Level 2 usefulness in the age of automated trading
- Thread starter bob2007
- Start date
In US, market depth or DOM is used in futures markets in stock markets its Level II eg you see a big order at a congestion area once that order comes up to the top of Level II if you do not see it on your Time & Sales window than it was a head-fake or a lesser chance it was made up of many small orders; then you should see a slew of small orders executed at that price...
traders have been buzzing about it on forums for a long time. I did it with Sierrachart scripting with regard to DOM total size about 5 deep above the going price vs below the going price.. it was interesting but no more useful than the existing similar indicators packaged with the software...
A lot of the respondents are providing you with misimformation.
You mention a timeframe for trades. what you are speaking about is the period of time you hold a trade from entry to exit.
Ten percent of the document being used by our trading room trading team is devoted to Level II. The person who runs the trading platform uses the Level II as her principle tool.
She logs key moments on a grid that contains 10 values that include and surround BestBid/BestAsk. Six symbols are entered appropriately as a shorthand. Attached please find the symbol list in the lower left.
In a nutshell, the Level II is used as a vernier for carving turns in hold/reversal trading.
As leading indicators signal nearness to an upcoming turn. Level II displays the limit of the momentum of the trend ending. (See symbol W). The trend cannot surmount this value.
Thus, a trader always knows the upcoming trend ending value precisely.
Trading profit segments of the duration the OP has chosen means that he misses 2 out of three W's on the Level II and he just holds through them. He can continue to ooperate this way but he should drop the timeframe of one minute bars.
gmst is incorrect in his suggestion of what the OP should do.
The hold duration (timeframe in the OP's words) will change if he adopts the use of Level II. In such happens, then he needs to use a 30 minute time frame for annotating the geometric containers of price.
Lastly, what makes the Level II work so precisely is the use of the independent variable on the Level II. All of the six symbols on the attachment apply to the independent variable.
What makes the Level II work so precisely is the ignorance of Big Money. In terms of market velocity. Big Money does not know how to calculate the capacity of markets at turns. As a consequence the user of Level II always knows in advance the precise value of the upcoming turn.
Most people regard my posts as gibberish or such. Good for most people. For those with the ability to use logic, then they can read this and understand how to automate it.
My 2 bps...
It clearly depends on your trading strategy. However I think in general there are things humans are good at, which computers are inherently bad at and vice versa.
Clearly processing a subset of R^2s within a large universe of securities is something very poorly handled by the human brain. However, quickly identifying and adapting to complex patterns is something the human brain can at times do better. I think this gets into the realm of certain NP-hard/NP-complete problem sets.
Here's an anecdotal example, a friend of mine who works for a very well known institutional data/software vendor is currently working on algorithm which attempts to outperform/replace execution traders. It's non-trivial (probably NP-hard). Here's the setup to the problem: suppose you have 50 brokers, each has 15 execution algorithms, and each algorithm has 5 parameters. If you work the combinations out, that number is massive. My point is there are just somethings the human brain can do better.
It clearly depends on your trading strategy. However I think in general there are things humans are good at, which computers are inherently bad at and vice versa.
Clearly processing a subset of R^2s within a large universe of securities is something very poorly handled by the human brain. However, quickly identifying and adapting to complex patterns is something the human brain can at times do better. I think this gets into the realm of certain NP-hard/NP-complete problem sets.
Here's an anecdotal example, a friend of mine who works for a very well known institutional data/software vendor is currently working on algorithm which attempts to outperform/replace execution traders. It's non-trivial (probably NP-hard). Here's the setup to the problem: suppose you have 50 brokers, each has 15 execution algorithms, and each algorithm has 5 parameters. If you work the combinations out, that number is massive. My point is there are just somethings the human brain can do better.