Ithink the best available tools for trading, for you puppets ,are the ones that based on metamodeling.The OMG might be among the choices.
Automated Trading - FAQs
- Thread starter Alpha Trader
- Start date
Thank you.
@garachen What do you use to determine your parameters to trade in that case?
It seems like Getco has merged and not doing so well, Infinium did not last and in a contrasting fashion, Jump is doing amazingly well. Below are the links JFYI.
On the note above, anyone that signs with Jump has to agree to their confidentiality clauses and are bound by that in contract. From speaking to another algo trader, it seems that not much information would be derived via this route. However, the creativity is appreciated.
http://www.reuters.com/article/2014/03/06/us-speed-trader-infinium-idUSBREA251Y520140306
http://www.bloomberg.com/news/2013-02-13/getco-profit-plunges-82-amid-slump-in-stock-volumes.html
http://www.bloomberg.com/news/2014-...out-this-story-on-hft-power-jump-trading.html
@jack hershey also suggested something similar and I think this may be a good option to take as I am only starting to discover this pathway.
Is it generally not a big deal because these are very simple strategies/models that tend to work and most traders would take these trades were they trading manually anyway?
And this is exactly my goal!
I question the need for complex strategies with all the bells and whistles attached to it. The links above for Getco and Infinium goes to show that complexity is not necessarily the right answer to success.
My whole idea is to utilize the "easy" stuff that has worked time and time again. The use of automation is solely for the purpose of overcoming my ability to not being able to manage more than an "x" number of stock at any given time or maybe to overcome the missing of setups that I would usually take.
If you are going to use interns, then read the article on page 61, etc in the latest "The Economist" Since they are eager beavers, they may talk you out of doing anything but a complete job.
Any studies or indicator/combination of indicator that with 100% of certainty show the regions throughout the trading session,where,statistically,the price tends to 'churn' a certain price diapason?For e.g.,the price diapason is 20 ticks meaning 20 price levels,what studies are to look at to capture that diapason with the 100 % accuracy?
I have a lot of experience in this area. I worked for a software company next to a college that was always keen to place their software engineering students into our company for work experience. It feels good to give an eager student the chance to build experience facing real-world challenges but even with the strongest of them we found that for the first six months or so they were more of a drain on the team than a help. Quite a few of them worked out well and were eventually hired, but it didn't take us long to realize that the 'free' resources were actually quite expensive and we had to keep the ratio of interns to regulars very low.
This is a great point to consider. I once came into a company that had many years worth of business logic written by their developers as SQL stored procedures. Initially it was a very attractive approach as the stored procedures could be developed and deployed fairly rapidly. As the systems grew in complexity over time the code became much harder to support, debug, enhance, refactor. On numerous occasions when we wanted to integrate our systems with third-party products such as rules-engines or other API's we were f'd because the calls would have to be made from SQL which either wasn't supported or had severe performance impacts.
Jack, you have written many times, “The market is always correct and it has no noise, no flaws and no anomalies.” If the market is always correct, then in a pedantic sense there is no noise, no flaws and no anomalies. How does one reconcile, without being pedantic, observations of noise and anomalies with your quotation above?
There was an afternoon this summer where the ES had a very narrow range lateral lasting three hours or more that sure seemed liked noise. The flash crash in the ES market in May of 2010 and the surprise Fed rate cut affecting the bond market in January of 2001 sure seemed like anomalies.
I ask you, in a very relaxed way, what exactly do you mean when you say the market has no noise, no flaws and no anomalies?
-river
I'm interested to hear Jack's response, but I think of noise/flaws/anomalies as a function of the fractal you are observing. Noise on this fractal? Zoom in or out.
Making money by capitalizing on the system of operation of the markets does depart from business logic in several ways. Secondly, there are no third party products which approach the power of taking the full offer of the market.
To reiterate in a different manner, the SQL for the system of operation of the markets has a core. (25 to 30 "tables") To the right of the core, there are four operations which terminate the segment just completed. This is feedback to the left of the core where three simple operations build a barrier to past data, clean up a data supply minor common error, and reset the system's references.
My next comments will try to deal with keeping on topic in this thread and at the same time putting some extrapolations into the space regarding the limitations of what is perceived to be automation.
My only work experience in a company was in the data systems division of IBM. within two years, I assumed an additional role of part time recruiter. My profile was considered emblematic of what IBM wanted staffing their main thrust. We sliced off the top graduates only. 50% of most interviews were devoted to explaining to candidates why they would not like working at IBM.
Having worked at EOP, I had to deal with large projects where management supported huge numbers of interns. Interns are gathered at presentations. Then they are winnowed rapidly by putting poor candidates on sub projects that will wither quickly. Candidates with knowledge and skills are made leaders of teams and rapidly cross fertilize. days and weeks produce results quite rapidly. Tasking is the key.
A trading system will have about 50 blocks. Blocks come in groups and teams do groups by repetition. Only a few of the blocks in a system are dynamic all others have defined static referential content. The two dynamic "tables" are: 1. the 81 bar "table" of the bar by bar monitoring and analysis and 2. the 15 to 20 record profit segment "table". Both have been published in ET. recently I had a 44 yo learn both dynamic tables in 3 days. He used all of the support tables to conduct the dynamic tables during that time. All the support tables have been published in ET.
Interns doing automation can use Access for most of the pre-coding before commands. If a person is automating his system, then he has all the information prior to beginning coding. In SQL these "guts" of the automation go into "tables". All table are composed of "fields".
All parts of automation require that all the fields are known. If anyone wants to automate, then just get 3 by 5 cards and sit and write out 100% of the fields.
Now, I can deal with the requests of recent posters. AND now, I can suggest how to get interns working at rail speed in a day or two.