ASX Stock Pairs Trade Journal

[SilverRanger]

New trades today:
Short BXB & long CEU
Short MND & long TCL
Long NWS & short NWSLV

Getting desperate for my yearly 100%,

 

[skc]

New trades today:
Short BXB & long CEU
Short MND & long TCL
Long NWS & short NWSLV

Getting desperate for my yearly 100%,

Definitely funny pairs you have there...

TCL and CEU both went ex-div today and may have triggered the signal. But having said that Aussie stocks tend to catch up the dividend drop off quite quickly so you never know. BXB has been strong in the last 2 months or so and I don't really understand why - I had a short BXB / long TOL trade some months ago and got cut badly both ways.

Actually closed and reversed my NWS/NWSLV position today. 1.1 ratio seems to be as low as it goes. Although I always wondered why NWSLV doesn't just trade at around the same price as NWS - considering that they receive the same dividend.

Hope you make the 100% mark.

 

[SilverRanger]

New trade for the month:
Long NWS and short NWSLV, this pair is hitting a new low of 1.08 and a new low in spread, but the correlation is picking up. All triggers met, so I'm hoping to close this trade before the end of the week

Didn't quite get my 100% annual profit, but luckily it didn't turn south either - managed to close my losing trades MCR/MRE and MGR/GPT at or close to breakeven

 

[skc]

New trade for the month:
Long NWS and short NWSLV, this pair is hitting a new low of 1.08 and a new low in spread, but the correlation is picking up. All triggers met, so I'm hoping to close this trade before the end of the week

Didn't quite get my 100% annual profit, but luckily it didn't turn south either - managed to close my losing trades MCR/MRE and MGR/GPT at or close to breakeven

I have traded NWS / NWSLV many times but I don't understand what should be their 'correct' relationship. They enjoy the same dividend amounts. NWSLV has preference in dividend and in liquidation, but it has no voting rights and poorer liquidity.

So what's the right discount for NWSLV / NWS? I would have thought the current ~8% is probably right. Obviously the market will fluctuate depending on the sentiment and short term supply and demand.

But because of this view, back in the days where NWS had a 15% premium to NWSLV, I would only long NWSLV and short NWS.

Profit season in Feb so tread carefully... I was too careful however and missed the MYR short last week

 

[SilverRanger]

I have traded NWS / NWSLV many times but I don't understand what should be their 'correct' relationship. They enjoy the same dividend amounts. NWSLV has preference in dividend and in liquidation, but it has no voting rights and poorer liquidity.

So what's the right discount for NWSLV / NWS? I would have thought the current ~8% is probably right. Obviously the market will fluctuate depending on the sentiment and short term supply and demand.

But because of this view, back in the days where NWS had a 15% premium to NWSLV, I would only long NWSLV and short NWS.

Profit season in Feb so tread carefully... I was too careful however and missed the MYR short last week

At the start of 2009 the "premium" was about 5.6%, then it gradually went up and peaked around 20% in 2010 and now going back down again, there's similar trend in the spread as well. But for all that matters, there's a market price for it and we are milking on its "mean reverting" properties

For the notorious confession and reporting period, I think I will just stay out completely (with the exception of this pair of course)

No need to for not hitting the jackpot, it could've been a nightmare as well

 

[skc]

At the start of 2009 the "premium" was about 5.6%, then it gradually went up and peaked around 20% in 2010 and now going back down again, there's similar trend in the spread as well. But for all that matters, there's a market price for it and we are milking on its "mean reverting" properties

For the notorious confession and reporting period, I think I will just stay out completely (with the exception of this pair of course)

Definitely tradable. But would be nice to know what is the right side of the trade.

 

[fingl]

skc, do u use a discretionary approach for taking signals? Whats your opinion of taking every signals for a pair to realize the backtest statistics compared to picking the best of what u think is best among your signals?

 

[skc]

skc, do u use a discretionary approach for taking signals? Whats your opinion of taking every signals for a pair to realize the backtest statistics compared to picking the best of what u think is best among your signals?

