R Backtesters: Quantstrat vs SIT

Question

I am learning R, and want to start using a backtester. I have spent about a day reading all I can about R backtesters, and it seems there are 2 main contenders:

1. quantstrat, which uses the packages blotter and FinancialInstrument, and
2. Systematic Investor Toolbox by Michael Kapler.

For anyone who has experience with either of these, can you tell me if they have done everything you expected.

[Edit] To clarify: I am writing some R code that needs to use a backtester, and from my research quantstrat and SIT are the 2 main contenders.

I am not looking for a religious debate a la Python/R :) but rather whether there is a general consensus as to which is more widely used, which is richer feature-wise, or whether both are worthy contenders.

Answer 1

While I've never used SIT, I have used quantstrat quite a bit and can attest to its strength. It has a solid developer community backing it (7 contributors on Github), is part of the TradeAnalytics project on R-Forge, and while it's still technically in beta, it should provide plenty of functionality. There is admittedly a pretty steep learning curve when you're first learning how to use it, but once you learn how to set things up it will manage portfolios and handle accounting rather gracefully. It has good documentation surrounding it with some solid examples to help get you going (see the QuantStrat TradeR blog). If speed is a priority, quantstrat can also make use of parallel processing functionality, though your mileage may vary depending on operating system.

Taking a look into SIT, it appears that it's mostly a tool built around a specific developer's needs/preferences, and might be slightly less mature at this stage. The documentation seems mostly limited to the SIT blog, and the author looks like the only project contributor on Github. Having not used it in practice, I can't speak directly to its strengths/weaknesses, but I get the subjective impression that quantstrat might currently be the more robust option of the two.

Answer 2

I've never used QuantStrat, but have used SIT for about two years. Michael's blog provides a great way to learn R, understand SIT, and learn about backtesting strategies. I've never found a mistake in his code, and that's how I made a living for 20+ years. It takes some real persistence to understand how to use it in depth, but you can easily set up tests by just copying and modifying one of Michael's own backtest stubs. Michael seems to have tapered off blogging recently, as has David Varadi, whose strategies Michael often tested. Still, I highly recommend it as a full featured tool that supports universe selection, top n selection, numerous weighting schemes, commissions, various rebalancing periods, leverage use, etc., all customizable and extensible.