In this issue:
I. Commentary by Steve Ward, CEO
II. This Trader Rocks
III. New add-on from Meyers Analytics
IV. Bowfort IBFeed Professional 3.1 Released
V. Bowfort Datafile Utility 1.1 Released
VI. Offices closed Labor Day
VII. One way to stop this newsletter
*******************************************************
I. Commentary by Steve Ward, CEO
Lately we received four emails from NeuroShell Trader users (well we received many more than that, but four I want to discuss).
Number 1 was from a user who has had the product about 5 months. He seems capable of using the software, but unsure of how to actually implement some of the more complicated strategies he wanted to build. He asked if we would take a programming contract to build them.
Number 2 was from a hedge fund user who is building trading models in NeuroShell Predictor. He was having problems understanding what was wrong with his models, and making some fairly common mistakes that NeuroShell Trader users get over in a few months.
Number 3 lives in Europe and was starting to use the software again after about a 2 year layoff. He wanted to know what he could read or what courses he could take to get some advanced knowledge about neural nets.
Number 4 wanted a list of our Trader customers in his area so that he could find a partner to work with.
We replied to number 1 that if we do programming contracts, we charge quite a lot, and that it would be much, much cheaper to attend our upcoming seminar where he could learn and ask questions not only of us but of other users as well.
We replied to number 2 the trading concepts he was struggling with were exactly like the ones that NeuroShell Trader users struggle with, and that since he lives on the west coast the upcoming seminar in Tucson would give him the boost he needs not so far from home.
We told number 3 that college courses and books delve mostly into the mathematics with little if any practical knowledge about how to use nets, especially when it comes to trading. We suggested he come to our seminar since already we have users coming from UK and Singapore.
We told number 4 that we do not give out user information because of our strict privacy policy, but that seminar attendees usually meet a couple dozen of like minded traders and often become long term friends with them.
You shouldn’t need a neural net to see the pattern here. Sometimes you have to strike while the iron is hot, as they say. http://www.neuroshell.com/seminar.asp
*******************************************************
II. This Trader Rocks
Leonid Velichkovski is using the same determination, curiosity, and artistic talent that made him a rock star in Russia in the early 90s to become a successful trader today. As part of the synth pop band Technologia, he released several albums, toured throughout Russia, and shot dozens of videos. Leonid played keyboards and was one of the two songwriters in the group. After several demanding years of touring and producing new material, the band slowly broke up.
About three years ago, Leonid got interested in trading. “I took trading classes with a very well known company. After that, I’ve been taking private lessons with Mr. A.Safin who was the author of a number of known trading systems. These systems are very effective and profitable. Mr. Safin taught me about financial markets and Forex. This got me very interested and excited about trading.”
“I had started trading real accounts several months after I took classes,” according to Leonid. Obviously, at the beginning I’ve been trading demo-accounts. After I gained a confidence in myself I turned to trade real money. But what I was taught in classes were ‘manual’ trading systems. In other words, you analyze the market using regular technical indicators and go through a complex process to make a decision about opening a position. In order to get an advanced understanding of what neural nets (NNs) are I took a class “Neural net applications in financial markets” taught at Moscow State University by NeuroProject (NeuroProject, Ltd is an authorized dealer of Ward Systems software in Russia). The class is led by Sergei Dolenko. Later I privately consulted with him.
“In order to get good understanding of neural nets I started with NeuroShell 2 and only then switched to NeuroShell Trader”, said Leonid. “NeuroShell 2 had helped me to understand how neural networks worked in general, the different types of nets and the nuances of training a network. NeuroShell 2 helped me a lot and I am glad that I started with it. After I switched to the Trader, it was much easier for me to understand how neural networks and the genetic optimizer work. I also learned a lot from technical support provided by Ward Systems. I even traveled to the United States for consulting from Steve Ward and Andrei Deviatov, who gave me a great deal of attention and assistance. On the other hand, even if you do not go to trading classes and just read the help file, ask questions through tech support and follow the rules, then you can also learn how to trade with NeuroShell. If you manage to understand and find a common language with the program, the NeuroShell Trader will do wonders for you!”
“There is a myth that you must be a neural net guru or a mathematician-programmer geek in order to use NeuroShell,” said Leonid. “As it turns out – not at all. Trading is mostly an art, just like song writing is an art. They have a lot in common – one needs to find non-trivial solutions. This is even truer when you use neural networks. Inputs in the neural net are just like notes in the song. In music, if you take a correct chord, a collection of notes, you’ll get a good sound. If you take random notes without thinking, it will not sound right. The same is true with neural networks. If you feed correct inputs to the net – it works well. If you take a random set of inputs – the net does not work. ”
“If a song is long and boring – you don’t want to listen to it, continued Leonid. “If a song is too short then it passes through you and you don’t even notice it. This is similar to neural net training. If you make the training period too short, the net will not learn anything. If the training period is too long the net will over-learn. Very similar,” according to Leonid. He noted that building a good trading system is like writing a good song. “They both bring money.”
