ASX Stock Pairs Trade Journal

[Freddy]

I have received the email for the new version 3.27. A lot of bug fixes but little else at this stage. I am hoping the "Pro" version has more. I am in constant touch with the owners and they are apparently monitoring this thread, so point them in the right direction with constructive bug fixes you need. Then we all benefit.

 

[KAO]

On the tools discussion - I dropped a couple of notes HERE about it so won't re-post them, but if you're looking for a more robust tool than PTF, I've had success with ArbMaker.

The documentation is published HERE and is fairly comprehensive.

Totally not trying to rubbish any tool, but it's definitely worth a look if you're staring down the barrel of forking out for a 'pro' version of PTF.

FWIW - I spent a couple of months developing a pairs framework in R with a Shiny web front-end and started to trade using that a little over a month back.

After reading this thread and seeing you guys all using commercial tools (it hadn't actually occurred to me there might be a COTS option and I started writing code before I even looked...) I did some research and cut over to AM. It has some convenience functions I like and wanted to focus on trading instead of spending more time coding.

Recently started to integrate my R framework with AM via it's SQL DB so will eventually end up with the convenience functions of AM and the flexibility to add features via the R framework.

Currently in GMG / ARF. Liquidity on the ARF side is a lot lower but have that as the short leg with an OTC CFD so it's less of an issue.

Watching WSA / IGO for a potential open at the start of the week with a 7.66% target on the table.

 

[KAO]

Slow couple of weeks due home relocation, getting back to some sort of routine now...

Closed L GMG / ARF on 26/10 for + 2.18% - ARF sold down to 1.808 for a brief moment on 24/10, had an opportunity to close the trade at ~1.90 and didn't take it. Tech issues on 25/10 prevented access from work location, missed further opportunity to close out during market hours.

Opened L ORI / RIO just now at 1.86 sigma, the pair back tests at 84% W/L over 25 trades with 1.82 ROIC and 3.8% reversion target currently on offer.

 

[skc]

Slow couple of weeks due home relocation, getting back to some sort of routine now...

Closed L GMG / ARF on 26/10 for + 2.18% - ARF sold down to 1.808 for a brief moment on 24/10, had an opportunity to close the trade at ~1.90 and didn't take it. Tech issues on 25/10 prevented access from work location, missed further opportunity to close out during market hours.

Opened L ORI / RIO just now at 1.86 sigma, the pair back tests at 84% W/L over 25 trades with 1.82 ROIC and 3.8% reversion target currently on offer.

FYI ORI reporting this Friday 4 Nov.

 

[KAO]

FYI ORI reporting this Friday 4 Nov.

Thanks... I've been using the Morningstar Corporate Calendar (http://www.morningstar.com.au/Stocks/CorpCalendar) which lists 16/11 as the Ann. Report, but doesn't carry anything for 04/11.

Yahoo.com (US) (http://finance.yahoo.com/quote/ORI.AX?p=ORI.AX) lists 05/11 as the Earnings Date, which makes sense given the TZ delta.

Yahoo.com.au (AU) (https://au.finance.yahoo.com/q?s=ORI.AX) lists N/A for Next Earnings Date, which is obscure... you'd think they'd match up.

The Orica Investors calendar (http://www.orica.com/Investors/Shareholder-Calendar) correctly lists as 04/11...

Out of interest, can you recommend a single, collated, reference source for earnings and corporate actions, or do you typically check the investor website for each company individually?

 

[VSntchr]

Out of interest, can you recommend a single, collated, reference source for earnings and corporate actions, or do you typically check the investor website for each company individually?

Fn arena. Pretty commonly used for traders.

 

[skc]

Out of interest, can you recommend a single, collated, reference source for earnings and corporate actions, or do you typically check the investor website for each company individually?

Yes you'd think that in the 21st Century, in a developed market like ASX, something like this would be available.

I read quite a few analyst reports daily and many of them have a calendar or weekly "what's ahead" type of section. You also build more knowledge over time so you will get to know some of the companies that report out of cycle like ORI. Lastly if you are not trading too many pairs, then a quick check to confirm the absence of scheduled announcements is probably prudent (either before putting on the trade, or certainly before close). You can simply scroll down announcements from the year before to know roughly when that particular company reports.

 

[KAO]

Fn arena. Pretty commonly used for traders.

Yes you'd think that in the 21st Century, in a developed market like ASX, something like this would be available.

