Hello, I have purchased Deep Learning and would like to train my strategy. I saw that this can be done using the "Send Target" function. However, I cannot find "Send Target" in my menu bar. Do I need to activate or configure something first?
Rename
Where did you see send target? To train a Model you click the Begin Training button. Did you watch the demo video?
Hi. Yes, I’ve seen the video.
Here, however, the author is training his existing trading system into the AI.
I can't find these functions. And sorry for the poor quality.
Here, however, the author is training his existing trading system into the AI.
I can't find these functions. And sorry for the poor quality.
This looks like one of DrKoch’s finantic extensions, not Deep Learning. Let me fix the post title and hopefully he will jump in.
Hi MrMiyagi328,
where did the presentation in Post #2 come from?
Where is Send Target?
The "Send Target" feature is implemented as a "Performance Visualizer" and appears after a strategy's backtest In the "Backtest Results" tab:
It is part of the finantic.IndicatorSelection extension.
(see https://www.wealth-lab.com/extension/detail/finantic.IndicatorSelection)
What does this Send Target?
The Send Target window collects all trades of a trading startegy and sends profit in percent, date and symbol to an embedded database for further use by the IndicatorSelection tool. There is an optional Transformation to remove outliers since these disturb all following algorithms.
What is it good for?
With tools like Indicator Profiler, Deep Learning or Indicator Selection we are looking for indicators that predict the future in some way.
In this context "Future" is defined by some price changes in a trading instrument we hold a position or we plan to hold a position.
The Indicator Profiler and Deep Learning use some shifted indicator (example: Rate of Change (ROC) for the next N trading days) as the "thing" they want to predict.
It is different with Indicator Selection: This tool uses concrete trades of a trading system as the "thing" to predict. This is better than just all ROCs of all symbols, because it removes some of the abundant noise and lets the algorithms concentrate on the price changes we are interested in: The outcomes of our trades.
where did the presentation in Post #2 come from?
Where is Send Target?
The "Send Target" feature is implemented as a "Performance Visualizer" and appears after a strategy's backtest In the "Backtest Results" tab:
It is part of the finantic.IndicatorSelection extension.
(see https://www.wealth-lab.com/extension/detail/finantic.IndicatorSelection)
What does this Send Target?
The Send Target window collects all trades of a trading startegy and sends profit in percent, date and symbol to an embedded database for further use by the IndicatorSelection tool. There is an optional Transformation to remove outliers since these disturb all following algorithms.
What is it good for?
With tools like Indicator Profiler, Deep Learning or Indicator Selection we are looking for indicators that predict the future in some way.
In this context "Future" is defined by some price changes in a trading instrument we hold a position or we plan to hold a position.
The Indicator Profiler and Deep Learning use some shifted indicator (example: Rate of Change (ROC) for the next N trading days) as the "thing" they want to predict.
It is different with Indicator Selection: This tool uses concrete trades of a trading system as the "thing" to predict. This is better than just all ROCs of all symbols, because it removes some of the abundant noise and lets the algorithms concentrate on the price changes we are interested in: The outcomes of our trades.
Remark: The last screenshot (4) in Post #2 shows a preview of the upcoming finantic.Learning extension. This one is not available right now but will be published in the near future.
Perfect, thank you. I was surprised—I had already seen the seminar with the tool on YouTube three months ago.
I have another question about deep learning! After approximately how many runs should you stop if you notice that nothing is happening at all?
I have another question about deep learning! After approximately how many runs should you stop if you notice that nothing is happening at all?
In Deep Learning, it saves the Model periodically when it achieves a better OOS result. So it's OK to let it keep training, but I typically use the auto-stop feature to stop it after 10,000 epochs without an improvement.
QUOTE:
I had already seen the seminar with the tool on YouTube three months ago.
Could you please post a link to that YouTube video?
I am rather surprised it exists, because right now no more than three people have access to finantic.Learning....
https://youtu.be/ifN7VtyGBLI?si=btzG--R3yjyRFws_
Hi, here is the video! It’s a German trader, Thomas Vittner, at WH Selfinvest! Everything is explained starting at minute 13:00. I hope everything is okay like this… 😁
Hi, here is the video! It’s a German trader, Thomas Vittner, at WH Selfinvest! Everything is explained starting at minute 13:00. I hope everything is okay like this… 😁
Your Response
Post
Edit Post
Login is required