New GFT Foresight AI service - any comments?

kentastisfx,

For years, I shared your opinion of there being no free cheese. However, I have recently changed my mind. First of all, this cheese is anything but free - it's about $500/month. However, that isn't my point.

My point is that if you approach trading as a business, you will soon realize that as with any business, you can't do everything yourself. Somebody opening a mechanic's shop doesn't attempt to make all of his own tools. He buys the tools and if they stop working, he tosses them out and buys some others.

After years of trying to build all of my own tools, I am beginning to realize that building the tools is not what I want to do with my time. I want to use the tools to trade. So I am suddenly in the market for some good tools.

I am aware that there is a great deal of crap out there and that any crook can market his MA crossover as the path to riches. However, there are also some very smart people who have worked on these problems for years and have tools to offer.

I always thought "If the tool is so good, why don't they just trade it instead of selling it?" But I realize now that trading with even the best tool is a risky and frightening process. A smart trader minimizes his risk / reward ratio, and that ratio is MUCH smaller selling a tool than trading it.

So, my eyes are open and anybody willing to let me demo their tools will get my attention. I'm not living in the woods anymore - I don't have to build every damn tool in the shed.

My two cents: I am getting an MS in Computer Science and have worked with Java over the past 4 years. IMHO the markets are NOT pattern driven and the argument that "traders trade patterns, therefore the market is pattern influenced" does not hold true.

The fact is that Forex is really a tool for banks and countries to hedge their bets about the future of the countries economies. Nothing more and nothing less. "Traders" make up a small percentage of the market.

That being said, I'd like to point out the NZD swings this past year. Economists were screaming "recession recession" in Q1, but New Zealand put in a number of provisions that lowered the unemployment rate from 70% to 30% in four months. THAT, my friends, is why currencies move up and down.

Hi Jshav,

I am interested in the idea of Neural Nets. I have created and trained around 145 different networks in the past six months (all with Matlab), and some of them actually do yield acceptable results. I would love to follow up the idea with someone whose brain is heading that way.
BTW, I've never heard of Joone Neural Net. is it just a platform for creating and training nets?

Have you played with FNTs (Flexible Neural Trees?) Supposedly they performed well with currency.

Thanks tdion;

I have not heard of FNTs, but will check it out.

jhsav

Khalilia:
Hope you got my reply, anyway to rehash, and hopefully generate more interest. Joone is a JAVA based neutral network framework with a very good editor, and although an understanding of JAVA would be helpful, it is not necessary. The editor allows one to configure and set try different networks "on the fly". With scripts, one can sequence through a moving window of training and testing data in seconds!!!.
Have not tried Matlab and would be interested your assessment of Joone versus Matlab.
********************************
"It's hard to defeat an enemy who has outposts in you head".

Coach Quincy

The training of NN's can take hours and it can take seconds. The gift of a community of traders such as forexfactory is that someone in all likehood is doing the same thing you and, may have found a more efficient/effective way of doing it.
For example, the Joone framework has a scripting capability that allows one to program both training and testing over a range of values. Don't give up the search yet for the "Holy Grail" the "Philosper's Stone" turning base metal into gold etc. etc.

**********************************************
'It's hard to defeat an enemy who has outposts in you head"

What software do I need to view the files in the zip?

GEOSTONE:

Not quite sure I undertand the question. If you download the Joone framework and unzip it, I think you can install it by going into windows, i.e. Word, clicking on the exe file. It should create a desktop icon for you (if you want).
If you have additional problems, let me know. I will reinstall it on my computer them get back to you since it has been a while since I went through the process.
Sould we start a AI thread, as opposed to a GTF AI thread?

JSHAV
******************
Remeber! It's hard to defeat an enemy who has outposts in your head .

what inputs will you use to train your nn? surely not just historical price action... i'd expect to have interest rates, oil prices, gold prices, etc...

I ran correlations on following pair, EUR/USD,USD/CHF,EUR/CAD/,AUD/USD,GBP/USD,AUD/JPY plus Russell Index, S&P 500, Dow Jones, spot gold and Brent crude. Outside of the usual correlations between certain forex pairs the only other significant correlation was between the price of crude oil and all of the forex pairs. I'm told that bond prices may have significant correlations but haven't checked it out yet.