I read quite a few analyst reports daily and many of them have a calendar or weekly "what's ahead" type of section. You also build more knowledge over time so you will get to know some of the companies that report out of cycle like ORI. Lastly if you are not trading too many pairs, then a quick check to confirm the absence of scheduled announcements is probably prudent (either before putting on the trade, or certainly before close). You can simply scroll down announcements from the year before to know roughly when that particular company reports.

Thanks guys... I've been checking them beforehand but probably not thoroughly enough. Started to check the corporate calendar on individual stocks investor websites as well as a couple of other sources. I consumed a FN Arena trial a couple of months back and will have to revisit it.

Opened a number of new trades this morning.

OPEN L PGH / ORA - 13 / 69.2 / 1.13, 0.0521
OPEN L SGP / FET - 5 / 80 / 1.17, 0.0302
OPEN L ORI / RIO - 24 / 83.3 / 1.68, 0.0283

The stats are: Trades in backtest window, Winning percentage, Net ROIC, % Deviation from 21 Obs. Mean.

The simple backtest configuration I'm using to screen is enter where 21 obs. Z-Score is 1.6-2.4 sigma and has been under reversion for at least one observation.

EDIT: I'd closed the earlier position in ORI / RIO prior to the ORI earnings for -0.6216%. This is a new position.

 

[KAO]

CLOSED L PGH / ORA +3.7040% ROIC (Net of commissions)

 

[KAO]

OPEN L ARF / VCX - 10 / 100 / 2.08 / 0.0526

There's basically no liquidity in ARF so I took it as an OTC CFD then wound up ordering in the middle of the VCX spread to get a match before having to leave the desk for a little while. I wouldn't have taken the trade given the liquidity and cap mismatch unless the backtest results and ratio delta were as good as they are.

Half closed the SGP / FET position in profit but couldn't get a fill on the SGP side without crossing the spread again and wound up staring the book down until 15:58 then dumping it in the middle. Didn't get a fill before close so I'm carrying the uncovered long overnight... we'll see how that winds up in the morning.

I spent some time today refining my trading workflow to identify opportunities to automate more where appropriate and refine the manual bits - particularly around being able to efficiently disqualify signals as early on in the process as possible where no trade will result.

I'm struggling a little taking signals / assessing opportunities out of a large-ish pool of mined pairs. Presumably the answer to this is to refine the list of pairs down to a manageable portfolio of 15-20 that I've already performed fundamental checks on in advance.

Some of you guys seem to be taking signals from quite large lists of pairs - I'd be really keen to understand how you manage signal qualification and pre-trade check workflows in a little more detail.

Happy to share mine in more detail if that's not talking out of school / useful...

 

[skc]

I'm struggling a little taking signals / assessing opportunities out of a large-ish pool of mined pairs. Presumably the answer to this is to refine the list of pairs down to a manageable portfolio of 15-20 that I've already performed fundamental checks on in advance.

Some of you guys seem to be taking signals from quite large lists of pairs - I'd be really keen to understand how you manage signal qualification and pre-trade check workflows in a little more detail.

Happy to share mine in more detail if that's not talking out of school / useful...

Can you better define what you meant by fundamental checks / pre-trade check workflow? What are you actually checking?

There are several times of the year when you know the fundamentals are definitely / potentially at play. Reporting seasons, ex-dividend dates, AGM season. A quick check of company announcement in last 2-4 weeks should show you if there was any major news explaining recent share price movements.

On the more core fundamentals (like is BHP or RIO better leveraged to I/O prices)... it's not possible to have in-depth understanding of every company... but with experience and longer exposure to the market you can certainly develop better "working knowledge" that works well enough.

 

[KAO]

The attachment is basically how I'm working right now.

0. Signal Enrichment & Shortlisting - Update all the pairs in the list and dump those that are +2/-2 SD from 21 observation mean to CSV. I've written a pricing tool in R that takes the delta of the ratio from the mean and uses it to calculate a rough trade potential dollar amount with my leg sizes and brokerage factored in. I take the trades that are worth looking at (generally anything 2.5%+ net brokerage), import them to a portfolio and backtest them with a simple strategy.

1. - Signal Validation - For signals with worthwhile potential and acceptable backtest results, I manually eyeball a bunch of statistical output to make sure the Z-Score is aligned with the spread/ratio (eg, reversion isn't just due to time decay, spread is actually reverting), PACF and Q-Q Normality (checking for normally distributed returns to validate the Engle-Granger results), manually validating the log price series over a couple of time frames to see if the price action makes sense.