Leonid trades FOREX, a market which he describes as complex and with frequent rule changes. “FOREX is very interesting and causes you to make non-standard decisions. Only neural nets allow you to accurately predict the currency pair GBPJPY using two separate pairs, GBPUSD and USDJPY. The neural net finds rules very well. May be that’s because the GBPJPY pair movement is represented by the GBPUSD and USDJPY pairs.” He’s also going to try stock trading in the near future.
Leonid currently uses intraday bars that range from 1 minute to 1 hour. “But my current trading is inter-day; that is a position cannot be closed the same day it is entered, only the next day or the day after that. I don’t like using daily bars because trades are infrequent and therefore not very dynamic. You need to wait for a long time and I don’t like that. My normal trading pattern is to make 2-4 trades a week on 1-hour bars. In the future, I’d like to try 1 minute and 5 minute bars and do a true intraday trading.”
“I am a big fan of NeuroShell add-ons, I own all of them. I also like newer indicators such as Causal Singular Spectrum Analysis (CSSA) from NOXA, a very promising, high quality add-on. It is, so to speak, the 21st century add-on – no one has made anything even close to it. I also like NOXA’s Entropy Indicators – they are very good for building predictions. I like add-ons from Meyers Analytics and Bowfort. Bowfort has Bowfort Adaptive Indicators – they are all adaptive and work well on small time-frames. One needs to note Meyers Analytics’ “The Nth Order Fixed Memory Polynomial System” – it works well on FOREX markets. Recently Dennis (Meyers) has released a new add-on “The Adaptive N Cycle Goertzel-DFT System”. I’m studying it at the moment. I think it is very promising.”
“Obviously, I am not a “trading wizard”, I keep learning. But I am very satisfied with how NeuroShell trades my systems. The most important thing is to understand and feel NeuroShell. Each trading system requires an individual approach. A trader must feel it and trust it. Even most savvy traders make mistakes, and so do I. But making mistakes happens to me less and less often now. It is just that you need get used to the program and get a feel for it.”
“As I said before, trading classes taught me about ‘manual’ trading. This implies that the trader must posses high endurance, colossal patience, constant attention, and steel nerves in order to make timely and correct decisions about entering the market. This is not very simple. In the end, I understood that ‘manual’ trading is not for me. On the Internet, I ran across the so called ‘intellectual’ trading – the NeuroShell Trader. A trader gives up his right to make decisions to NeuroShell – thus liberating himself from such a nervous and difficult action. And by saving his nerve cells (which we all know do not regenerate), the trader prolongs his life which is not so unimportant in our difficult times!”
“My favorite tool in NeuroShell is the prediction capability. It works very well when you select the correct inputs. I also like using the Neural Indicators add-on as a part of trading strategies. Neural indicators do not predict anything, they just follow the market. They are especially good for straight pairs (currency to dollar). And the most important thing is the genetic optimizer in NeuroShell, which finds correct solutions much faster than a regular optimizer.”
Leonid says judging the success of a model is a very delicate and complex issue. “Each trading system has its own criterion of success, and there are more than one. First, look for a gradual rise with no significant bumps or drops in the system equity curve. Second, look for precise market entry signals, even if they are slightly premature. Third, have a small number of inputs to a prediction. And, most importantly, do not over-train and over-optimize!”
*******************************************************
III. New add-on from Meyers Analytics
The Adaptive N Cycle Goertzel Discrete Fourier Transform System
The Adaptive N Cycle Goertzel Discrete Fourier Transform (nCycleGZ) System for short term & intraday trading This package contains the advanced mathematical technique called the Goertzel algorithm for discrete Fourier transforms. This mathematical technique is currently used in today’s space-age satellite and communication applications and is applied here to stock and futures trading.
While the mathematical technique called the Goertzel algorithm is unknown to many, this algorithm is used everyday without even knowing it. When you press a cell phone button have you ever wondered how the telephone company knows what button tone you pushed? The answer is the Goertzel algorithm. This algorithm is built into tiny integrated circuits and immediately detects which of the 12 button tones(frequencies) you pushed. Here we will use the Goertzel algorithm to detect the frequencies of the price series we give it.
Previous researchers using MESA constrained themselves to using MESA to find only the cycle with the highest amplitude and called that cycle the dominant cycle. In this system, we will find and use the frequencies with the N (N is a strategy input selectable from 1 to 20) highest amplitudes to create a noise filtered signal curve that we will follow to create our strategy buy and sell signals. The Goertzel Discrete Fourier Transform (nCycleGZ) super fast DLL finds the N cycles(frequencies) with the highest amplitudes at each price bar, and creates a x bar ahead(x is user selectable) noise filtered projected momentum curve. This process gives a more robust noise filtered signal than a single frequency (dominant cycle) procedure. You are no longer constrained to using only a single frequency. The nCycleGZ algorithm is faster and superior to MESA in finding cycles in noisy price series.
