Pair Trading > Good & Bad Pairs

Entry / Exit rules

<< < (2/2)

admin:

--- Quote ---so if the ratio is above/below the second deviation (+2 or -2) you will execute the trade, right?
--- End quote ---

right!


--- Quote ---Do you think 15 days are enough? maybe the correlation could be broken in these few days, no?
--- End quote ---

It depends. For some pairs 15 days is fine, for others it may be not. Personally I prefer pairs which give positive results for all reliable settings. You may want to try the Study feature which executes backtests for many period settings at the same time.

damiano:
Perfect! I did this backtest: https://www.pairtradinglab.com/backtests/UfGCZ-moDaRIDuDp

I would like to know how do you calculate the "Thresholds + Delta", because I saw you use it to do the entries.

As you can see there is only one entry in the 2008. (please do not pay attention to the pair, it is only a test)

Looking at the chart I noticed when the black line break the -2 deviation the entry was made. Ok, if it is true, I did not understand what you said before.

I read that you do the entry applying the standard deviation to the ratio. In the Thresholds + Delta what is the black line? it does not seem the ratio. How do you calculate it?

Thank you for the clarification.

admin:
Yes, I admit it is a bit confusing in this chart (it is old and will be updated in future):

Delta = spread = (R-MA)
Thresholds (green "band") = Entry and Exit thresholds multiplied by StdDev

From the mathematical point of view it is equivalent here to work with Z-Score and constant threshold or with Delta (spread) and thresholds multiplied by StdDev...it is just a different graphical representation of the problem.

So yes, the trade was opened because the spread crossed the -2 deviation point. That's mathematically the same, as if the Z-Score crossed the -2 line.

The chart for ratio model backtests will be updated in future also to show the Z-Score too (same way as the chart for Residual model, which already shows the Z-Score graph).

Navigation

[0] Message Index

[*] Previous page