2. Event Risk - Checks company announcements, news digest, corporate and dividend calendars for upcoming events. I'm basically just using Morningstar for this - still learning what sort of events will impact price movement (eg. AGM before or after reports being released etc...). I'll disqualify a trade with a significant event in the trade horizon. I'll also stay out of a trade with a fundamental trigger, for example the IOF announcement on 14/11 that triggered a load of REITs.

3. Fundamental Fit - Checks company summary description / Morningstar analyst report for validation of fundamental thesis (eg. general stuff like avoiding miners in different locations / metals, pairs like PGH and ORA that are a statistical fit with JHX and BLD but a poor fundamental one..).

Review basic ratios like P/E, P/B, and checking out DCF to avoid taking a long/short position in a relatively over or under valued leg. I've started using SimplyWall.St to quickly form a position on whether one leg is fundamentally stronger or weaker / over or under valued than the other and using this as a filter to avoid trades that aren't on the correct side of the assessment.

4. News Flow - Quick check of a bunch of open news sources (I totally need to sign up for FN Arena... actually... I'm gonna go do this tonight) for anything that would keep me out / wasn't apparent in the announcements.

5. Trade Execution - General execution and pre-trade management.

As you call out, getting speed is probably more a function of longer term market exposure (I've only been at it a few weeks...), but anything to improve the workflow / process is a bonus...

 

[skc]

The attachment is basically how I'm working right now.

KAO, when you said "workflow" I thought you were using too strong a word... but from what you've posted it is very well structured and comprehensive. Hats off to you.

FWIW, it's way more than anything I've ever done whether from the beginning (due to ignorance) or years down the track (due to experience/laziness).

My comments below in blue.

0. Signal Enrichment & Shortlisting - Update all the pairs in the list and dump those that are +2/-2 SD from 21 observation mean to CSV. I've written a pricing tool in R that takes the delta of the ratio from the mean and uses it to calculate a rough trade potential dollar amount with my leg sizes and brokerage factored in. I take the trades that are worth looking at (generally anything 2.5%+ net brokerage), import them to a portfolio and backtest them with a simple strategy.

I am not totally sure what this means but it sounds like you are just making sure the potential profits in the trade is worth taking. I am guessing this doesn't take a lot of time and it's something I do as well. However, a quick eyeball of the ratio chart in Pairs Trade Finder will suffice most of the time.

1. - Signal Validation - For signals with worthwhile potential and acceptable backtest results, I manually eyeball a bunch of statistical output to make sure the Z-Score is aligned with the spread/ratio (eg, reversion isn't just due to time decay, spread is actually reverting), PACF and Q-Q Normality (checking for normally distributed returns to validate the Engle-Granger results), manually validating the log price series over a couple of time frames to see if the price action makes sense.

This seems sound and I like the bold part. Perhaps another quick way to check is a simple "Total return" comparison (see image) of the 2 stocks in question plus any other applicable peers over a week and month timeframe. So you get a sense of how the current divergent compares to longer timeframes.



Again, eyeballing the ratio chart will reveal much of this information at a glance.

2. Event Risk - Checks company announcements, news digest, corporate and dividend calendars for upcoming events. I'm basically just using Morningstar for this - still learning what sort of events will impact price movement (eg. AGM before or after reports being released etc...). I'll disqualify a trade with a significant event in the trade horizon. I'll also stay out of a trade with a fundamental trigger, for example the IOF announcement on 14/11 that triggered a load of REITs.

A must do and it will get done quicker over time.

BTW, not sure if the announcement by IOF on 14/11 is that meaningful for the sector...

3. Fundamental Fit - Checks company summary description / Morningstar analyst report for validation of fundamental thesis (eg. general stuff like avoiding miners in different locations / metals, pairs like PGH and ORA that are a statistical fit with JHX and BLD but a poor fundamental one..).

Much of this should be done before the pair are backtested. Once you are comfortable enough to put the pair into your pairs monitor, you don't need to do these again everytime you get a signal. So this process seems to be in the wrong place of the workflow.