The nature of intraday price movements is constantly changing due to current economic surprises, events and trader sentiment. Also, the time of year changes the nature of intraday markets, such as the seasons, holidays, vacation time, etc. As such, the periods or frequencies found on intraday prices three months ago may no longer be the same as the frequencies found on today’s intraday data. We expect the frequencies found on intraday data to vary over time.
For The Adaptive N Cycle Goertzel strategy we create an indicator that walks forward one bar at a time. The strategy will take a fixed number, M, of closing prices and use the Goertzel Algorithm(GA) to find the N frequencies with the highest amplitudes. Using those N frequencies, amplitudes and phases, we will construct a new price that forecasts the price x bars ahead. We will save this x next bar forecast value, or forecast point. Next, we move the M closing prices forward one bar adding the next bar and dropping the last bar so that we have exactly M closing prices again. With this new M closing prices window, we compute the next bar forecast value from the N frequencies found by the GA and save it. We keep marching the M closing price window forward one bar at a time, calculate and save the new forecast point until we reach the end of our data. We then connect all the generated forecast points to produce a curve that creates the N cycle next bar forecast as the N frequencies used to create the next bar forecast change over time. Thus, this curve adapts to the closing price’s changing frequencies and projects x bars ahead so that it’s lag is minimized when things change.
The nCycleGZ system is orientated to short term trading in all bar ranges (1 min, 5min, etc., daily) and with any tradable price series. The Input parameters to the system and indicator are changeable and optimizable so that the user can develop his own parameter set on his price series and time frame of interest. Although the nCycleGZ system gives results for the parameters that the E-Mini future was tested on, the user can easily use the system on any future or stock or on any time frame.
For NeuroShell Trader/DayTrader Pro, the Trading Strategies and Indicators are directly imported into NeuroShell via a special setup exe file and are fully disclosed in the Manual, Indicator wizard “MA_nCycleGZ” category and in the Trading Strategy Wizard “MA_nCycleGZ” directory. The C++ DLL code is not disclosed. The Input parameters to the strategies and indicators are changeable and optimizable so that the user can develop his own parameter set on his price series and time frame of interest. Although the strategy results will give parameters for the intraday or daily futures the strategy was tested on, the user can easily use this system on any tradable or on any time frame.
The accompanying manual consists of:
How To Setup A New chart With Data Series, Indicator & Trading Strategy.
A complete description of the N cycle Goertzel Indicator & Strategy Input Parameters
The N cycle Goertzel Strategy Applied to E-Mini (ES 5 min bars)
The Adaptive n Cycle Goertzel DFT Math and Derivation.
A Summary Table of All Strategy and Indicator Names
For further information please contact:
Meyers Analytics
1622 N. LaSalle Drive
Chicago, IL 60614
(312) 280-1687
in**@me*************.com
www.meyersanalytics.com
*******************************************************
IV. Bowfort IBFeed Professional 3.1 Released
Bowfort Technologies has just released a new version of its data feed for Interactive Brokers. It now supports Tick / Range / Volume bars with up to 1 year of historical data, as well as Second bars and all the usual timeframes. IBFeed Professional is the only datafeed for NeuroShell that provides Market Depth / Nasdaq Level 2 data streams within NeuroShell. These new data streams enable you to determine bias in the order book around the current price, a natural choice to feed into the Neural Networks in NeuroShell.
The data is free if you have an IB trading account, with the addition of a monthly fee for the datafeed interface from Bowfort Technologies (IBFeed). Another advantage is that you can analyze the same data you’re trading (removes the disparity between separate brokers and data providers). IBFeed can also be combined with the automated trading feature in NeuroShell DayTrader Professional release 5.4 (a beta for release 5.4 will be available soon).
For more info visit http://www.bowfort.com/Products/index_html
*******************************************************
V. Bowfort Datafile Utility 1.1 Released
Bowfort Technologies has just released a new version of it’s .wsg and csv import utility for NeuroShell. This new version now supports importing from Metastock and MT4 format files. This utility can be used to export / import NeuroShell’s datafiles (“.wsg” files) to/from Comma Separated Value Format (“.csv”). You can also truncate your “.wsg” files, and batch process many .csv/.wsg files in one operation.
For more info visit http://www.bowfort.com/Products/index_html
*******************************************************
VI. Offices closed Labor Day
The Ward Systems Group offices will be closed Monday, September 3, in observance of Labor Day.
*******************************************************
VII. One way to stop this newsletter
It is really easy. Just change your email address and don’t tell us.
*******************************************************