Home Creators Posts Import Register Favorites Logout
haven't archived this post yet. have a subscription? use the importer!

Content

Bamboo-ta

As some of you already might have noticed. In my Bollinger Bands Trend indicator trading strategy I made use of a couple of indicators that are not available in other TA libraries.

Instead I created my own Python library that contains the technical indicators I like and think are worthwhile to keep.

At the moment I am at version 0.1.1 of this library, which already contains many indicators.

However this library is still in it's infancy stage. Which means that a lot of development has to be done.

For example, I collected a lot of indicators that were included in strategy code found left and right in Freqtrade trading strategies. And why not collect them into a central Python module, for ease, comfort and a standardized way of calling these into current and future trading strategies.

Bollinger Fast Trend Indicator with MA changes

To make everything standardized, this also means to reassess older indicators that were already used and distributed. In this specific case I also had to change the way the Bollinger Bands Trend and Fast Trend indicator is called from the Freqtrade strategy code.

To be specific. In the earliest versions of the bamboo-ta library the colums in the indicator dataframe were called with capitals (see BBTrend etc below).

But to make everything consistent and also make a visible difference between calling the Indicator and calling the columns from its created dataframe, I changed this to lower case letters.

See image below:

This also counts for the MACD indicator in this case. I created my own MACD indicator and call this from the bamboo-ta library. In the old way these were columns like 'MACD' and 'MACD_signal'. But with the new default way they are called by their lowercase 'macd' and 'macd_signal'

So if you update the babmoo-ta library to the most current version (0.1.1 at the time of writing) then know that the Bollinger Trend strategies might get broken by this upgrade.

Python module and GIT repo

The python module can be downloaded and installed with: pip install bamboo-ta

And all the code is available in my github repository: https://github.com/DutchCryptoDad/bamboo-ta

Future Ideas for strategies with this module

There is also another reason why I created this python module...

I want to also focus more on trading strategies from other sources like Youtube videos and Tradingview. And a lot of these sources use indicators or ideas that are not available in the current TA-libraries. So I decided to make this a double-edged sword.

I can now create reusable indicators in a module that can be used by anybody. And due to their simplicity they can be used in regular data analysis and also Freqtrade strategy building (maybe other bots too...)

Besides this I can now also learn more on programming with Python and development as a process. Which can be beneficial for creating more advanced trading strategies.

Suggestions and ideas

Your suggestions for trading indicators or other specific functionalities on trading are welcome. But I will handle these on a best effort (and opportunistic) base.

Reported bugs will be handled with a higher priority though. Don't want to have a faulty strategy ;-)

But first...

But first the holiday seasons are coming and it will be time to recharge. Most of you know that I have a busy family live and also a normal day job besides this venture. So after my vacation I will pick things up where I left these and will continue to search for the best and most profitable trading strategy/algo.

Btw. The forward testing trading bots are still running but I stopped streaming. I will come back with updates in the future on Patreon though

Thanks for being my Patron and until the next post!

Files

Previews only

Comments

Jan

Hello Dutch Algo Trading, I've been using your website for nearly a month and find the information very valuable. However, I'm struggling to run the strategies as many use functions that are renamed or deprecated in the latest version of the Freqtrade bot. This requires me to fix numerous errors in the code. I tried using the Docker image provided in the "Shop," but it always pulls the latest Freqtrade image. I have also attempted to fix the code in the strategy.py file, but the dry tests have been disappointing. In 25 days, only one strategy has been profitable, and the rest are not. I am currently dry testing the top 5 strategies from your website, and maybe I made some errors in the code. Could you please upload the updated strategies, at least the top 5 from your ranking? Alternatively, specify the Freqtrade version required to run these strategies. Thank you.

Dutch Algotrading

I have made a special post and video specifically for this problem. It's true that the development of the bot surpasses the adaption of the strategies themselves. There is a function that updates code to the bots version, but does not take additional modules etc. into consideration. This is why I still run version 2023.7 btw...

Marc

Same here! Honestly I love your tests but the patreon access is a bit questionable in my opinion. I can just confirm: 1. it's not very comfortable to test your strategies as you got to fix each one 2. i don't see a point in installing your own package, especially I don't want to create a new Dockerfile just for that 3. I got errors trying your strategies, e.g. ``` /freqtrade/user_data/strategies/Obelisk_Ichimoku_ZEMA_v1.py:179: FutureWarning: A value is trying to be set on a copy of a DataFrame or Series through chained assignment using an inplace method. The behavior will change in pandas 3.0. This inplace method will never work because the intermediate object on which we are setting values always behaves as a copy. For example, when doing 'df[col].method(value, inplace=True)', try using 'df.method({col: value}, inplace=True)' or df[col] = df[col].method(value) instead, to perform the operation inplace on the original object. ``` 4. my backtesting was way different then yours, e.g. - BB_RPB_TSL_RNG_2 and Obelisk_Ichimoku_ZEMA_v1 triggered only 14 trades in 624 days - NASOSv5_mod3 only 2 trades in 624 days (using Bybit Backtesting data for ETH and BTC) 5. download files are pretty messy not sure what I do different than you but there were also no hyperopt files so it should be pretty much the same?

Dutch Algotrading

I agree that there are a lot of issues nowadays with the strategies. Let me explain them point by point: 1. FT (Freqtrade) is always developing and sometimes will deprecate older code. Contact the developers of FT about this inconvenience. This is also the reason why I currently stick with version 2023.7. It just works... And i made a video and a post about backwards compatibility for this reason 2. I created my own python package since other packages (TA-lib, pandas-ta etcetera) do not have some of the indicators I develop. Yes my package is not included in the default FT docker image, but I posted solutions about adding additional modules to the Docker images. Btw. did you know that most of the indicator packages are not updated over several years now? 3. See the Internet about this issue. If I test this and can publish the results, I can run it. So it's probably a 'local' problem. 4. There are numerous causes for this like other config files, pairs, datasets, timeranges and more. I use multiple machines and even they sometimes have different results. Hence my disclaimer to DYOR. 5. Its all in one package and filenames should be obvious. I just add the complete output package of my tests as a 'prove' of my results. I know not a solution to all of your questions. But at least you know the reasons...