you want to do here. There are no arb opportunities in this space whatsoever imho. Now, market making less liquid pairs could be a serious revenue driver but it can also cost you dearly if you don't have ownership over intelligent algorithms. You can find very willing counterparties that are able to pay a multi pip wide spread in, for example, South Asian ccys. But you may also need to set up a very unconventional autohedger. I don't want to go into too many details because I am also exploring this area. But Stat arb and particularly triangular arb imho is totally out of the question in fx space.
Thank you for your detailed response! Very interesting information. Like I said, my experience in equities and options so this is quite new to me. Seems like an interesting space. If I understand correctly, a simple cross venue arb between different liquidity pools is basically completely absorbed by the banks? Seems like the more exotic currencies would still be inefficient.