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

Algorithmic Trading: The Play-at-Home Version

wsj.com

Post reply on HN
Print view View original

101–110 of 120 posts

Re: Algorithmic Trading: The Play-at-Home Version

#101

Post by throwaway161803 »

lukebuehler wrote: post #36

It absolutely can be done. I know a handful of people who are doing well, or have spun a small hedge fund out of their DIY trading system. Having said that, it's hard. Most people simply don't have the bandwidth to do it properly. Parsing daily Yahoo prices and having a R script or two somewhere will absolutely not put you in the "I do successful algo trading" camp. Or running TradeStation with a handful of pair stra…

Intraday data is often unnecessary and, worse, leads otherwise intelligent people to trade on spurious relationships. Recall that a sufficient statistic for the drift of a Brownian (or geometric Brownian) price process over a period is its return over the period. Hence intraday data provides essentially no additional information on the drift of prices, though it is useful for measuring the strength of covariate relationships and responses to intraday events.

Re: Algorithmic Trading: The Play-at-Home Version

#102

Post by sokoloff »

gnaritas wrote: post #88
jeffreyrogers wrote: post #66

Earlier quoted context omitted.

Yeah, that seemed bizarre to me. You'd need to have a huge edge (extremely unlikely) to make money after fees with only $10k. Plus you're very limited in what you can actually trade with that little capital. For example, many options contracts trade for more than $10k.

10k is plenty, you're forgetting leverage; equities I think is generally around 3x, futures usually offer around 10x leverage, forex in U.S. 50x, forex outside of the US, 200x plus.

Calculate out what the projected hourly rate might be, after discounting the gains on a broad-based index fund.

If you can get 30% annually instead of 10%, that extra 20% on $10K would be $2000. If you worked just 200 hours in a year (seems a low estimate), that's a $10/hr "job" and you have a non-trivial risk of a substantial drawdown during the year that could put you near out of "business" (as happened in the article).

Borrowing on margin, you're going to pay at least 7.5%, so if you're borrowing to 3x, you're paying 15% on your account balance. (Small margin loans are higher rate.) - [1]

Forex doesn't have the margin loan problem, but you can't generate much income on a $10K yen/USD carry trade.

Round trip commissions will be a much higher drag on a $10K portfolio than a $500K portfolio.

If you have the talent to develop an algorithmic trading system that can 3x the market, you can make more than $10/hr in any number of easier and riskless manners.

1 - https://us.etrade.com/investing-trading/pricing-rates#margin

Re: Algorithmic Trading: The Play-at-Home Version

#103

Post by sokoloff »

kittenfluff wrote: post #95
murbard2 wrote: post #75

Earlier quoted context omitted.

That's $20,000,000 which is enough to warrant at least 10 people working on it.

Remember that a hedge fund only sees 20% of the profit that it generates on behalf of its investors, and a large part of that goes into staffing and infrastructure costs, not to mention that quant traders like to be paid sizable bonuses (and therefore would not want to work on a trade with a small upside). I find it extremely unlikely (almost inconceivable, in fact) that a hedge fund would divert 10 researchers to wo…

"Standard" hedge fund compensation is 2-and-20 (2% of funds under management and 20% of gains), so a $2B fund would yield $40M in the 2% management fee. That's the "keep the lights on money".

Re: Algorithmic Trading: The Play-at-Home Version

#104

Post by lifeisstillgood »

noname123 wrote: post #43
lifeisstillgood wrote: post #34

Earlier quoted context omitted.

I don't trade but I am interested in understanding what HFT and other algorithmic trading actually is (having been fooled by The Flash Boys into thinking it's something it's not and having been frustrated trying to find out what it really all is. Any pointers gratefully rcvd

Np, lifeisstillgood. I'm busy at work today, so am afraid I can't give full-length explanations. However I'll provide some links to some of the ideas. HFT: Most HFT strategies involves making a market or rebate trading. https://en.wikipedia.org/w/index.php?title=Market_maker&sect... https://en.wikipedia.org/wiki/Day_trading#Rebate_trading They leverage being co-located next to the exchange servers and trading algorit…

Thanks to both of you- lots to sink my teeth into

Re: Algorithmic Trading: The Play-at-Home Version

#105

Post by murbard2 »

kittenfluff wrote: post #95
murbard2 wrote: post #75

Earlier quoted context omitted.

That's $20,000,000 which is enough to warrant at least 10 people working on it.

Remember that a hedge fund only sees 20% of the profit that it generates on behalf of its investors, and a large part of that goes into staffing and infrastructure costs, not to mention that quant traders like to be paid sizable bonuses (and therefore would not want to work on a trade with a small upside). I find it extremely unlikely (almost inconceivable, in fact) that a hedge fund would divert 10 researchers to wo…

Ten people, not ten researchers, but that would be towards the upper limit. Point is, $20M is nothing to sneer at, even for a billion dollar hedge fund.

