Pairs Trading UK & US Stocks Journal

[the blades]

LOL ! It isn't that there are insufficient opportunities among the shares - rather that I had already been thinking of opportunities in certain indices and commodities and was interested to know whether the software evaluates these.

I wasn't being "funny", really. I just doubt that there'll be many opportunities here. Especially when you consider the number of traders following the main movers, quickly removing any price discrepencies. As I say, it's worth a look, so good luck

UTB

 

[Lado]

S77:

When backtesting pairs in PTF to find suitable pairs to follow, is there an unwritten rule regarding the minimum correlation in the past 365 days that is acceptable in order to reduce the risk of a pair diverging? For example, if a pair has had a correlation greater than 90% for the past 3 months but had a correlation of 10% 9 months ago, is that an acceptable pair?

Also, is there a preferable range for the ratio? I know that a good pair is one for which the ratio has been trading sideways. Many pairs have ratios that are trending up or down. Do you look for a particular range (high ratio/low ratio)?

Lado

 

[stocktrader77]

S77:

When backtesting pairs in PTF to find suitable pairs to follow, is there an unwritten rule regarding the minimum correlation in the past 365 days that is acceptable in order to reduce the risk of a pair diverging? For example, if a pair has had a correlation greater than 90% for the past 3 months but had a correlation of 10% 9 months ago, is that an acceptable pair?

Also, is there a preferable range for the ratio? I know that a good pair is one for which the ratio has been trading sideways. Many pairs have ratios that are trending up or down. Do you look for a particular range (high ratio/low ratio)?

Lado

Generally you want to stick to pairs that have an 1 year avg correlation of 80% or higher, if a pair has a had poor daily correlation readings in the past that is ok, it may have been a news event that affected prices for awhile. Yes for the ratio chart, I like to see regular and consistent oscillations around the mean, doesn't have to be a particular % range, try and pick ratio charts that look like a hearbeat monitor, don't trade strongly trending ratio charts.

 

[stocktrader77]

Opened new pair trade in the US small cap biotechnology sector

Long Entry CBM - 2.63

Short Entry OMPI - 6.28

 

[pedro01]

Opened new pair trade in the US small cap biotechnology sector

Long Entry CBM - 2.63

Short Entry OMPI - 6.28

That's an interesting pair - looks like only 40% correlation prior to November...

How do you protect yourself against a situation where the sector takes a nosedive and all players in the sector appear to be more correlated ?

Also - the big bear market we are currently in - most instruments are going down but that doesn't mean they are correlated per se - just that everything is going down.

I'd have thought that a strategy like this would be better in a non-trending market. How do you distinguish between truly correlated instruments and those simply moving with the market trend ?

 

[stocktrader77]

That's an interesting pair - looks like only 40% correlation prior to November...

How do you protect yourself against a situation where the sector takes a nosedive and all players in the sector appear to be more correlated ?

Also - the big bear market we are currently in - most instruments are going down but that doesn't mean they are correlated per se - just that everything is going down.

I'd have thought that a strategy like this would be better in a non-trending market. How do you distinguish between truly correlated instruments and those simply moving with the market trend ?

Yes thats a good point and its not something I fuss about too much, if stocks are moving with the market then thats good because we will achieve market neutral so its not a bad thing should stock be correlated to a sector which is correlated to the general market. Most importantly we rely on fundamentals to prove there is real correlation as there are many correlated pairs from different industries, for instance ive seen a trucking stock with 95% correlation to a clothing stock, I wouldn't trade this pair as different industry-macro factors affect each business, so as long as our pairs are in the same industry we only use the technical correlation reading to confirm its relationship.

 

[stocktrader77]

Sold BPZ - 3.65

Covered PDC - 4.15

 

[stocktrader77]

Opened new trade in US Oil & Gas Drilling Industry

Long Entry PDC - 3.85

Short Entry RDC - 12.22

 

[stocktrader77]

Entered new trade in US Gold Industry, Im finding most of my UK pairs are decreasing in correlation at the moment for those wondering why I haven't taken many UK trades.

