Search
× Search
Saturday, February 8, 2025

Archived Discussions

Recent member discussions

The Algorithmic Traders' Association prides itself on providing a forum for the publication and dissemination of its members' white papers, research, reflections, works in progress, and other contributions. Please Note that archive searches and some of our members' publications are reserved for members only, so please log in or sign up to gain the most from our members' contributions.

Cointegration for Pairs Trading - Part 2

photo

 private private,

 Thursday, June 2, 2016

Cointegration for Pairs Trading - Part 2 - now released. http://bit.ly/CointPart2 This article moves on from Part 1 to look in more detail at the underlying price action hidden behind cointegrated pairs and introduces the concept of stationarity - a precursor to being able to perform the cointegration calculations.


Print

14 comments on article "Cointegration for Pairs Trading - Part 2"

photo

 Oleksandr Medviediev, Portfolio Analyst – FIBI

 Friday, June 3, 2016



Part 2 turned too academical. Hope to see more practical implications and profitable ideas in Part III, like this example for instance http://bit.ly/1TZqGg4


photo

 private private,

 Friday, June 3, 2016



Hi Oleksandr. Thanks for the feedback. Part 3 is already underway and it will be practical. Rather than using a package like R to calculate the Augmented Dickey-Fuller p value I will show how this can be done in Excel step-by-step enabling people to fully understand how and why Dickey Fulller works. Always good to get feedback so thanks. Martyn


photo

 private private,

 Friday, June 3, 2016



If only price series was stationary.


photo

 Gaurav Singh, AVP at Tradeweb

 Friday, June 3, 2016



Andrew , the individual price series would always be highly unlikely to be stationary, however spreads between them might turn out to be stationary . Check out my blog for details.


photo

 Abel Parker-Geisman, Undergraduate Student Bentley University: Eco-Fi, Minor:Math

 Friday, June 3, 2016



so for groups with low P-values you would short when above and buy when below? why do you need this done on pairs?


photo

 Daniel Uhlemann, CFA, Aviva Investors

 Saturday, June 4, 2016



There have been many papers on this subject, also going into details of ADF vs Johanson test and VECM. I prefer SAS and VARMAX function. What would be interesting is to see how pair trades work when you are getting regime changes...how can those be anticipated.


photo

 private private,

 Saturday, June 4, 2016



Yes Andrew. We'd all be rich!


photo

 private private,

 Saturday, June 4, 2016



Hi Abel. The p value merely tells you that the pair are cointegrated i.e that the spread between them is mean reverting in nature. It doesn't tell you which way to trade each instrument in the pair. For that you need to look for a divergence in the spread and the short one of the instruments and link the other in the expectation that the spread will revert back to the mean. There's a lot more to it than that but that's the general gist.


photo

 private private,

 Saturday, June 4, 2016



Hi Daniel. Thanks for taking the time to comment. Regime changes. yes that's the big question. In the end, as with the majority of trading systems it comes down to a probabilities game. hopefully most of the time the spread mean reverts but when it doesn't for some underlying fundamental reason we have to have our stop loss strategy ready to kick in.


photo

 Mithil Pradhan, CMT, Futures Trader & Founder at Clover Edge Academy of Technical Analysis

 Monday, June 6, 2016



Too much complexity spoils the cake. Not all pairs are mean reverting. In fact spread ratios r more secular than most people think. Mean reverting strategies need only one disaster to invalidate the theory.


photo

 private private,

 Monday, June 6, 2016



Hi Mithil - That's where a good stop loss strategy comes in surely?


photo

 Oleksandr Medviediev, Portfolio Analyst – FIBI

 Tuesday, June 7, 2016



Agree with Mithil Pradhan, CMT. In Pairs Trading your harvest is - many small aggregate profits 3-5-15 pips range each. "Let the profit grow" approach doesn't work here. So if hit "a good stop loss" - might take forever to recover.

Bottom line - aim your system to be so good//robust to work without SL.

Also to note in Part-3: correct lot-sizing as critical stand along issue in any statarb. Formula for lot-size must include instruments pip-values and adjusted to recent volatility (in addition to Risk Management).


photo

 Ajay Kakarania, Project Manager at Avaloq Sourcing Asia Pacific(Singapore) Pte. Limited

 Wednesday, June 8, 2016



I guess it also about forecasting and with that hope ...:Did u spend anytime comparing Johansen framework against Engle granjer?... How about beta modelling , did you use something propriety ?? Just curious :-)


photo

 private private,

 Thursday, June 9, 2016



Hi Ajay - I have only used Dickey-Fuller to date and that will be the subject of Part 3, but I'm the kind of person that likes to learn, so at some point I'm sure I will...

Please login or register to post comments.

TRADING FUTURES AND OPTIONS INVOLVES SUBSTANTIAL RISK OF LOSS AND IS NOT SUITABLE FOR ALL INVESTORS
Terms Of UsePrivacy StatementCopyright 2018 Algorithmic Traders Association