Re: Algorithmic Trading: The Play-at-Home Version

#106

Post by jwally »

patio11 wrote: post #35
kingnothing wrote: post #32

Are there any free or cheap sources for historical stock data in a computer-friendly format?

I do not endorse trading with it, but for many types of uses, you might take a look at the Yahoo finance CSV "API." It has per-day resolution going back about 15 years or so.

I'd take Yahoo's data with a grain of salt. Was tinkering with options prices there, and the data wasn't clean which lead to a lot of false positives (e.g. a lower strike call was selling for 1/2 of what a higher strike call sold for).

Checkout TradeKing. You have to open an account to use it, but (to my knowledge) you don't have to fund it.

Re: Algorithmic Trading: The Play-at-Home Version

#107

Post by lukebuehler »

throwaway161803 wrote: post #101
lukebuehler wrote: post #36

It absolutely can be done. I know a handful of people who are doing well, or have spun a small hedge fund out of their DIY trading system. Having said that, it's hard. Most people simply don't have the bandwidth to do it properly. Parsing daily Yahoo prices and having a R script or two somewhere will absolutely not put you in the "I do successful algo trading" camp. Or running TradeStation with a handful of pair stra…

Intraday data is often unnecessary and, worse, leads otherwise intelligent people to trade on spurious relationships. Recall that a sufficient statistic for the drift of a Brownian (or geometric Brownian) price process over a period is its return over the period. Hence intraday data provides essentially no additional information on the drift of prices, though it is useful for measuring the strength of covariate relat…

I did not say that the overall model must be built on intraday data.

But most models will simply be useless without proper execution. And IMO, the model should be back-tested with simulating the transaction costs as close as possible.

So the overall model might be fine and interesting, but if you can't close the loop with a proper execution model, which usually affects the overall model as well, you simply don't have much.

Re: Algorithmic Trading: The Play-at-Home Version

#108

Post by lukebuehler »

e15ctr0n wrote: post #48
lukebuehler wrote: post #36

It absolutely can be done. I know a handful of people who are doing well, or have spun a small hedge fund out of their DIY trading system. Having said that, it's hard. Most people simply don't have the bandwidth to do it properly. Parsing daily Yahoo prices and having a R script or two somewhere will absolutely not put you in the "I do successful algo trading" camp. Or running TradeStation with a handful of pair stra…

> Or running TradeStation with a handful of pair strategies or whatever. You can now buy trading strategies from their TradingApp Store. Supposedly, they've been written by professional algorithmic traders. https://www.tradestation.com/trading-technology/tradestation... https://tradestation.tradingappstore.com/search/all/Rating

This goes exactly against the point I'm making.

Those strategies are not a bottom up system that combines a tailored infrastructure with sophisticated trading models. BOTH are needed, TradeStation is absolutely NOT a sophisticated infrastructure. And I doubt that the models are sophisticated enough to rake in alpha, especially in terms of execution.

Maybe if you put in a little money you'll break even with the subscription cost. If you put in a lot, those systems will break down.

Re: Algorithmic Trading: The Play-at-Home Version

#109

Post by lukebuehler »

siavosh wrote: post #23

How is this new? The first time I setup my own linux server was at my parent's house, so I could write a crawler to collect market data in the hopes of applying some AI technique for market predictions. This was 10 years ago. It's a lot of programmers dream to write a little program that can print money. Afterwards I went and worked on wallstreet, and realized how foolish this was.

Yes, it's foolish for most people. But again, there still are some BYO everything HTF outfits, where essentially one man teams still do well - they do get co-loc and feeds usually though.

And there are some that can pull it off at home, algo trading slower time scales, but they are usually programming geniuses.

Re: Algorithmic Trading: The Play-at-Home Version

#110

Post by tunesmith »

karmapolice wrote: post #92
tunesmith wrote: post #70

I've always been confused why so much of algorithmic trading centers on technical indicators. What about fundamentals? Wouldn't it be easier to cobble together a system that checks for healthy companies that are low in their PEG ratio historical range, and then buy-and-hold? There are super-boring companies with reliable earnings history out there, with stock prices that go up and down throughout the year. So you'd b…

How easy is for a bot to read fundamentals? I do not know if there is some place to gather this kind of data. I mean, in "The intelligent investor" the author points out that lots of companies bury important information in side notes in their annual reports, that can totally change their attractiveness. I wonder how easy is to check these things nowadays.

That data set is harder to come by, which really sucks... morningstar has something like that - historical fundamentals - and they charge a ton. But I think Quantopian gives access to it? Not sure if others do. And there might be ways for other services to bundle it somehow across many subscribers.
Post reply on HN