Is it possible to hire a Wealth-Lab expert?
Author: Comodco
Creation Date: 6/15/2015 6:59 AM
profile picture

Comodco

#1
Hi,

I am wandering if there are any wealth-lab experts that I can pay to assist me with my general ongoing questions, e.g part time or on a per hour basis, and is anyone else doing similar?

I'd prefer to focus on the "non-wealth-lab" aspects of my strategy rather than the coding and wealth-lab specific implementation required to code the models before testing them.

For example, I'd like to build an index of cement manufacturers/builders etc listed on XYZ developing country index (where I am currently residing) and test their price correlation to monthly housing approvals data of that country.

What I'm looking to do generally is outside the wealth-lab community and it feels a bit unreasonable to ask for help all day long on these matters.

No doubt I'll become more familiar in time but my intention is more to invest profitably than be a wealth-lab expert per se.

Does anyone have any suggestions for me here? Should I employ a programmer on a part time basis or am I better to cut to the chase and start with someone with real experience in wealth-lab who can give me faster and more accurate answers?

Thanks very much, Brian
profile picture

airation1

#2
Excellent, excellent question! I would like to join your quest Brian! The Wealth Lab idea has lured me into moving money to Fidelity in order to use it. The idea of building a rules based strategy with a simple wizard that uses standard indicators is great! I'm a rules-based trader,and building/controlling my own strategy is very appealing. I have built an automated trading strategy with an independent coding contractor for my forex trading program. It works great and I have traded it daily for 6 years.

I have 40 years experience in trading. I know what I want to do. I have experience doing it manually with several brokers, and in my automated MT4 trading program in forex. But, getting Wealth Lab Pro to function automatically with equities is a different story.

I am not a programmer and don't want to be! Trading and programming are 2 distinct disciplines. I tip my hat to those of you who are both! If you want to excel at something, you have to focus on it. So, programmers out there,,,, we need to come together!
profile picture

KEVINP

#3
Why are you looking for a Wealth-lab expert if your focus is "non-wealth lab"? Are you instead looking for a trading systems expert?
profile picture

akuzn

#4
Friends, just has got an email from Wealth support team and by the luck have loaded main page with hyper link about expert.
So let me combine "exchange idea" invitation with some true words.
--
Wealth-lab is just an apllication with charting capabilities, strange optimization interface and absolutely strange (would say idiot) portfolio simulation. You cant simulate real money flow using standard features. Let s say standard design is technical analysis tool based on super modern language and compiler.
So if you really want to do something valuable - certainly you have to start with model (financial, trading etc), draw it by stylus on the paper :).

I ve been trying to use Weatlh and i think it is the most interesting app mostly because of c# and Bill Gates developing .Net.......
But in reality noone metastock, amibroker and even WealthLab will help you.

First of all you have to start from beginning: are you mining for strategies? you have ideas that need automation? etc etc.

But in general i would like also add additional 50c - no one neural network, genetic optimizer will help you, if you have not found explanatory features or variables.....
As for neural networks in Wealth and in general - approach is very slow, 30 - years old ideas from maths....If you are familiar with Nobel laureat - father of machine learning Gerbert Simon - he was always against this idiotic approach. Yes in theory every function may be presented as linear combination of other functions .......................bla bla bla bla. Every data set may be fitted by conveniend polynom of 5000x power or higher. Isnt it idiotic? Certainly he was right... But many people are thinking if they load computer, let it work it will resolve something.
And at the same time there are much faster methods and simplified approaches.
Modern statistical theory, machine learning theory, artificial intelligence just remember it but for example nobody in machine learning world cup use it.
There are methods known from 1995 th resolving classical XOR progammers problem in 200 x faster than any minimal convenient neural network...After using tricks like caching, sparsing etc faster and faster.
Regression - is the first task for all learning, fitting, classification etc..............
In Wealth Lab you do not have any of these modern features.

What i m really using - C# 2015, i have developed general testing application and unfortunately had to recreate everything from beginning.Design, thinking, methods....
1. If you prefere human style thinking approach - human questions and answers - genetic programming - reconstructing formulas of expressions from giving dataset - first test Pithagor formula in 1 - 10 second, bid ask diffusion...30 pages of differential equation programming or some minutes of GP ( i use modified version wich works with expressions stack like compiler) giving similar symbolic result ...
Another application - reconstruct combinations of dynamical expressions wich can be even as genetic algorithm, another genetic program, expression with dynamically optimized parameters. And it gives possibility to keep knowledge presented in expressions: logical, mathmetical and other wich can be created......But in online trading this solution is too expensive.
If you are affraid to miss something - just let s mutate existing rules or their parts.
Or if you want to resolve old options exercise: why seller of put options delta neutralised by long position in underlying asset is always in profit?
Or for example in wich conditions it will happen in current market?
I think everybody wanted it looking in WealthLab GA interface :)
Science in general calls it symbolic regression.
2. Functional Regression. Modern approache is very difficult programmatically because you have to work with linear algerba.
But what is most valuable - online algorithms or even better say online least squares for different task...........Modern online algorithms may be well extended for large dataseries mining ....
Naive linear algebra compared to optmized even under C# has 200x speed up and more, but without native it is nothing - C# doesnt have any possiblities to really use SSE and AVX CPU registers - i would say this only one "-" of C#. But the difference is significant. Best Matrix * Matrix i ve got under c# unsafe, super optimized pointers version and 8 cores notebook - about 4 GFLops, and only one core but native compiler 5.13 GFlops. That means i could use 8 parallel hypothesis testing. GPU for large operations is more attractive. The most funny think - it is not too difficult to call C/C++ code from C# - but still have to spend some time......It is mostly about data sets in backtest.
As for me i switchted to online recursive versions - they dont need to reconstruct new matrices and are super fast, do not consume memory - garbage collector is left without job....
--
Btw modern approaches to regression erase needs of all technical indicators and are killing many old methods of math analysis!!!

--
Interesting fact - author of adaptive boosting has decided to partcipate in small tournament. What he did: decision trees looking for extracting rules based ot technical analysis tools like RSI, stochastic and other s...t. - kind of genetic programming. The most funny thing - it was rare situation when guys from university have proved mathmetically that if Adaptive Boosting has weak learners as input with 55% pronostic probability, enseble will give you 99% probability for classification or regression. Another team was using more natural inputs like orderbook data and was working mostly with regression models......And btw no one has used WealthLab - MathLab, C++.......
Certainly well found explanatory variables approach has won...No one test with technical indicators even done by strong academical researchers doesnt give any significant result.

--
So Wealth - is super charting application with some extensions.
Weatlh Lab will not help you it data mining. It doesnt have anything for data mining, modern machine learning and can be helpful mostly in chart decoration.
Strategies, features, portfolios - you have to do this job manually and visualize it in WealthLab.
Modern user interface is highly complicated - it will save you a lot of time.

But better start from beginning. What kind of index, formula, how it will be presented?
What kind of cashlow management, of results etc do you expect?

If you know the formula - i am sure final code will need not more than 20 lines and you ll get it easy.
But do not expect to much:)