HFT Myths

Quote from onelot:

how does your firm handle pre-trade risk? do you get a fixed allocation, and then are limited via a software routing library? or via a hardware check?
More...
The clearing firms essentially trust us to handle our own pre-trade risk controls.

There's many ways to handle the checks, but in general it's done via your own code and works as you would expect (MinPNL, MaxOrderSize, etc, get checked against your orders and the market).

This is where exchanges/regulators could make some welcome and publicizable changes to HFT - by enforcing these checks at the exchange level to prevent fat finger trades and Knight-like fiascos.
 
I believe you said you trade futures exclusively. Say you had to switch over to equities. Would that be a big deal for you or would you not have trouble transferring you edge/finding a new one
 
Quote from jb514:

I believe you said you trade futures exclusively. Say you had to switch over to equities. Would that be a big deal for you or would you not have trouble transferring you edge/finding a new one
More...
I've done both and it is quite different between the two, but not a complete dead-end. The high-level process to finding edge remains the same (how to analyze microstructure, how do I manage my orders, what markets/signals are my competitors looking at, how can I run this business efficiently), but obviously the details diverge from there in many ways.

It's not as drastic as say, a move from high-frequency to medium-frequency, or making markets to hitting markets. In those cases it's very hard to transfer knowledge and processes over.
 
After a strategy has been coded up, to what extent does human intervention come into play? I assume you have someone 'monitoring' the strategy during market hours to make sure nothing goes haywire, but is there a constant tweaking of parameters based on market conditions or the strategy's current performance? Or is all that already baked into the cake?
 
Quote from mo3pro:

After a strategy has been coded up, to what extent does human intervention come into play? I assume you have someone 'monitoring' the strategy during market hours to make sure nothing goes haywire, but is there a constant tweaking of parameters based on market conditions or the strategy's current performance? Or is all that already baked into the cake?
More...
Each HFT firm/trader/strategy varies on this. There are certainly some that are conpletely black box with just an on/off switch, and others that have lots of buttons and knobs that can and are tweaked while trading. Just comes down to your style. As you'd expect, most people prefer to move towards the black box solution when possible, but there are plenty of gray box trades running successfully out there as well.
 
Nice thread. Do you feel it is still possible to make money using the consolidated feeds (assuming colo, etc). Or do you feel the latency of those feeds takes you out of the game.
 
Quote from mr19:

Nice thread. Do you feel it is still possible to make money using the consolidated feeds (assuming colo, etc). Or do you feel the latency of those feeds takes you out of the game.
More...
If by consolidated you mean some 3rd-party feed that consolidates endpoints then yeah I think it's too slow for pure HFT trading (rebates, scalping, inter-exchange arb).
 
Quote from rwk:

@hft: Have you read Haim Bodek's book? Do you agree/disagree with his assessments?

"The Problem of HFT - Collected Writings on High Frequency Trading & Stock Market Structure Reform"
http://www.amazon.com/Problem-HFT-C...DSS4/ref=sr_1_1?ie=UTF8&qid=1369832601&sr=8-1
More...
I have not read it. I glanced at some previews though.

I agree with them somewhat, particularly on the special order types. I'll preface by saying that I don't use them currently since I'm not truly focused on equities trading. From what I know about them (which admittedly isn't much), I don't like special order types and think they create more confusion than they're worth. I also think exchanges should be more transparent with their existence and usage. However, I know several strategies and firms that don't use them and still make good money, so they're not vital. That said, they can obviously give you some additional edge if used correctly. It is certainly on my list of things to analyze, just not the highest priority right now, though maybe that should be rethought.

So what I'm saying is that they should either be removed (not likely) or just be documented more clearly so they can be used more readily.

I don't know if/when I'll actually read the book. Anything specific in there that sounds interesting besides the different order types?
 
Quote from hft:
>>@hft: Have you read Haim Bodek's book? Do you agree/disagree with his assessments?

I don't know if/when I'll actually read the book. Anything specific in there that sounds interesting besides the different order types?
More...
I forgot you're not trading equities when I wrote that. A lot of the problems in equities trading aren't so bad where you have a central limit order book. Bodek was trying to compete in equities HFT and losing. He thought his problem was speed, but he concluded (after a tip) that it was order type, specifically obscure order types created just for HFT. Speed is necessary ... but not sufficient ... to generate consistent profits in equity HFT. I never understood how speed alone could account for the profits the HFT firms were pulling in. This seems to explain it.

Bodek makes recommendations, one of which is that for-profit exchanges should not be self-regulating. That seems to me a no-brainer. DUH!

Bodek is also critical of the maker/taker model because of the complexity of order handling and market fragmentation.

Some of the reasons I switched from equities to futures/options are market fragmentation, tax reporting, and preditory practices in stock trading.
 
You must log in or register to reply here.
Back
Top