ASX Stock Pairs Trade Journal

[bochman]

PT, I was just joking in a sense.

And please don't get hit by a bus!

 

[cooper1308]

Would it be possible to run PTF backtester over the 12 month period, with 5 min data?

I guess EOD data is sufficient if you are only trading EOD. However, I believe the more data over the same time period, the better the edge. I watch the markets all day and managed to open/close several trades at favorable levels that are showing as an open loss in PTF..

I'll exaggerate to make a point..... Imagine every trading day the long leg in your pair opened +5% and short opened -5%, but they always closed EOD at 0% & 0%... As I am watching the markets I would have closed the trades 99% of the time at the +5 -5 levels and not held on... Yet the backtesting data would always convey something different.

You may get lucky every now and again because using EOD you keep a trade open that otherwise would have been closed.....and then the next day both move considerably further in your favour.

But overall your system I believe is at a significant disadvantage.

EOD = 1 price level per day
5 min = 360 odd price levels per day

Agree/Disagree/thoughts?

 

[skc]

Would it be possible to run PTF backtester over the 12 month period, with 5 min data?

I guess EOD data is sufficient if you are only trading EOD. However, I believe the more data over the same time period, the better the edge. I watch the markets all day and managed to open/close several trades at favorable levels that are showing as an open loss in PTF..

I'll exaggerate to make a point..... Imagine every trading day the long leg in your pair opened +5% and short opened -5%, but they always closed EOD at 0% & 0%... As I am watching the markets I would have closed the trades 99% of the time at the +5 -5 levels and not held on... Yet the backtesting data would always convey something different.

You may get lucky every now and again because using EOD you keep a trade open that otherwise would have been closed.....and then the next day both move considerably further in your favour.

But overall your system I believe is at a significant disadvantage.

EOD = 1 price level per day
5 min = 360 odd price levels per day

Agree/Disagree/thoughts?

360 price level x 250 trading days per year = 90000 data points.

100 pairs = 9m data points.

Will crash your computer in no time.

The truth is if it works EOD then it works if you are monitoring intra day as well. Esp since triggers are already generated intraday.

 

[cooper1308]

360 price level x 250 trading days per year = 90000 data points.

100 pairs = 9m data points.

Will crash your computer in no time.

The truth is if it works EOD then it works if you are monitoring intra day as well.

Agree

Esp since triggers are already generated intraday.

Triggers are generated intraday yes, assuming your PC is on throughout the session.....otherwise it will revert EOD.

I'm looking more toward back-testing though. While I believe optimizing systems through backtesting isn't hugely important, I think we could get some interesting difference in performance.

 

[oby_2]

I think the time based stop is an interesting one. I have just implemented a 10 day stop after some testing and was stopped out yesterday on my first trade. Today that stock jumped in my favour significantly and what was closed for a small loss yesterday could've been closed for a nice win today.

Lesson learnt - As PT says, it is not one trade but the results of many trades that counts. I continue with the strategy....

Another lesson learned with IIF on which I was short. The evening i had openned the trade I found one reference (speculation) to it being a potential takeover target. In the coming 10 days it gained from 32c to 48c. I closed out yesterday when it came back under 40c, but at this time have still seen no solid news or information to support the sudden price change. Given other news during the time was a property value downgrade and the sale of stock by the CEO, the increase in price is still a little strange.

Lesson learnt - when one hears the phrase takeover against your short position, close out and get on with other more profitable trades. Better a black swan avoided than a small profit gained.

Cheers,
oby

 

[Pairs Trader]

Guys, no trades open for me at the moment as I'm traveling, thanks for keeping the thread alive and interesting, pleasure to read, will comment more on discussed topics when I have time.

 

[TradePro]

Hope you are all having a fun and successful Pair Trading!

I have been Pair Trading ASX 200 stocks and have been wondering what sort of risks may be associated with shorting ASX 200 stocks. From risk of shorting, I mean the fact that one takes unlimited upside risk when shorting.

Would ASX 200 stocks be considered fairly safe for shorting, ie, what would one need to be prepared for in an event of a strong increase in the short leg of the trade, practically speaking? Like say a 100% increase overnight in the value of a ASX 200 stock, or is this scenario impossible?

 

[skc]

Hope you are all having a fun and successful Pair Trading!

I have been Pair Trading ASX 200 stocks and have been wondering what sort of risks may be associated with shorting ASX 200 stocks. From risk of shorting, I mean the fact that one takes unlimited upside risk when shorting.

Would ASX 200 stocks be considered fairly safe for shorting, ie, what would one need to be prepared for in an event of a strong increase in the short leg of the trade, practically speaking? Like say a 100% increase overnight in the value of a ASX 200 stock, or is this scenario impossible?

Nothing is impoosible. The tail end of ASX200 stock can be very small companies with small market cap and high volatility. You will also find many fallen former blue chips there that are quite volatile. So it is certainly possible for a ASX200 stock to move up 100% over a matter of weeks. See EHL for a recent example: 32c low on 8 Jul, 90c high on 12 Aug.

To control (not eliminate) this risk , I adopt several rules which may be of interest to you.

1. Never short potential takeover targets
2. Cut the short leg if price move confirmed with news and strength continues
3. Implement a hard stop if pair goes against me to 10% of account size (luckily this hasn't occured yet)
4. Not trading stocks 1 week either side of reporting
5. Depending on chart pattern, I also delay shorting shares in a "potential breakout" mode as a matter of caution

So basically I filter the signals with a dose of fundamental knowledge and basic charting skills. Sometimes that means I miss out of good trades, other times I avoid potential disasters.

 

[TradePro]

Thanks SKC. Advice much appreciated ...

