Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Ask HN: Is it feasible to do high-frequency trading as an individual?

news.ycombinator.com

Post reply on HN
Print view View original

51–59 of 59 posts

Re: Ask HN: Is it feasible to do high-frequency trading as an individual?

#51

Post by what »

SkyMarshal wrote: post #44
what wrote: post #18

A little OT: People here are always talking about trading stocks, anyone dabble in forex?

A caveat emptor on FX trading worth reading: http://tickerforum.org/akcs-www?post=23331

Yeah, there are lots of bucket shops.

The claims about costs are stated in a weird way. The absolute cost of the trade doesn't change based on your leverage, just the cost as percentage of the margin requirement. I guess they're just trying to say you can blow up your account faster with high leverage.

You can also find brokers that don't do roll over and will even pay you interest on open positions for certain pairs.

Re: Ask HN: Is it feasible to do high-frequency trading as an individual?

#52

Post by wglb »

vecter wrote: post #26
nudge wrote: post #4

It's not a dumb idea, but you certainly should give up. The high frequency game is one you will lose, because many of the large investment banks have relationships with the stock exchanges allowing them to get information faster - and respond to that information faster - than other traders. See this NYT article (including the graphic) for more: http://www.nytimes.com/2009/07/24/business/24trading.html

That article is 85% hype. It reads like an editorial, not a news article. What exactly do you mean by "relationships with the stock exchanges"? If by that, you mean they pay the exchanges for colocation (which anyone can do), then yes, I would agree. If you mean the banks and exchanges are cronies and that the exchanges give the banks free perks because they're golf buddies, then that's wrong. The edge you are talkin…

There are several problems with this article; some errors (or simply incorrect terminology) and some things that are no longer true.

There is a difference between "algo" trading, also known as "program trading" and HFT. The algorithms in HFT kind of boil down to "get it there fast".

The thing that is no longer true is the flash orders (pointed out elsewhere in these comments) are no longer allowed. This is the reference to flash orders.

The trick to being a HFT is to be big enough and fast enough that the exchanges pay you to trade. Thus, you don't necessarily have to make a profit on these trades. Then you are really a liquidity provider. Being an HFT is about size, speed and execution mostly, and ideas not so much. My advice is to not go after that market. It is likely that you are three or five orders of magnitude too small.

Other strategies are more interesting for the small guy. Maybe become a MFT (medium frequency trader), whatever that is.

Re: Ask HN: Is it feasible to do high-frequency trading as an individual?

#53

Post by wglb »

anigbrowl wrote: post #43
lrm242 wrote: post #39

Earlier quoted context omitted.

Good thoughts, and I don't particularly disagree. Regardless of any derivative you might obtain from the data, most folks want to retain high-fidelity historical data for two reasons: (a) Future analysis techniques are unknown. Today you might be using method X, but tomorrow you might want to try method Y which calls for an entirely different massaging of the raw data. (b) Backtesting and replay. Simulations and repl…

Seems like hosting a large hifi dataset in the cloud (eg rolling last 5 years) and charging a small fee to crawl it might be a good opportunity. Or maybe there's no margin in it when people are prepared to pay $$$ as you describe even if they're reinventing the wheel in the process. Not that I'm a mathematical or economic genius of any kind, but I continue to be surprised at how primitive financial analytics seem. Wh…

It is likely that you can't share that tick data--there are redistribution restrictions.

Re: Ask HN: Is it feasible to do high-frequency trading as an individual?

#54

Post by steveplace »

lrm242 wrote: post #23
steveplace wrote: post #20

Since everyone has said no so far, I'll go ahead and say it's possible, but very difficult. HFT algo commission is much lower than $10/trade. IBKR offers 0.008/sh, and it can get much lower if you trade in volume. Like others said, it is a game of milliseconds. So you need to own a server near the exchanges in new york (i.e. http://www.ubiquityservers.com/data-center/new-york.php ). The best way to start learning is…

IB's unbundled commission structure is cheap, but I wouldn't recommend them for HFT. HFT, as defined by the industry, is basically very heavy on the order flow with the majority of those orders going un-executed. IB charges an insane cancel fee for orders that are direct routed. If you let IB route the order with their SMART algorithm then you have no idea where it might land. I'd speculate that IB charges such a cra…

IBs cancellation fee on options is pretty absurd. For the solution needed there are much more specialized firms; however, starting off it's a pretty decent start.

Re: Ask HN: Is it feasible to do high-frequency trading as an individual?

#55

Post by drallison »

High frequency low latency trading is an fascinating area. If you do decide to make a go of it and have the algorithm, the collocation site, the relationships, and the money, you might want to work with Maxeler Technologies, http://www.maxeler.com, which has the acceleration technology that is need to make it happen.

Re: Ask HN: Is it feasible to do high-frequency trading as an individual?

#56

Post by pbhjpbhj »

wglb wrote: post #53
anigbrowl wrote: post #43

Earlier quoted context omitted.

Seems like hosting a large hifi dataset in the cloud (eg rolling last 5 years) and charging a small fee to crawl it might be a good opportunity. Or maybe there's no margin in it when people are prepared to pay $$$ as you describe even if they're reinventing the wheel in the process. Not that I'm a mathematical or economic genius of any kind, but I continue to be surprised at how primitive financial analytics seem. Wh…

It is likely that you can't share that tick data--there are redistribution restrictions.

Stock exchange data is closed? I find that rather strange.

Re: Ask HN: Is it feasible to do high-frequency trading as an individual?

#57

Post by pbhjpbhj »

larsberg wrote: post #37
dhyasama wrote: post #12

Stay away from HFT, big players will eat your lunch. HFT isn't even a game of milliseconds anymore, it's a game of microseconds. You can still do well with algorithmic trading where speed of execution isn't an issue. Work on your strategy, test it on historical data, execute it manually, if it works then automate part or all of it.

Yup. NASDAQ will even tell you what the current latencies are for their coloc: http://www.nasdaqtrader.com/trader.aspx?id=inet Microseconds indeed. Alternatively, you might consider going overseas if you aren't prepared to pay coloc fees and do custom FPGA or realtime work. Some of the exchanges, particularly in Asia (except Japan) still have clearing times measured in seconds.

Why doesn't government step in and quantise trading terms to be a whole day (or similar time)? It seems wrong to me that with potentially equal performance two traders can get different returns based on the microsecond difference in their latency between their server and the exchanges central server.

Does anyone still by the whole line about exchanges being their to create fluidity in the market.

It all seems a big scam, the more I learn about it the less I feel the highest returns go to those who perform the greatest scams fleecing off value created elsewhere (manufacturing, services, ..).

Re: Ask HN: Is it feasible to do high-frequency trading as an individual?

#58

Post by lrm242 »

pbhjpbhj wrote: post #56
wglb wrote: post #53

Earlier quoted context omitted.

It is likely that you can't share that tick data--there are redistribution restrictions.

Stock exchange data is closed? I find that rather strange.

It's not closed, but to purchase re-distribution rights is very expensive.
Post reply on HN