Ask HN: Quants and traders: Best OS backtesting platform for trading strategies? | Hacker News Reader

Ask HN: Quants and traders: Best OS backtesting platform for trading strategies?

10 pointshodder4 comments

I'm looking for some open source software to dive into for backtesting trading strategies in/out of sample which produces useful tear sheets, nice built in functions, and quality documentation.

I have looked Quantopian/Alphalens/Zipline, but it seems like it has been left to wither and die outside of Quantopian's online platform. The documentation is frankly terrible.

I have briefly looked at Quantconnect/LEAN and it looks more promising but havent dived in to much detail.

I am also looking at the Quantstrat package in R which appears to be quite promising as well.

I'm involved in the futures markets myself but data access is not an issue. I have Bloomberg so can source my data from them.

So, Quants and traders of Hacker News who aren't using in-house corporate solutions, what are you using?

I use Zipline. It serves the purpose. Typically, you need to decide what instrument you want to trade, and look at what backtesting software provides the best support for that instrument. Not all s/w cover all instruments. If you have access to Bloomberg for data. Find s/w that has a plug-in for Bloomberg.

I really like pyFolio tear sheets.

Backtrader is pretty good, but eventually you will find that there really isn't anything out there that will do everything you want it to do.

So imo the best strategy is to roll your own.

Thanks. I’ll look into backtrader. Eventually I’ll be building my own or stapling onto existing solutions.

Well backtests depend on everything including assumptions about your data, your trading reality, corrections for multi hypothesis testing, comparisons to a billion benchmarks, working out why your advantage works when it does, etc. I dare say every team has their own rig which takes into account as many of their trading realities as possible.

To that effect I’m not sure there’s a useful answer to your question which won’t mislead you.

Reply on news.ycombinator.com