Long Entry ANV - 4.00

Short Entry GRS - 7.76

 

[Lado]

S77:

I am surprised with your latest pair. The correlation graph on PTF shows that correlation has peaked and is now declining. Your comments please.

 

[stocktrader77]

Lado, the correlation chart doesn't have to be making all time highs, a little dip is ok, its the most likely the divergence that caused this little dip, whats important is the latest reading is higher than the 1st reading on the chart so there is a long term uptrend, if there was a strong short term trend down I wouldn't take the trade.

 

[Lado]

From what I have read on a similar thread at elitetrader.com, the experienced traders say that correlation has to be trending upward for a pair to be worthy of consideration. For this most recent pair, correlation has peaked and is heading downward. May be others get it but I don't.

Lado

 

[the blades]

From what I have read on a similar thread at elitetrader.com, the experienced traders say that correlation has to be trending upward for a pair to be worthy of consideration. For this most recent pair, correlation has peaked and is heading downward. May be others get it but I don't.

Lado

well for there to be an opportunity in the first place the correlation , by definition, has to have fallen away over a short time frame - surely?

UTB

 

[Lado]

UTB:

Does your statement implicate that you are betting the correlation trend will reverse whereas if you select a pair with rising correlation you are betting that the trend will continue? Maybe I am placing too much emphasis on correlation.

I am trying to learn what makes a good pair but haven't accomplished that yet.

Lado

 

[the blades]

UTB:

Does your statement implicate that you are betting the correlation trend will reverse whereas if you select a pair with rising correlation you are betting that the trend will continue? Maybe I am placing too much emphasis on correlation.

I am trying to learn what makes a good pair but haven't accomplished that yet.

Lado

it depends what timeframe's you're looking at. Over the "shortest" timeframe one of the pair must have risen against its' mean, and the other fallen, for a trade to be signalled. This is negative correlation. So discussions about correlations must be timeframe dependant.

UTB

 

[stocktrader77]

I wouldn't worry too much about correlation, so long as its high and both stocks are in the same industry thats good enough for me, yes increasing correlation is a plus, I know the guys at ET preach it, but its not a strict mandate imo.

Closed one trade

Sold UA - 13.77

Covered BEBE - 4.87

Opened trade in US Oil & Gas Drilling Industry

Long Entry ME - 7.40

Short Entry NTG - 15.04

 

[stocktrader77]

Took trade in US Life Insurance Industry

Long Entry MET - 13.72

Short Entry TMK - 18.08

 

[stocktrader77]

New trade in US REIT Retail Industry

Long Entry SPG - 29.36

Short Entry TCO - 15.60

 

[JGH03]

Presumably PairTradeFinder tells you when the ratio between the pair you are trading has moved sufficiently close together again and it is time to take your profits. Does it also give a signal if the pair keep on diverging and it is time to cut your losses ? And if so, on what basis is that signal given - correlation falling below a particular threshold, perhaps ?

Also, what is your win / loss ratio ? Do you have a rough idea of what that ratio would be if you didn't apply any discretion but followed all the signals blindly ?

 

[stocktrader77]

Presumably PairTradeFinder tells you when the ratio between the pair you are trading has moved sufficiently close together again and it is time to take your profits. Does it also give a signal if the pair keep on diverging and it is time to cut your losses ? And if so, on what basis is that signal given - correlation falling below a particular threshold, perhaps ?

Also, what is your win / loss ratio ? Do you have a rough idea of what that ratio would be if you didn't apply any discretion but followed all the signals blindly ?

Yes it will trigger an exit signal when the pair comes back to its mean, it doesn't use a stop loss, because a rolling mean is used instead of a static mean it will simply exit once the pair comes back, profit or loss.

About 70% of my trades are profitable, not sure what the ratio would be if I didn't apply filters, it may be the same but the trades I take would have a lower risk profile.

Closed a trade

Sold PDC - 3.90

Covered RDC - 11.32

New trade

Long Entry ACE - 31.87

Short Entry CB - 35.24