If I only pair stocks in ASX200 with market capitalization greater than say, 1 Billion, would this substantially reduce the shorting risk?

Also, there are some companies that are large market cap but price below a dollar, for eg. Goodman Group, GMG with market cap 1.5 Billion but with current price only $ 0.545. Does shorting risk increases if price is very low, or would this be irrelevant...

Since shorting is a part of pair trading, I want to ensure that I am not taking some unknown high risk (risk of losing more than say, 100% on a trade in the worse case scenario).

 

[bochman]

I also have a general question regarding pairs trading.

would you pair trade a stock when it is about to go ex-div?

If you long, your share will drop, if you short, you have to pay the dividend; either way, it doesn't look good?

 

[skc]

I also have a general question regarding pairs trading.

would you pair trade a stock when it is about to go ex-div?

If you long, your share will drop, if you short, you have to pay the dividend; either way, it doesn't look good?

It doesn't make any difference as you "should" be neutral when it comes to ex-dividends. i.e. Lose 10c per share but gain 10c per share in div.

The only thing you need to watch is that the signal isn't generated because of one leg or another going ex-div.

 

[skc]

Thanks SKC. Advice much appreciated ...

If I only pair stocks in ASX200 with market capitalization greater than say, 1 Billion, would this substantially reduce the shorting risk?

Also, there are some companies that are large market cap but price below a dollar, for eg. Goodman Group, GMG with market cap 1.5 Billion but with current price only $ 0.545. Does shorting risk increases if price is very low, or would this be irrelevant...

Since shorting is a part of pair trading, I want to ensure that I am not taking some unknown high risk (risk of losing more than say, 100% on a trade in the worse case scenario).

Personally I don't trade any of the fallen REITs. They are just too volatile.

May be trade stocks with comparable volatility... e.g. GMG with GPT, but not GMG with WDC. Use ATR% as a guide. The Pair Trades Finder has a volatility chart you can plot... but I never quite make sense of it.

You can never fully eliminate the risk... in either the long side or the short side. I manage that by getting to know the companies I trade reasonably well. With higher volatility pairs (e.g. pair of coal stock) I would also reduce position size to say 1/2 or 3/4 of normal.

P.S. GPT is a potential target of LLC btw.

 

[bochman]

It doesn't make any difference as you "should" be neutral when it comes to ex-dividends. i.e. Lose 10c per share but gain 10c per share in div.

The only thing you need to watch is that the signal isn't generated because of one leg or another going ex-div.

But how can you guaranteed being neutral with 2 different stocks.
They may not have the same level of dividend, or may be 1 doesn't pay dividend?

 

[skc]

But how can you guaranteed being neutral with 2 different stocks.
They may not have the same level of dividend, or may be 1 doesn't pay dividend?

I am a bit lost as to what you are trying to ask...

The ex-div effect is neutral within a single leg. But stocks going ex will affect the program's signal so you will have to take that into consideration on the pairs you trade.

 

[awg]

Some observations.

The best trades I have had, on an annualised return basis, occurred on 1-2 day turnaround, mostly on a signal triggered the day before an unexpected announcement, then closed on the announcement...lol

took profits of 3-4 % and often closed before the program indicated.

other good trades on highly correlated stocks, with 100% success shown in backtest.

Go better if I am familiar with the stock

have had good trades on volatile stocks, MCW has been a successful half

lots of dummy account trades, found the long duration ones often eventually come back to smaller losses, but then you have to factor in the differential finance carry cost on the short leg.

more number crunching/testing to be done.

I also tend to think that this method of trading will underperform in a trending market, but outperform in volatile.

sideways, I think it may outperform, but as we havent had one, cant tell yet.

can place a stop-loss or even GSL on the short leg, if you are concerned about takeover target

comments welcome

 

[cooper1308]

Thanks SKC. Advice much appreciated ...

If I only pair stocks in ASX200 with market capitalization greater than say, 1 Billion, would this substantially reduce the shorting risk?

Depending on your circumstances, US can be a far better IMO.

ASX has just over 100 stocks $1b+

NYSE over 1100+ stocks 1b+...........not even including NASDAQ

Factor in cheaper trading fees and lower financing costs and things are looking good.

Yes you'll probably know alot less about the companies initally, but it's all part of the journey...

Hours also get a little challenging. But still good

 

[SilverRanger]

Anyone mind to share their faverite aussie trading pairs?

Mine would be the nickel mining stocks (MRE, MCR & MBN) and the big 4 banks

 

[Freddy]

"technical correlation may not be that strong, however the direction is plus whenever the 2 lines diverge then tend to converge again."

PT you talk a lot about "the two lines". What is that exactly?

I am new to this game and am confused with references to two lines diverging/converging but I only see one line on a correlation chart. Can I have a very simple explanation about diverge/converge and on what charts are we referring to?

Also, with Back Testing, how do you get data from the left hand box of pairs into the right hand section? I have tried left click right click and drag without success.

 

[skc]

"technical correlation may not be that strong, however the direction is plus whenever the 2 lines diverge then tend to converge again."

PT you talk a lot about "the two lines". What is that exactly?

I am new to this game and am confused with references to two lines diverging/converging but I only see one line on a correlation chart. Can I have a very simple explanation about diverge/converge and on what charts are we referring to?

Let me try to answer this... although I can't guarantee this to be correct.

The 2 lines referred to are the ratio line and the Moving average of the ratio. Entry signal is generated when the ratio line moves far enough away (diverge) from the moving average of the ratio, while exit signal is generated when the ratio line crosses back (converage) with the MA of ratio.

 

[Freddy]

Thank you skc, that makes things clearer. I have watched charting using an 11 day exponential moving average so this is similar - my PTF is set to 150 days and 14 day lookback, is that normal?