JSHAV

An interesting read here on the correlation between currencies and commodities.... Focus is on crude oil and gold:

http://www.investopedia.com/articles...Currencies.asp

I would suggest a new thread for your Neural Network ideas. It is an extensive topic that readers will miss if they are scanning the board and see only the original subject of this thread.

The GFT tool that I started this thread about is different than all of my experience with neural networks. It fascinates me for that reason. It is a timing tool more than an "up / down" tool or even a "support / resistance" tool. They do it a disservice, in my opinion, on their website. They present as being too good to be true.

The fact is that it is not too good to be true. It is a tool that requires manipulation and interpretation, just like any good tool. You don't trade it robotically, you compare it to the other signals you are used to trading and make decisions based upon what you know of each.

THAT'S why I started this thread. To hear other people share their experiences with interpreting this tool.

Tdion:

Thanks for the info on the article, very interesting. Looks like a a good web site to check periodically.

I thought this would be a GFT Foresight discussion thread! Everytime I come back it is full of programming stuff on neural networks.

Come on, people. Open a new thread. Or maybe Coach, we should open another fresh thread with information on Foresight alone?

Anyway, your comments are absolutely correct. My experience (of 2 months) using Foresight AI are the same. IT IS FASCINATING. Not the Holy Grail, but a special tool in my box. It's like one of these special tools you buy for construction: you need to know HOW to use it and when to use it. Once you get the grip you are one step ahead.

I agree with you that the marketing hype does not help this tool to take off, because it is a good indicator. I combine it with other trading techniques (and that is how they recommend it and the inventor, Larry Pesavento, too).


Peter.

Hello everyone,
Just the other day I thought of the idea of combining Foresight A.I. with Vantagepoint. I have been experimenting with Vantagepoint since around March 06 when I bought it. What I have been backtesting recently is the idea of using Vantagepoint's Predicted Next Day Highs & Lows and plotting them on Foresight's prediction. The results have been quite promising. Below are two examples....

The first example is from the EUR/USD 12/28/06 Foresight A.I. prediction. Vantagepoint predicted a high of 1.3159 which was already blown out of the water by the beginning of the Foresight forecast which could then recommend to stay out of the EUR/USD for that trading day.

The second example is from the USD/JPY 12/20/06 Foresight A.I. prediction. Vantagepoint predicted a high of 118.56 and a low of 117.56. In this example, the price in the beginning of the Foresight prediction is well below the predicted high of Vantagepoint and well above the predicted low. Foresight is projecting a pretty solid trend for the trading day from the start, signaling a possible buy opportunity since both indictators are in somewhat of an alignment.

Again these are really basic examples with a lot more research and testing to be done. And as I'm sure you all know, Vantagepoint's forecasts are far from perfect but they do give a pretty good general idea of where the market is going and will be the next day. I currently do not own Foresight A.I. and am interested in meeting up with some Foresight users online who currently use the software. Thoughts?

I looked at GFT's website, and the tool looked accurate. They are showing the best predicted models to potential buyers. Much like VantagePoint's website shows the best trending markets after crosses.

For those of you that bought the tool..... how ofter does the tool accurately predict the following day's price action?

Thanks,
Tom

Not always. Just look at how "well" EUR/USD went for Foresight A.I. on 12/28/06.

You don't buy this tool, you rent it for a cool $500/month.

About the 28th. Keep in mind that AI is mostly a timing tool. Notice that the local high for AI matched up with that big drop in the actual price. My support/resistance methods called for that drop and I took it.

Funny thing is that I was in a live web meeting with a rep at GFT as that move was taking place and since I don't have AI yet, that was the only way I could see the curve before the move.

After the big move, the actual price lined up with my support/resistance methods, and since the AI curve showed it as a local bottom, I rode it back up.

I have been demo trading for years. I just opened a live account with GFT and will start using this tool next week. I have a real sense that while not the Holy Grail by any means, it is tool that deserves to be sitting at the top of my toolbox. If I can't make $500 a month with it then I'll chuck it and start looking for another.

Interesting. So if you have a pattern that dips down and then reverses, you wait until what appears to be the completion of the dip on your chart.... and then go long?

Thanks,
Tom