It has to be discretionary for me, because you always need to take into account company specific news and events... e.g. rumours on takeover, capital raising, writedowns, upcoming dividends, ex-dividends, quarterly reporting etc etc.

The backtest statistics will contain false information as it cannot discriminate price movements due caused by the above.

 

[fingl]

It has to be discretionary for me, because you always need to take into account company specific news and events... e.g. rumours on takeover, capital raising, writedowns, upcoming dividends, ex-dividends, quarterly reporting etc etc.

The backtest statistics will contain false information as it cannot discriminate price movements due caused by the above.

Other than news, events and other information, do you use chart patterns and indicators to decide which signal has more potential and which has less potential for profit?

And are not all these news and events accounted on the trading backesting statistics? Backtesting does not really care about all these news and events and take all valid signals using the specific rules.

 

[skc]

Other than news, events and other information, do you use chart patterns and indicators to decide which signal has more potential and which has less potential for profit?

And are not all these news and events accounted on the trading backesting statistics? Backtesting does not really care about all these news and events and take all valid signals using the specific rules.

Chart patterns yes. Support / resistance yes. Indicators No.

The backtest doesn't care about a lot of things and you will have false profit/loss.

e.g. Backtest entry signal was to long NAB/short CBA the day before CBA went ex-div. The following day the exit signal is generated because CBA fell 2.5% on going ex-dividend. The backtest says a profitable trade. You've made no money...

The backtest also doesn't show what you've risked to take the reward.

E.g. Share A spiked up on a takeover bid, generating all sorts of signals to short. You can short it for a possible few % profit but risk a bidding war. It's up to you if you want to believe the backtest results as gospel.

 

[edman79]

Other than news, events and other information, do you use chart patterns and indicators to decide which signal has more potential and which has less potential for profit?

And are not all these news and events accounted on the trading backesting statistics? Backtesting does not really care about all these news and events and take all valid signals using the specific rules.

Hi fingl,
Things I have found that cause the backtester to differ from real results :
1. Dividends
2. Consolidations.
3. restructuring of the company - ie westfield spliting. (the backtester thinks that the shares had dropped from about $12 to $9 but in reality you would have been issued with WRT shares.
4. Errors in the historical data - I have found when there is an error in the yahoo data feed for 1 day it always results in a trade opening that day, closing the next day (as the data returns to the correct amount) for a hefty profit...

 

[fingl]

Should the correlation be calculated using raw prices or percentage change? The plot of the two method is quite different using same bars calculation. Which method is more effective for generating successful trades?

 

[skc]

What a crazy day on the market. 3 periods of 40pt swings that we really haven't seen for a while. Volatility is truely back and it's a good time to pairs trade.

My account went from down 1.5% at the open and by 11am it was up 1%... amazing day.

Should the correlation be calculated using raw prices or percentage change? The plot of the two method is quite different using same bars calculation. Which method is more effective for generating successful trades?

Not sure what you mean. Can you elaborate on your question?

 

[fingl]

What a crazy day on the market. 3 periods of 40pt swings that we really haven't seen for a while. Volatility is truely back and it's a good time to pairs trade.

My account went from down 1.5% at the open and by 11am it was up 1%... amazing day.



Not sure what you mean. Can you elaborate on your question?

Raw price represents the closing price of two securities and percentage change represents the percentage change of closing price for specific number of bars.

The main different is for the second calculate the percentage change for 20 bars for individual security before calculating the correlation.

skc, did you write on what filters you use for the trading? What value do you require the correlation to be above for trading? Correlation does not seems to relate alot to the pnl of individual trade. There are many winning trades with low correlation and losing trades with high correlation. Whats your opinion and why do you carry on using correlation for selecting trades?

 

[fingl]

skc do you focus your pairs trading on Australia market? Any preference for stock price?

 

[skc]

Raw price represents the closing price of two securities and percentage change represents the percentage change of closing price for specific number of bars.

The main different is for the second calculate the percentage change for 20 bars for individual security before calculating the correlation.

