[throwaway7808]

Now you are just bullshitting. A 130k is a huge loss and a phenomental outlier. You've never had even a 100k loss on your strategies. And by current standards your strategies are not even HFT. I've seen you talking milliseconds. Nowadays people are talking sub-microseconds.

[sillysaurus3]

Sub-microsecond transactions seem impossible unless you're physically jacked in to the trading datacenter's network. If you try to transfer a message (like a string buffer) as quickly as possible from program A running on core 0 to program B running on core 5 on your server-grade computer, the best benchmarks I could achieve were "99% of measurements executed in fewer than 150 nanoseconds." And I worked hard on this problem. It seems like maybe you could execute a trade in ~400 nanoseconds at best, if you have a server that's connected with 10GigE directly to a NASDAQ (or whoever) box and you're bypassing how Linux normally does packet handling, but I wouldn't be surprised if the overhead of the other party's server pushes that to at least 1 microsecond in all cases.

But... now that I've examined the facts, I admit what you say may be true. Are HFTs really operating in sub-microsecond time for trades nowadays?

[yummyfajitas]

The goal is to be faster than the other guy - HFT is a race, thanks to the subpenny rule. So if network latency is 2ms, you still care about making your trades happen in 2,050us vs 2,100us.

Tl;dr even if trades happen in ms, you still care about us.

[throwaway7808]

Yes, you are correct. On an average Linux box to have a single cache line data transfer between the cores under 150ns 99% of the time is about the best that you can get. Especially if you are running stock kernel that eats this 1% and creates huge outliers ;). There is some talk though, about crazily-expensive switches with integrated FPGAs...

[theyoungestgun]

These switches aren't really a secret.

And the crazy expensive switch is guaranteed to be the cheap part. Now add the all the ip required to make that fpga smart enough to place orders, and you're talking huge bills in dev hours and third party licensing.

[Animats]

There are HFT trading systems implemented in FPGAs. They're written in Verilog. Here's one system:

http://www.bittware.com/fpga-dsp-applications/applications-s...

Want to work in that field? Know VHDL/Verilog, Linux, networking down to the wire level?

https://www.linkedin.com/jobs2/view/1821955

[foobarian]

Light moves at 1 ft per nanosecond... I sure hope you're close to the NASDAQ data center :)

[sillysaurus3]

It's 1ft/ns in a vacuum, but my understanding is that it's closer to 0.5ft/ns in a wire, so what you say is doubly true. But I think HFTs are running on servers close to the NASDAQ datacenter, and that people pay buckets of money for such privileges.

[TrainedMonkey]

Pretty sure NASDAQ rents servers directly on site for serious dough, so distance is not that much of an issue.

[kasey_junk]

Yes and No. Nearly every electronic exchange now offers co-location services (including NASDAQ).

The "serious dough" part is a little harder to quantify. I've not looked into NASDAQ specifically but server colocation is usually on the order of a couple of thousand dollars a month. This is a drop in the bucket compared to the real costs of a professional trading outfit (namely employees and margin/risk costs).

Anecdotally, it's also almost exactly what I paid for a tier 1 co-located server at my first job in a startup during the first dotcom boom.

[steven2012]

Wire? It's infiniband and other bufferless network protocols over fiber.

[jagger27]

Fiber is still only about 66% of c.

[iaw]

I worked at a prop shop where my teams worst 2-day losses totaled more than $2MM. The worst period for any team on the firm was a $30MM loss with an additional $60MM of risk over the course of 5 days.

No one was happy about it, but it was part of the job and considered reasonable volatility for a portfolio of our (middle) size. While I wasn't a fan of Consz's tone, his numbers check out from my personal experience at a trading firm. Not HFT, but in the end it's all about return on capital and the P&L's should be similar.

[kasey_junk]

One of the big problems about conversations around "HFT" is that there is no definition of what that actually is. I like to put computerized trading systems on a scale of 3 different things: 1) trade volume 2) latency sensitivity and 3) human interaction.

So for instance, I've worked on systems that were "fairly" latency sensitive (~tens of micro seconds in latency budget), not very high volume (100s of trades a day) and had virtually no human interaction. I've also worked on systems that were "not very" latency sensitive (co-lo'd but never actually measured tick to trade times, which were assumedly in the tens of milliseconds), high volume (10s of thousands of trades a day) and had a team of clerks looking over it.

I would call both of those systems HFT systems, but some people wouldn't consider either of them HFT.

In both of those cases a 130k loss would be an outlier, but not a phenomenal one. I've worked with people who have lost millions of dollars in seconds and it blew up the group. I've met others that million dollar swings on a given strategy was par for the course (and there hedge funds doing manual trades that won't notice that trade in a graph).

That is, the world of computerized trading is pretty varied. If you are going to say someone is or is not HFT, let them know what your definition is first, as that is the only way the discussion can move forward.

[smrtinsert]

People are downvoting you, but as someone who only worked on a FIX application in a tiny finance shop, microsecond was barely acceptable. The comment rings false with me as well.

[theyoungestgun]

What do you mean that 'microsecond as barely acceptable'? That was too long for what?

I'm skeptical that there are really that many players out there who are in a position where a single microsecond would really make a difference. There are definitely companies out there in this situation, but not many.

[consz]

okay

[throwaway7808]

Duh. Common. Describe me just one automated strategy for which $130k would not be a phenomenal outlier. And you were talking multiple strategies...

[consz]

I was talking about 130k in any one strategy. I'm sorry you misunderstood me.

A single strategy is easily capable of losing that much. Any bond futures or cash bond strategy post-FOMC. Any equity index futures strategy during Twitter "flash crash". Any strategy taking large size in a big future during early-mid October this year. Any strategy taking large size in a big future during August 2011. Any Nikkei futures strategy right after the recent QE announcement from Japan.

All of those are easily capable of dropping 130k in a day.