- ago
I've read over the description but it just seems to be another optimizer like the ones that already come with WL, or like the Monte-Carlo extension? What's the difference? TY.
0
455
Solved
3 Replies

Reply

Bookmark

Sort
- ago
#1
Neuro Networks (NN) are primarily used for modeling nonlinear behavior such as event data seen in fundamental and economic (FRED provider) statistics. Traditional linear time series analysis (like you see with WL indicators) are not well suited for modeling nonlinear data. So if you employ market sentiment with your WL strategies (a good idea), you should consider building your "sentiment indicator" with NeuroLab. And the output of NeuroLab is designed to look like a standard WL indicator.

Now you can still use NN to model linear data as well, but employing traditional WL indicators would probably be a much faster choice for traditional time series analysis.

There is one other NN advantage. Many traditional time series (linear) indicators are not adaptive. That's a big drawback, and you might want to avoid some of the older indicators for this reason. However, NN models tend to be "inherently" adaptive, which is a "short-term" plus, although the "long-term" minus is that they tend to forget. So that inherent behavior can be a double-edged sword when designing your NN model.

---
I might add there are several AI services that broker sentiment market data generated by NN models. Be careful when subscribing to these services; they are not all the same. You want sentiment data that's predictive--not reactive. Your trading model needs to know the future to be at its best.
1
- ago
#2
https://www.wealth-lab.com/Discussion/How-to-use-NeuroLab-8259
0
- ago
#3
Another explanation:

Imagine, you found a few working filters for your trading strategy.
Say, it works better if the SMA(5) is below Closes, the ATR(2) is above 4.2 and ROC(4) is below -0.2.

In short: there is some complex relationship between the indicators SMA, ATR and ROC and the profits of your strategy.

This is a situation where "Machine Learning (ML)" (or in more fancy terms "Artificial Intelligence - AI) can help you.

These algorithms are able to "learn" the complex relationship between a set of indicators and profit of your trading system.

After the learning phase you may use the algorithm to predict future profits: You give the indicator values as input and get the predicted profit as output.

Of course you'll enter the trade only, if predicted profit is positive!

Sounds good? Yes it is!

And a neural net is just one of many ML/AI algorithms. There are more to come to WL...

2
Best Answer

Reply

Bookmark

Sort