HFT Myths

zmax · Sep 10, 2013

HFT,

Are you using the so called ultrafast/low latency mkt data feeds or creating your own orderbooks from all the connections you have to all banks/brokers out there. If using mkt data feeds can you name the fastest for fx, fut markets?

thank you!

jb514 · Sep 11, 2013

Quote from zmax:

HFT,

Are you using the so called ultrafast/low latency mkt data feeds or creating your own orderbooks from all the connections you have to all banks/brokers out there. If using mkt data feeds can you name the fastest for fx, fut markets?

thank you!

He said he uses direct feeds and doesn't think you can be competitive with consolidated feeds

xbaha · Nov 2, 2013

Hello,

Thanks HFT for the valuable information, i hope you are still around to answer my questions:

1. how long in a calm market it takes in ms or micro sec from the time you send the order, until you receive confirmation of a fill?

2. i would like to know more about the risk profile, how many ticks would be a maximum stop loss for an HFT in average?

3. this could be a stupid question, but i heard that HFT can be tricked to buy @ ask , sell @ bid, and they would do it over and over again, what would be the kind of strategy that tricks HFT to do such a thing?

Thanks.

hft · Nov 5, 2013

Quote from xbaha:

Hello,

Thanks HFT for the valuable information, i hope you are still around to answer my questions:

1. how long in a calm market it takes in ms or micro sec from the time you send the order, until you receive confirmation of a fill?

2. i would like to know more about the risk profile, how many ticks would be a maximum stop loss for an HFT in average?

3. this could be a stupid question, but i heard that HFT can be tricked to buy @ ask , sell @ bid, and they would do it over and over again, what would be the kind of strategy that tricks HFT to do such a thing?

Thanks.

1. Depends on which exchange. At CME for instance, you can get a response in less than a millisecond.

2. 'HFT' is too broad to classify a number of ticks stop loss. There's strategies with stop losses set anywhere from 1 tick to flash-crash type levels.

3. As far as tricking your own HFT, put something in the code that inverts its PNL (MyPnl = -RealPnl). Then let it think it's raking it in by buying the offer and selling the bid.

. Seriously though, flipping a sign somewhere in your code can have this effect, though it's usually mitigated by some risk limit down the line. As far as tricking someone else's HFT, your guess is as good as mine.

xbaha · Nov 7, 2013

hi hft, nice to hear from you,

my questions is regarding globex:

1. what i meant by risk profile is, if your avg profit each trade is $x, what would be the avg max risk $x ?

i also have more questions if you don't mind answer them:

1. What is the average hold period for a wining position, and a losing position?

2. usually when the bid (ask) flips, i see many buy orders coming at the best ask ( sell orders @ bid), i was wondering, are these orders is a 1 trader order? or is it that he initiate a buy, then other hft follow until the price flip to the next bid level? and if it's a mix, what % is it a 1 trader order?

3. you said making market net up in losing little money and earn rebate, how do other hft make the market in globex es ? since there is no rebate?

thanks a lot...

slickpick · Nov 8, 2013

Quote from murphmack:
And how does HFT dink around with the open/close auctions? Any hints? [/B]

This one is no secret, so I'll go ahead and post it for those who don't know.

For VIX futures, if one is so inclined to, someone who isn't me would blast the strip prior to final settlement thus skewing settlement calculation.

lastlook · Nov 17, 2013

Quote from hft:
About backtesting, there are a small minority of people in HFT that lean on backtesting results in market-making, but I am not one of them. [/B]

HFT, this sounds quite odd. How else would you be able to develop and debug a new strategy, improve an existing one, or know that you have sufficient edge to startup the entire operation?

hft · Nov 18, 2013

Quote from lastlook:

HFT, this sounds quite odd. How else would you be able to develop and debug a new strategy, improve an existing one, or know that you have sufficient edge to startup the entire operation?

My definition of backtesting might not match yours. I define it strictly as simulating how a strategy performs through historical data. It's really hard to do this well with a high-frequency market-making strategy. Instead of that, I rely other forms of analysis (and admittedly 'educated guesses').

- Run blind: Simply rest small size and see if it makes money
- Trial and error: Tweak some settings and run it concurrently with a control strategy to see if it's more effective.
- Post-trade analytics: Analyze the trades you got into in order to improve the strategy.
- Historical data analytics: Analyze historical market data, but not necessarily simulated fills as in a typical backtest.

Some of that is quite vague, which is both to not reveal secrets as well as indicative of the ad-hoc nature of the process anyway.

WinstonTJ · Nov 18, 2013

Quote from hft:

My definition of backtesting might not match yours. I define it strictly as simulating how a strategy performs through historical data. It's really hard to do this well with a high-frequency market-making strategy. Instead of that, I rely other forms of analysis (and admittedly 'educated guesses').

- Run blind: Simply rest small size and see if it makes money
- Trial and error: Tweak some settings and run it concurrently with a control strategy to see if it's more effective.
- Post-trade analytics: Analyze the trades you got into in order to improve the strategy.
- Historical data analytics: Analyze historical market data, but not necessarily simulated fills as in a typical backtest.

Some of that is quite vague, which is both to not reveal secrets as well as indicative of the ad-hoc nature of the process anyway.

That sounds exactly spot-on.

rwk · Nov 18, 2013

Quote from hft:
My definition of backtesting might not match yours. I define it strictly as simulating how a strategy performs through historical data. It's really hard to do this well with a high-frequency market-making strategy. Instead of that, I rely other forms of analysis (and admittedly 'educated guesses').

This is something I discovered for myself some time ago. Less experienced traders often look to backtesting to tell them how an idea will turn out if they implement it. Testing doesn't do that very well, and especially not with HFT. Testing can sometimes tell us when an idea is really bad. I believe achieving alpha starts with intuition, and that takes experience.