I think using correlation of price makes more sense than correlation of the percentage changes.

You can check using an example yourself. Say stock A goes up 1% day one and -0.5% on the next day, then up 1% again on day three (and so on for 20 days). And stock B does exactly the same except it starts with -0.5% on day one and +1% on day (and so on for 20 days).

You will have to say that the price is correlated over 20 days, but the percent change are certainly not.

skc, did you write on what filters you use for the trading? What value do you require the correlation to be above for trading? Correlation does not seems to relate alot to the pnl of individual trade. There are many winning trades with low correlation and losing trades with high correlation. Whats your opinion and why do you carry on using correlation for selecting trades?

I don't have strict criteria for correlation. But I have reasonably tight criteria on fundamental correlation (i.e. same industry, same macro factors, same metal etc). Price correlation is often a proxy for fundamental correlation.

At the end of the day the most important on is your ratio chart - hopefully one that is range bound with a moving average through somewhere near the middle.

skc do you focus your pairs trading on Australia market? Any preference for stock price?

Yes only Australian market. With stock price the bigger the better just from an ease of entry/exit and crossing the spread point of view.

You will always under perform the system if you trade too many 55c shares. The system say exit at 55c and you are there at the back of the queue with 2m shares in front of you... happened to me today with GMG and missed my exit

 

[Pairs Trader]

Hey guys, glad to see this pairs journal is still alive and well, just a quick update from me;

For a few years we have wanted to get co-integration into Pairtrade Finder, however after many futile attempts we put it into the "too-hard basket" for those of you that have seen any co-integration code you will understand what we mean, however it has been our biggest feature request for the last year and again we faced up to this tedious task, luckily we've found a capable financial programmer who has experience with co-integration and who actually developed his own co-integration software, we have paid him handsomely to integrate co-integration into Pairtrade Finder and it should be available to all members within the next few weeks.

For those of you who don't know what co-integration is and why it's so important to pair trading here is a quick analogy;

A man and a women are walking down the street together at the same pace, the walking relationship between the man and women would show high correlation, now on the other side of the street a man and a women are walking down the street, however the women stops to look in a shop window and the man keeps on walking, soon after the women runs to catch up to the man, this is co-integration, whereas in this case correlation would have broken down. So to sum up, co-integration measures the tendency for a relationship to revert back to it's mean average. I'm sure you can understand the importance of this measure when selecting appropriate pairs to trade, we want pairs that have a high tendency to converge back to a normal relationship after diverging, it's the ultimate measure of mean reversion in any 2 variable relationship.

This is why every statistical arbitrage(pair trading) hedge fund uses co-integration to scan for profitable pairs to trade.

The co-integration value will show between 0-100 in Pairtrade Finder, similar to the current correlation value shown, and our financial programmer has confirmed it's increase in profitability, in his own words "high co-integration + high back test profits = superb pair to trade"

We highly anticipate the added feature of co-integration into PTF and are quite confident it will increase the profitability for all pair traders, including those that pursue this art full time.

Watch this space...

 

[skc]

We highly anticipate the added feature of co-integration into PTF and are quite confident it will increase the profitability for all pair traders, including those that pursue this art full time.

Watch this space...

That will be awesome... I am (very) pleasantly surprised that you still have time to improve the PTF when you seem so busy doing many other things.

 

[Pairs Trader]

Thanks SKC! I too can't wait to have it in my own trading, plus to improve our pair trading signals service too, which is 2yrs+ running profitable, and yah im very busy but ive learnt to multitask, outsource and automate a lot more these days, the need came out of necessity, there is still a few things more I want in PTF and if you or anyone else has any suggestions I'm all ears...

 

[skc]

Thanks SKC! I too can't wait to have it in my own trading, plus to improve our pair trading signals service too, which is 2yrs+ running profitable, and yah im very busy but ive learnt to multitask, outsource and automate a lot more these days, the need came out of necessity, there is still a few things more I want in PTF and if you or anyone else has any suggestions I'm all ears...

Real time prices for ASX shares!!! I will have that over anything else.