Why Do Pairs go wrong?

[JAWS]

I like to trade V, MC pair. I used to like to trade CVX, XOM pair. However, one day these pairs all blow up and you end up giving back all your gains.

What methods do pair traders on this website use to detect when a pair is going bad? Any links to papers or ideas are welcome.

 

[lawrence-lugar]

one day these pairs all blow up and you end up giving back all your gains

Pairs, straddles, automation, or combinations thereof...All blowup, or takes a significant hit, one day or another,

Trading requires constant vigilance and intuition and skill and keeness.
If it were that easy to essentially print money...everyone would be doing it fruitfully, effortlessly into perpetuity,

You need to be an ET extraterrestrial to survive and thrive in the market,

 

[JAWS]

Pairs, straddles, automation, or combinations thereof...All blowup, or takes a significant hit, one day or another,

Trading requires constant vigilance and intuition and skill and keeness.

I agree. Although I like to take all the intuition out of my trading. Once I have my model, I let my program keep my levels for me in and out.

The problem is in knowing when to get out and take the loss, or add. And in many cases, when to put the pair aside until it proves to be in a trade-able range again.

Those are the things [exit, add, put aside] I am hoping we can discuss on this thread

 

[newwurldmn]

I like to trade V, MC pair. I used to like to trade CVX, XOM pair. However, one day these pairs all blow up and you end up giving back all your gains.

What methods do pair traders on this website use to detect when a pair is going bad? Any links to papers or ideas are welcome.

Each pair would have a different reason and catalyst.

If you are doing short term like intraday that will require a certain signal. If you are doing longer that will be different.

 

[R1234]

you should avoid holding the pair during earnings announcements and to a lesser extent ex-dividend days.

Are you trend following the pairs or mean reversion?

 

[tommcginnis]

"Pairs" do not go wrong.

Pairs trading is Reversion-to-Mean trading -- it observes a history of consistent behavior and keys off of working an observed inconsistency to trigger an entry.

You can't have a recovery to historic behavior -- a reversion to mean -- without having, first, an *excursion* from historic behavior.

 

[JAWS]

"Pairs" do not go wrong.

Pairs trading is Reversion-to-Mean trading -- it observes a history of consistent behavior and keys off of working an observed inconsistency to trigger an entry.

That's incorrect. You can regime trade trend-follow pairs, or you can regime trade mean-revert pair. They go "wrong" [for you] when you are playing them to be in one regime or the other, but the market goes against your chosen regime.

You can't have a recovery to historic behavior -- a reversion to mean -- without having, first, an *excursion* from historic behavior.

Again this is wrong. However, if you remove the "mean-reversion" part, it is then correct, but since it is just a definition it is therefore tautologous.

"In rhetoric, a tautology(from Greek ταὐτός, "the same" and λόγος, "word/idea") is a logical argument constructed in such a way, generally by repeating the same concept or assertion using different phrasing or terminology, that the proposition as stated is logically irrefutable, while obscuring the lack of evidence or..."

https://en.wikipedia.org/wiki/Tautology_(rhetoric)

 

[JAWS]

you should avoid holding the pair during earnings announcements and to a lesser extent ex-dividend days.

Are you trend following the pairs or mean reversion?

This is a good suggestion. There is no inexpensive way to get historical earnings dates. Do you know of any?

 

[tommcginnis]

"Jaws" -- you confuse trading in correlated assets with pairs trading. You also confuse a precedent condition with the circular tautology. No biggie.

 

[JAWS]

"Jaws" -- you confuse trading in correlated assets with pairs trading.

How am I confusing them?