Back

New Backtesting Cloud

In a few minutes we're going to deploy a new cloud to production. This is a scalable cloud with predictable speeds which should give you more reliable backtesting. You shouldn't experience any downtime but feel free to let us know of any issues you find.

Super perk of the new cloud will be batch optimization with hundreds of VM's is now much much easier. We'll be look at building this in 2018 =)

Update Backtest






The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by QuantConnect. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. QuantConnect makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances. All investments involve risk, including loss of principal. You should consult with an investment professional before making any investment decisions.



The cloud has been successfully deployed with 0-downtime. Nice work Andrew Hart!

Image result for standing ovation gif

1

The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by QuantConnect. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. QuantConnect makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances. All investments involve risk, including loss of principal. You should consult with an investment professional before making any investment decisions.


Well done folks. It seems to work like a breeze. Mega Kudos =-)

1

I've been having problems saving algos and loading live paper trading algos, everything seems slow :( even tried different browsers.

2

This should be resolved now Noel
1

The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by QuantConnect. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. QuantConnect makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances. All investments involve risk, including loss of principal. You should consult with an investment professional before making any investment decisions.


Everything(especially saving) is working for me now, thank you!
Cloud scalability is a really exciting upgrade for the community, thanks QC team!

0

Hi Jared, is that just me or is the backtesting platform slow again. Each of my simple backtests are taking ages to run . that is when they run at all (one equity, 3 days of data). What's creating this?

1

Yes, backtesting has been slow for me for 2 days now. It takes hours for backtests that used to be completed in 5 min, unfortunately.

1

Sdoofsam jaafari - there were teething problems for the scaling to keep up with demand the first night but now running as normal. 

0

The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by QuantConnect. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. QuantConnect makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances. All investments involve risk, including loss of principal. You should consult with an investment professional before making any investment decisions.


Seems a lot faster now than prior!

Thank you for looking into the speed issues!

I know this is a separate topic that I'm about to mention - is there a feature request/wishlist thread? Speed and debugging has been my biggest hurdle with quantconnect. Seems the speed is getting sorted out and would love to have better debugging mechanism. It's very time consuming finding what part of an algorithm is throwing the errors (or why).

Thanks for all that you do!

1

Hi Charles,

 

I second your motion.  BTW, here's what I'm doing for as a workaround. There may be a better way, so I'm all ears as to better methodologies.

rather than being clueless as to where my errors occurred or trying to decipher extremely crpytic error codes produced by the log files, I do this

at top  of algo use a constant

VERBOSEMODE = True

Then inside each procedure or function I use try except statements, with "reached line number" and optional varriable logging code. Here's the pseudcode sample:

def someFunction(self,somevar):

    try:

        someVar = 1

        if VERBOSEMODE: self.Log("got past line 5 of someFunction with somevar = {}". format(someVar)  )  

        someOtherBrokenVar = 42.000 / 0 

    if VERBOSEMODE: self.Log("got past line 7 of someOtherBrokenFunction with somevar = {}". format(someOtherBrokenVar)  )  

     

   except:       
          self.Log('An unknown error occurred running someFunction  ' +  str(sys.exc_info()[0]) )                 
 

So this code would pass first level variable definition and bomb on second one (division by zero). You would have a clue up to where your code ran fine and an idea about type of error causing problems

If you later get tired of seeing all the log files, simply erase those lines or turn VERBOSEMODE to False.

Hope that helps someone. Now if some bright python guru could tell me how to automatically generate the linenumber (so I don't have to handcode them) as well as function name, it would be even simpler.

0

Hi

I have been having lots of issues with the new backtesting system. Either the backtest will take ages to launch or it will launch as normal but the backtest will crash. Additionally, the usual backtest tearsheet is not being generated. Are you aware of similar issues?

Thanks

Jonny

1

Update Backtest





0

The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by QuantConnect. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. QuantConnect makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances. All investments involve risk, including loss of principal. You should consult with an investment professional before making any investment decisions.


Loading...

This discussion is closed