Gem’s bot

Status
Not open for further replies.
ok, Gem have auto system running at http://www.sophocleous.com/usa/webinar.htm

put in name and email and some interest survey questions to get you in, running P/L is in top left corner

use it to confirm your signal or just watch, remember the bot can flip the trade faster than you can

33vhfti.gif
 
I have uploaded today’s data in a personal site that I use throughout this period, where you can find data for the last few months and lots of explanations regarding the system that I use.

If anyone is interested the site is http://www.sophocleous.com/usa/

As of tomorrow you can also see the data live online as it being produced in a webinar that I run on a daily basis. You will find all details in the above site, which with the kind permission of the Elite Trader moderators I just posted. By the way, everything is absolutely free and without any obligations.

It is my intention to have all written discussions exclusively in this forum and it will be my pleasure to answer here any questions or respond to any comments that you may wish to put forward.
 
What kind of scheme is used for dynamically weighting the factors that are applied to the indicators? I mean, what makes the system to overweight, say, RSI over Stochs at any point in time?
 
Frequently I call the Market Strength the backbone of the system. Along these lines your question refers to the Heart of the system. It is indeed the vital point that can make the system work or not.

Most of the coding that is run every 5 seconds has to do with the variable weighting that the system does on the indicators used. There are so many points taken into consideration that even if the full code is published, which I do not intend to do at present, it will take someone a long time to comprehend what is happening.

All I can say is that the Time and Sales ticks play an important role in the whole process and there is a continuous feedback and auto correcting of the weighting factors. This is the reason why I use a short refresh rate when running the algorithm. The performance of the system is inversely proportional to the refresh rate used.

Now that anyone can watch the trades produced by the system live throughout the market hours, it will be educational for all of us if trades produced by other system are compared to the bot. So please watch the video and compare
 
Quote from gem333:

All I can say is that the Time and Sales ticks play an important role in the whole process and there is a continuous feedback and auto correcting of the weighting factors.
More...

So, it sounds like the weights of the indicators are re-calculated frequently so that their combined output reflects the market price. In a way, instead of calculating the indicators with the fixed parameters and using it to predict what the market will do, you seem to be doing the opposite: look what the market has done and from there figure out which combination of indicators would have predicted that market move. In effect, the parameters of the indicators and the indicators themselves are "reverse-engineered" from the current market state. I presume that the underlying premise is that most of the short-term trading is driven by the technicals and therefore figuring which particular technicals to use at any moment is valuable. Does this describe the fundamental principle of your system, or am I completely off-base in interpreting it?
 
Describing the principles of the system can be done in many ways. Certainly what you mention is correct, although your emphasis seems to be in the past tense as opposed to the future.

The system is not trying to see what would have been the ideal weighting of the indicators but tries to establish what weighting will be most suitable for the immediate future. Like all trend following systems the objective is to try to determine when a new trend occurs, stay with it as long as it continues and exit when the trend is either weak or reverses.

I do think that probably we are both talking about the same thing from a different point of view. But even if we see it differently, the proof in the pudding is when you eat it. This is why I show the data that my system produces live (and keep daily records of anything that transpires) so anyone can judge whether or not it is a useful system. I will be the last person on earth to say that it is a perfect system. There is always lots of room to improve it. However I do think it works and it is up to you to judge it.

There have been a number of observations from various people who have been watching the system and I spend lots of time trying to analyze them and possibly incorporate them when I will do the semi annual revision. To that end any suggestion for improvement will be appreciated.

Thanks for your interest.
 
It's an interesting implementation. Thanks for sharing it.

Is there any way to check out the P&L for the day's trade set in realtime. All I see are the open trades -- it does trade a lot I can see.

I tried to get an answer from the webinar but got no response...
 
Quote from gem333:

Describing the principles of the system can be done in many ways. Certainly what you mention is correct, although your emphasis seems to be in the past tense as opposed to the future.

The system is not trying to see what would have been the ideal weighting of the indicators but tries to establish what weighting will be most suitable for the immediate future.
More...

I am still missing the factor that drives the weighting. Perhaps the following example will clarify this:

For simplicity sake, suppose that the system uses only 5 indicators, and they are all exponential moving averages: 1-min average, 2-min average, 3-min average, 4-min average, and 5-minute average. At time t, the price happens to be above the 1-, 2- and 3-minute averages, and below the 4- and 5-minute averages. How do you decide what the "most suitable" weighting of these indicators is for time t+delta?
 
Sorry I cannot be all the time in the webinar myself. After all it is an automatic system. Will try to cover all questions here, particularly on weekends.

To give a simple answer the program may choose that at that particular time the x minute is more important than the others and may disregard the others. It gets more complicated when it assigns a weighting factor to it and then moves to other indicators.

I have to repeat what I have already mentioned above, there are so many points taken into consideration that even if the full code is published, which I do not intend to do at present, it will take someone a long time to comprehend what is happening.
 
The market can only do a finite number of things.

Once you have a method for determining what the dominant trend is in the moment, what you need is have a greater than 50% indicaton of a move being made in the direction of the previously established trend to attain a higher number of winning trades.

It also helps that the information is calculated on a 5 second intervals, it gives the system the appearance of being able to anticipate and front run the market.

Man, this stuff sure is fun!

Best Regards,

JJ
 
Status
Not open for further replies.
Back
Top