Hi Temjin,
I was thinking about the potential of 24/5 trading during the day andvI've decided that this is the direction in which I want to be headed in as I look to the future.
If you are trading big money its probably worth it to pay somebody to look after the system just in case anything goes wrong eg. blackout, internet connection problems, etc. If they need to make an intervention even 1-2 times per month, the money that saves will probably make it worth it for you to employ them. Obviously this depends on the scale of the operation.
They say theres no better sleep than sleeping knowing you're gonna wake up paid
All the best on your journey.
Hey Nizar. Heh, I thought you originally focusing on long term trading systems.
There are lots of virtual or dedicated private servers out there for hire and it is just as simple to upload the trading program via remote access to the server and have it run 24 hours a day, 7 days a week non-stop. My opinion is that it is far better to outsource such requirement to professional providers who are far better equipped and can provide better securities and 99.99% connection uptime to the internet. It is no different to web hosting essentially.
As for paying someone to look after the system, this would only comes when the operation become big enough. I.e. to the stage of getting registered as a CTA and operate a managed future as a full time basis for other people's money.
Thanks, it's definitely needs hard work but who say wealth is easy to create? (I mean to...*cough* earn from others losses) haha
Temjin,
I don't quite understand what your comment about position sizing?
My point is that the goal of a high frequency mechanical system is low variance (risk), the trade off is expectancy. Low variance affords you high leverage, which in turn over comes low expectancy. It's a zero sum game unless you utilize leverage, which increases your exposure.
This one is rather difficult to explain because all these stuff have always been deep in my head.
(there are tons of writers who are far capable of explaining difficult ideas in a simple way)
How about instead of answering your questions, I give you my points for developing a high frequency system.
My understanding of expectancy is that it largely depends on how your system trade (entry and exit) and has absolute no relation to being high frequency or not.
Let's assume I use a basic position sizing strategy of risking no more than 1% of my account per trade. Let's also say I am trading a high frequency and high probability system which yield me an expentancy of 0.4R after cost and give me around 3 opportunities per day on average. (determined historically from valid back testing results)
My expectunity would be 1.2R per day and "in theory", by risking 1% of my account per trade, I would expect to receive a return of 1.2% per day.
How does leverage comes into play then?
As explained in one of the other thread, I insisted that leverage allows me to trade my system by using less capital than otherwise would be needed. I can always limit my total explosure by limiting the number of opened trades.
e.g. 1% risk per trade, I don't open more than 10 trades in any direction at one time, thus, max lost is 10% assuming no spillage in a highly liquid market and no overnight risk. Not to mention no sudden increase in margin requirements, which has been a major risk with the current global credit contraction.
Sorry I cannot explain the leverage side further, I need to get my head around this to explain it more clearly.
One fundamental difference between funds & retail is brokerage - which has a huge impact on this kind of system - not to mention resources, inside information, etc.
Trading costs for currency pairs (only spread, no commission) and futures from discount internet brokers have largely closed the gap of funds' large bargaining power over retailers. But it is true that less spread can give me a huge advantage over time and large quant funds are able to negotiate for better spread.
And I don't see why I have a need for inside information to trade a mechanical system anyway.
Big positions + short holding times requires a market that's liquid + volatile : not THAT many out there and fewer again with enough similarity to be traded by the same system.
That's a worry that I don't mind getting into one day. That's like when I have over a $1 billion to trade or something and I have to start worrying if my trade size will move the market. hahah
I'm interested to hear more about adaptive systems though, I'v read a few journal articles about this but nothing truly adaptive I found was able to make significant profits.
You can try reading
Trading Systems and Methods by Perry Kaufman. It has a section on basic adaptive trading techniques.
As long as your back testing results have determined your trading signals are robust, and they are by itself adaptive to changing market conditions, and that significant changes in parameters (if any) do not affect the overall "edge" of the signals, then you could potentially make a very profitable system out of it over the long term.
Remember that any static with parameters trading systems can be robust and profitable over a period of time. However, they will all eventually meet their death as market condition changes. Adaptive trading systems only extend its longetivity and is by no mean immune from system death as well. There is no such thing as a "truly" adaptive system.
Hope this explains everything.