ATMA Discussion Forums
ATMA Discussion Forums for Affiliates and Members
I have a fair level of experience in strategy automation and algo trading. Please feel free to fire up your queries.
Some benefits of automated trading include:
1 Save Time
2 Save Money (trading costs)
3 Minimize Emotions
4 Inculcate Discipline
5 Intellectual Engagement
6 Minimize Human Errors
7 Reduce Gambling Tendency
8 Diversify Trading
9 Deploy Complex Strategies
Accepted AnswerSushil KediaOfflineIts wonderful to see Saurabh Lohiya taking over as the chief moderator of the ATMA Discussion Forum. Back to the point of this post: Saurabh, What would you enlist as the top disadvantages of a systematic trading approach vis a vis a discretionary trading approach, since nothing comes for free. If there are benefits of algorithmic and or systematic trading they must be coming at the cost of giving up somethings? What are the things we are giving up when taking the algo approach?
Accepted AnswerThanks for the warm welcome. Algo trading is simply the deployment of new technology in trading; technology by itself cannot be disadvantageous, the problems comes from the wrong deployment of a particular technical solution. So to talk about the demerits of automation can be done only on a case to case basis.
Accepted AnswerVignesh EswarOffline
Very generous of you Saurabh. The big stumbling block to adopt Algo- based trading is, in my view, lack of enough felicity with coding and back-testing trading systems, for most. Unless you have a background in programming, the learning curve is steep and often slippery. I much appreciate any pointers you can offer about this.
Most good Technical Analysis software offer a Array Processing lanaguage for easy coding of strategies. Which means you can operate on complere arrays of High/Low/Close/Volume etc. using a simple line of code.
For example, MA crossover can be simply coded as:
where MA stands for Moving Average. This is the code for AFL (Amibroker). Similarly TradeScript (Modulus) or EasyLanguage (Multicharts) are easier to learn within a year.
There is still one problem- it will be hard to find regular classes or structured courses for these languages, because these do not find much traction in mainstream computer science. But this can be easily overcome by taking help from professionls... that is why these forums are important!
Accepted AnswerPrashanth KrishOffline
To me, the big issue is not with constructing the programs themselves but that of Bad Data. Data quality for running a Algorithmic based Trading system are higher than one where there is a manual construct between the program that generates the Signal and the process of putting that into the system as a Order.
Most simple programs cannot distinguish between good data and 1 single bad data tick. But that one tick can be enough to trigger a trade which shouldn't have happened. While we are able to ignore those when there is a mix of Algo and Manual, a fully Auto setup would need detailed procedures including crosschecking with secondary sources when data for current bar is way away from last bar.