We are not affiliated to QuantGo (www.quantgo.com) in any way. RobustTechHouse is a leading web & mobile app development company in Singapore focusing on ECommerce, Mobile-Commerce and Financial Technology (FinTech).
Renting Historical Data For Backtest
We first read about QuantGo on Ernie Chan’s blog (http://epchan.blogspot.sg/2014/11/rent-dont-buy-data-our-experience-with.html). Go Ernie!
Instead of buying and paying a high price for historical data upfront, why don’t you rent the data at a monthly price? Now that makes a lot of sense if we have an idea we want to try and not sure if it would really fly, so just rent it for a while to test it out first.
Test Drive of QuantGo
We decided to give QuantGo a try.
Their team consists of a few very hardworking, knowledgeable, credible and helpful guys. They know the initial setup might be confusing to quant guys who may not have experience using AWS or Virtual Machine setup, so they refuse to just throw you in the deep end of the pool of documentations and let you survive on your own. We suggested that we want to play with the platform ourselves to figure out, but they decided to patiently walk us through demos, talk through how to use the system so we can hit the ground running and be productive with strategy back-testing immediately.
Their ever helpful support team responds very quickly to your support queries as well. They have deep knowledge from the quant industry of trading strategy research and know their infrastructure setup very well, so they can actually provide you very good advice on how to use QuantGo to perform your back-tests.
Amazing Arsenal Of Data
Australia, Austria, Bahrain, Bermuda, Bosnia and Herzegovina, Brazil, India, Canada, Cayman, Chile, China, Colombia, Cyprus, Czech Republic, Denmark, Ecuador, Egypt, Estonia, Finland, France, Germany, Greece, Hong Kong, Hungary, Iceland, Indonesia, Ireland, Israel, Italy, Jamaica, Japan, Jordan, Kenya, Kuwait, Latvia, Lithuania, Luxembourg, Malaysia, Malta, Mexico, Morocco, Netherlands, New Zealand, Norway, Oman, Pakistan, Panama, Peru, Philippines, Poland, Portugal, Qatar, Romania, Russia, Saudi Arabia, Singapore, Slovak Republic, South Africa, South Korea, Spain, Sri Lanka, Sweden, Switzerland, Taiwan, Thailand, Trinidad and Tobago, Tunisia, Turkey, Ukraine, UAE, UK, US, Venezuela, Vietnam
Wow! That is an impressive list of countries to have historical tick data in. Really gives us an itch to try back-testing on all of them now.
Actual User Experience With Back Testing on QuantGo
The QuantGo team has worked hard to provide the toolsets and documentations. The documentation and tools aren’t pretty but they definitely have all the functionality and information you need. There were some initial glitches during account setup but QuantGo team solved it very quickly for us.
Once you remote into the VM to do your work, it is like your own desktop as usual. You can setup any software or back-testing platforms including our usual Dropbox, Anacondo for Python, R and R Studio etc. You probably can install the back-testing software you need to use yourself, be it AlgoTrader (http://www.algotrader.ch), MarketCetera (www.marketcetera.com) or your own customized Zipline.IO library (http://zipline.io) if you prefer.
QuantGo makes it very clear that the historical data is provided only on rental purposes. You can copy out your own back-test results but you should not try to copy out the data using any means. They have their own ways to track whether you might be copying copious amount of rented data off their VMs, so please do not do that.
- Initial charge of USD50 for setup.
- Monthly charges for running the virtual machine on AWS. Their costs would be AWS costs + their overhead. You can see the example at https://quantgo.com/vqlpricing/. For initial testing, we are happy with a medium instance, so we pay about USD50 monthly for the Windows VM.
- Monthly data rental charges. Each exchange / market might cost about USD100-250 monthly to rent. However, there is a minimum of 3 months or 6 months for renting the data. You can try out the demo data they provide but if you want the real access to all their historical data, you can’t just rent for a month. You have to rent for at least a few months.
Like AWS, you can check your QuantGo dashboard to transparently see the costs you are incurring. No surprises with how much you need to pay.
I Really Wish
Rental Minimum: The minimum length of time for rental is shorter. Minimum 6 months could be too long for us and some other users. Sometimes if quant models have been developed for one country and we just want to try it out quickly on a new country, all we need is a month to decide if we want to proceed. If we don’t want to proceed, then we are still locked in for a few months unnecessarily.
History of data: For many of the exchanges, the data mostly starts from 2009 onwards. We wish it included 2008 credit crisis period which is quite a different regime that we need to back-test for strategy robust-ness as well.
VM Location: QuantGo’s current AWS instances are in NY. With a roundtrip latency of 250ms or so from Singapore to NY, remote desktop experience can be quite annoying. Sometimes acceptable, sometimes we have to consider doing most of the coding on our own workstations, copy the code over to QuantGo VM and running and debugging on the VM. It likely won’t be difficult later for them to make AWS instances available in different regions including Asia eg Singapore since AWS has server farms in Singapore.
Conclusion – Would we use it?
For those countries where we have historical data, we don’t need to use QuantGo.
But for those countries where we don’t have historical data and they are not available cheaply, we would definitely consider QuantGo and renting data from them.
In many ways, I do find Quantopian (https://www.quantopian.com/) or QuantConnect (https://www.quantconnect.com/) would be much easier for newcomers. But they are extremely limited in market access (only US equities and maybe some FX). For those serious into this and want to explore markets not supported by Quantopian and QuantConnect using your own toolsets, you are better off doing it yourself if you have the data or using QuantGo if data is very expensive to buy one off and you are happier to rent.
Together with Quantopian, QuantConnect, I believe QuantGo is wonderfully contributing to the democratization of quant trading and making it more accessible to the masses. We look forward to more exciting tools and data access from QuantGo team.
If you like our articles, please follow and like our Facebook page where we regularly share interesting posts and check out our other blog articles where we write about programming, eCommerce, mobile-commerce, FinTech, Machine Learning and other interesting topics.