Share New Research 
Hi Everyone,
In this strategy, we model a Pairs Trading Process as an Ornstein-Uhlenbeck Process in order to derive the Optimal Entry and Liquidation values. I had the chance to ask the co-author of this strategy, Professor Leung, the Director of the CFRM department at the University of Washington (which is where I am attending right now), for help on this strategy. I will attach a backtest, and the full writeup can be found here. Furthermore, this one comes with a video walkthrough! The URL for the video can be found on the page linked earlier or here.
Best,
Shile Wen
Aaron Janeiro Stone
Great work! I think this is my favourite addition to the library so far!
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.
Jovad Uribe
Hey Shile, this is awesome!
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.
Stan M
Nice!
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.
Quant Stratege
Interesting study, how is it possible to extend OU optimization to a meanreverting portfolio ? Maybe it could be usefull if you can upgrade library mlfinlab that have this feature ?
https://mlfinlab.readthedocs.io/en/latest/optimal_mean_reversion/ou_model.html
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.
Shile Wen
Hi Quant Stratege,
It is definitely possible to extend this to a portfolio, so we would need to have the portfolio values be a linear combination of more than just two time-series of price data. However, please note optimization runtime tends to be O(k^n), so it might take much longer when more equities are involved. Furthermore, I have seen the mlfinlab post, and it is possible to use it to re-implement this strategy (QC would need to update our mlfinlab version first however) using this method.
Best,
Shile Wen
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.
Spacetime
AlMoJo ,
Also, if you would like to explore more CBOE data sets, then you can use the Nasdaq Data Link provided below.
Correction in my last post: meant to say CBOE and not Quandl (Quandl was acquired by Nasdaq)
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.
Spacetime
AlMoJo , Just posting our QC CBOE Datasets link below for your reference if you need it.
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.
Vladimir
Spacetime,
There are only 6 trades for SHV in your version of the code.
Can you try changing in the code the VIX and VXV data retrieval in a way like in the attached sample.
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.
Spacetime
Hi Vladimir,
Hmm… not entirely sure why there were only 6 trades executed.
The above backtest which I have attached was taking a “bit” to complete for me and I was engaged with other work, so I ran the above model (shaerd by pangyuteng) with a start date of self.SetStartDate(2020, 11, 1) just to speed up things.
Hmm… I have extended the date range to start from self.SetStartDate(2019, 1, 1) and it does not have too many trades either. (backtest attached)
If you have noticed from the above example shared by pangyuteng , then it does not trade that often either. [ self.SetStartDate(2019, 1, 1) ]
But, the fixes which I have shared above does work as the error which was being received by AlMoJo is no longer occurring.
Hmm… maybe someone else might have a better idea on this, but I will try to look into this further. Hmm… maybe the LSTM configuration needs to be looked into or the volatility data resolution and other resolutions needs to be changed to minute time slice if possible.
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.
Vladimir
Spacetime,
I backtested cloned pangyuteng algorithm with only changes in my_custom_data
url_vix = "http://cache.quantconnect.com/alternative/cboe/vix.csv"
url_vxv = "http://cache.quantconnect.com/alternative/cboe/vix3m.csv"
It generated 13 SPY-SHY trades and metrics as your last one, but the performance is not comparable to pangyuteng's results.
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.
Spacetime
Vladimir,
Hmm… If you want to compare it with pangyuteng's results, then you have to provide an end date of self.SetEndDate(2020, 1, 1) because pangyuteng's model ends around that time. Please note, pangyuteng's post is time stamped somewhere in December 2019.
Try to run it with end date parameter and check to see if performance is comparable.
I have attached a backtest run starting from self.SetStartDate(2019, 1, 1) and ending at self.SetEndDate(2020, 1, 1) and it only traded once.
Hmm…
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.
Vladimir
Spacetime,
I definitely like the metrics from your last backtest, but my attempts to reproduce them for some reason were unsuccessful.
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.
AlMoJo
Hello everybody,
Pangyuteng this is an amazing result. Jack Simonson I also think technical indicators would help a lot.
I tested Pangyuteng version on the 2022 crash and it is not so good at dodging it. I taught maybe going hourly timeframe instead of daily could help even if I know that trading more frequently is not really good.
Anyone has an idea on how to change the timeframe? I tried to pass all the .Daily to .Hourly but obviously it wasn't that simple.
Kind Regards
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.
Varad Kabade
Hi AlMoJo,
To change the timeframe of the entire algorithm we recommend adding the following to the Initialize method:
Note that the algorithm uses higher frequency bars to create lower-frequency bars. Therefore to use indicators or consolidators at the hourly resolution we need to have the universe at the minute, seconds, or hourly resolution.
Best,
Varad Kabade
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.
Will Berger INVESTOR
Hey Guys and Sherry,
Thanks for this incredible thread. So many interesting ideas!
Will
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.
To unlock posting to the community forums please complete at least 30% of Boot Camp.
You can continue your Boot Camp training progress from the terminal. We hope to see you in the community soon!