Unable to replicate backtest performance | Chapter 8 02_backtesting_with_zipline

[benhsampson]

Running the notebook myself, the backtest performs worse. Note that I've imported the notebook and not made any modifications.

My run

From repo

Environment Python 3.8 Here's the important dependencies in requirements.txt file after running pip freeze > requirements.txt

TA-Lib==0.4.28
zipline-reloaded==2.1.1
pyfolio-reloaded==0.9.4

• OS: Windows
• Version: 11

[BlockchainPunks]

if possible - would you mind exporting your .yml files for your ml4t and backtester environments then attach them here.

you can do that using conda command.

The install instructions are messed - i'm looking for working windows exact package list which steps below will give here are steps: https://conda.io/projects/conda/en/latest/user-guide/tasks/manage-environments.html#sharing-an-environment

this will help allot of the windows people - so would really appreciate your help

[stefan-jansen]

Can be differences in data (which may change depending on when you download), differences in the trained model, etc.

I"m sorry but the goal is to show you how you can run this, not to claim specific performance so I can't help much further, thanks for your understanding.