A book like Trading Evolved will without a doubt have errors. On top of errors, updates may also be needed as new versions of dependent software packages are released. I will keep a running log on this site of issues found, and suggested solutions. After this list grows long enough, I will update the book itself, and put a version number on it.

This article will be continuously updated.

 

Book Version 1.0

Version 1.0 of the book was released on August 8, 2019.

Unable to Install Zipline with Conda version 4.7.10

Conda version 4.7 seems to break some dependencies and may result in an inability to install Zipline using the syntax “conda install zipline -c quantopian”.

Workaround 1 – Downgrade Conda

https://github.com/quantopian/zipline/issues/2514

Workaround 2 – Install using pip

https://www.zipline.io/install.html#installing-with-pip

 

Typo in “Patching the Framework”

On page 96 there’s a typo and the wrong file name is stated. This was the very last thing I did on the book before shipping it, as the issue that needed patching appeared at the last moment.

The file to be changed is “benchmarks.py”, not “loader.py”.

Here is the complete, corrected version.

Zipline, like most software in Python world is not only free, but open source. You can see and edit all the code as you like, and one big advantage of this is that we can make changes and even fix bugs when needed. You may wonder why you would ever want to go and edit someone else’s code, but it can be very useful to have this possibility.

While writing this book, an interesting example of such a situation came up. One day, all the backtests suddenly stopped working, with an error message about failing to load benchmark data from some web address. As it turns out, there is a web call executed during backtests, where Zipline tries to pull data from an online source. This source, much like what happened to many other free data sources, suddenly stopped working without warning.

As of writing, this issue has not yet been addressed. But no worries, we can do it ourselves easily. We don’t actually need this benchmark data for anything that we will do in this book, and the easiest fix is to simply remove the web call and just return empty data.

My version of Zipline is 1.3, and if you’re working with the same version, you likely need to do this fix manually. It won’t take long.

First locate a file called benchmarks.py, which should be in your Zipline installation folder. You can use the file system search tools to locate it. If you are on Windows, the path will probably be similar to this.

C:\ProgramData\Anaconda3\envs\zip35\Lib\site-packages\zipline\data

It’s always a good idea to keep a backup of the original code, so rename this file to keep a copy. Then make a new file, with the name benchmarks.py, and put the following code in it before saving.

This code will simply return a data series from 1930 to 2030 with all zero values, which will effectively sidestep this web call issue, with no negative impact.

Addition to Patch

You may also need to bypass the cache in loader.py, in the same folder:

https://github.com/quantopian/zipline/issues/2480#issuecomment-504387554

 

Zipline Installation: failed with current_repodata.json

Error:

Collecting package metadata (current_repodata.json): done
Solving environment: failed with current_repodata.json, will retry with next repodata source.

Solution:

1) Update Conda to the latest version:

Trouble Ingesting Quandl

Error:

“ValueError: Boolean array expected for the condition, not float64”

Solution:

Downgrading Pandas to 0.18.1

 

Jupyter version 5.0.0 not starting

Solution:

Downgrading “tornado” from 5.1.1 to 4.5.3

 

Trouble ingesting Quandl on Mac

Try

 

Typo Chapter 23 – Bundle Name

In chapter 23, you’re advised to ingest the random stocks bundle using

But that’s not the name of the bundle that we just created a few lines above. The name in used in the bundle was random_stock_data, so we need to use that name when ingesting.

Similar typo in regards to the futures bundle. The name of the file in the sample code is random_futures_data:

And

Note that you can call your bundles whatever you like, but the file names need to match the references, and the registered bundle name needs to match the ingest call.