You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Pair Trading Strategy Journal
- Thread starter jonnysharp
- Start date
There are several principles used to select the pairs, but the one pairtrading finder uses is based on the standard deviation, if you asume complete ramdom samples 95% of all samples will fall in the range of 2 standard deviation from the mean value of the samples, this is considered a normal distribution or bell curve.Quote from waltbx:
How is Spread used as a confirmation signal in pair trading?
Apparently during the end of Dec into January, spreads become wider. How does that affect pair trading?
Walt B
The spread by itself does not give you a confirmation signal, is the way this spread has changed during the lookback period (14 days) what determines the standard deviation and when the stock ratio is in this boundary there is a great possibility that the ratio goes back to the mean value, but in reality this ratio is not a random value.
I remember something I read awhile ago about pair trading:
A man take his dog for a walk, they are going in the same direction but the dog during the walk takes different routes but eventually the are going to meet again somewhere, in the house's door or in the backyard.
The crazier the dog the higher the satandard deviation and you are going to find higher spread between the man and the dog during the walk.
I have some books and papers about the math behind pair trading but unless you are a mathematician once it gets into the heavy math you get lost.
Quote from waltbx:
My figures show a profit of 0.9%
RSI dropped to 29 yesterday, reversed and is rising today. May have a little more profit.
Walt B
I am notorious for taking profits early, so that certainly may be the case. But bulls,bears, pigs and all that jazz...
Depending on how you calculate, that prolly isn't far off. I show +1.05% if you take total capital deployed, but on market neutral where funds hit the account to offset the long, it's a decent gain.
As an aside, and fwiw, you may want to rethink the standard 14 period RSI. I have found it to be virtually worthless in short-term trading.
Quote from bentedges:
I am notorious for taking profits early, so that certainly may be the case. But bulls,bears, pigs and all that jazz...
Depending on how you calculate, that prolly isn't far off. I show +1.05% if you take total capital deployed, but on market neutral where funds hit the account to offset the long, it's a decent gain.
As an aside, and fwiw, you may want to rethink the standard 14 period RSI. I have found it to be virtually worthless in short-term trading.
What period RSI are you using, and why?
I used end of day quotes for the entry, may be different than what you actually did.
wb
Quote from magon:
There are several principles used to select the pairs, but the one pairtrading finder uses is based on the standard deviation, if you asume complete ramdom samples 95% of all samples will fall in the range of 2 standard deviation from the mean value of the samples, this is considered a normal distribution or bell curve.
The spread by itself does not give you a confirmation signal, is the way this spread has changed during the lookback period (14 days) what determines the standard deviation and when the stock ratio is in this boundary there is a great possibility that the ratio goes back to the mean value, but in reality this ratio is not a random value.
I remember something I read awhile ago about pair trading:
A man take his dog for a walk, they are going in the same direction but the dog during the walk takes different routes but eventually the are going to meet again somewhere, in the house's door or in the backyard.
The crazier the dog the higher the satandard deviation and you are going to find higher spread between the man and the dog during the walk.
I have some books and papers about the math behind pair trading but unless you are a mathematician once it gets into the heavy math you get lost.
Ok, I sort of understand some of what you are saying. But bottom line, how do I use the Spread Graph? Good, bad when trending, rising, falling, gapping? Any suggestions? Seasonally, when spreads increase, is that a positive or negative? I still don't understand how spread is related to convergence/divergence in the pair prices. Increasing spread results in increasing divergence? So I would want a spread that is decreasing with a history of ???
wb
Quote from waltbx:
What period RSI are you using, and why?
I used end of day quotes for the entry, may be different than what you actually did.
wb [/QUOTE
I've found a 2-7 period RSI works much better for short-term trading, but as I've stated previously, I don't put much emphasis on that oscillator. Just my 2 cents.
Quote from bentedges:
Quote from waltbx:
What period RSI are you using, and why?
I used end of day quotes for the entry, may be different than what you actually did.
wb [/QUOTE
I've found a 2-7 period RSI works much better for short-term trading, but as I've stated previously, I don't put much emphasis on that oscillator. Just my 2 cents.
Strange, PTF RSI chart is identical for lookback of 7, 14, and 28 days.
wb