Review basic ratios like P/E, P/B, and checking out DCF to avoid taking a long/short position in a relatively over or under valued leg. I've started using SimplyWall.St to quickly form a position on whether one leg is fundamentally stronger or weaker / over or under valued than the other and using this as a filter to avoid trades that aren't on the correct side of the assessment.

I am not sure this is helpful considering the timeframe and size of movements you are trading. Is there any reason to suggest that a higher / lower PE stock will over/underperform over a random period of 3 to 15 days?

And if you are caught on the wrong side occasionally... that's what a stop loss is for.

4. News Flow - Quick check of a bunch of open news sources (I totally need to sign up for FN Arena... actually... I'm gonna go do this tonight) for anything that would keep me out / wasn't apparent in the announcements.

May be... but I don't do it during the trading day. Analyst reports (or summary of them) are often useful and they should be skim read to build up knowledge on particular stocks. You will catch things like... there's an government inquiry into private hospital charges, or XYZ has a blockbusting drug trial results due next month, or OPEC has a meeting next week, or a big overseas takeover in similar sector etc etc. It's not possible to know everything, but the more you know the better the outcome over the long term.

This protects you mostly from any known unknowns. You will still get caught by the unknown unknowns... and that's what your positions sizing strategy is for.

5. Trade Execution - General execution and pre-trade management.

This is something that will take a long time to improve as well. What is the market doing? What's the depth doing? Should I cross the spread, take the centre point or wait on the bid/ask? Should I take 10,000 shares in one hit or 5 x 2,000 shares? Can I go long first and sell the short leg after it moves up another 2 ticks?

Master this and you are almost better off as a day trader!

As you call out, getting speed is probably more a function of longer term market exposure (I've only been at it a few weeks...), but anything to improve the workflow / process is a bonus...

It's a balance between what is ideal and what is practical... plus what pairs style you are actually trying to achieve.

Are you are doing 50 trades a year part time using end of day prices? Or are you doing 200 pairs a month arbing some short term noises? Or are you trading fundamentally over 3-6 months using statistics as a trigger? These different styles will impact where you should aim for on the spectrum of perfect signal and practicality.

 

[VSntchr]

The attachment is basically how I'm working right now.

This is impressive. A great road-map that aspiring pairs traders would do well to take note of.

It's a balance between what is ideal and what is practical... plus what pairs style you are actually trying to achieve.

Are you are doing 50 trades a year part time using end of day prices? Or are you doing 200 pairs a month arbing some short term noises? Or are you trading fundamentally over 3-6 months using statistics as a trigger? These different styles will impact where you should aim for on the spectrum of perfect signal and practicality.

Some great comments in your reply SKC. Reading through KAO's post and then your reply had me reflecting on how many different tools/sources are actually required in the pairs trading process.
Just reading some of these made me realise that I am already doing many of these things, but can probably improve the efficiency. By clearly understanding the goal I am trying to achieve with each process I can then select the best tool for the job rather than just using the tool I have always used as a matter of habit.

 

[SilverRanger]

Hi guys,

Since for the month of December I'm a full time trader I thought I should try to be more active here.

I'll start with some interesting pair trades I took recently:

VOC/TPM - Decided to catch this falling knife at the start of the month, as I wasn't sure all the VOC troubles were all that VOC-specific. Closed yesterday for a good profit (TPM fell more than VOC), and glad I did! Plenty of trading opportunity for these two today as well (went long for both at open, following TPM's report).

AHG/APE - This one is still on-going, more of a merger arbitrage trade than technical (although I did get a signal when I bought). One pending risk out there is the ACCC decision on the car insurance commission anytime, but since this is gonna hit both so I'm not too concerned, and the change itself can be another reason for consolidation. APE currently holds 20% of AHG, but haven't done anything for a few years.

SLF/GPT - Nothing special other than SLF, quite an useful instrument to couple with a single strong/weak AREIT stock. The downside, SLF is not shortable most of the time, and can't really trade it in open/close auctions.

 

[SilverRanger]

Interestingly, Iron Mountain (INM) is now part of the A-REIT index (XPJ)!

Maybe this could be a good buddy to WFD and GMG due to its international footprint (69% revenue in US, 17% Europe)

 

[skc]

Interestingly, Iron Mountain (INM) is now part of the A-REIT index (XPJ)!

Maybe this could be a good buddy to WFD and GMG due to its international footprint (69% revenue in US, 17% Europe)

INM has always been a REIT in US and pays quarterly dividend... perhaps it's a trust of industrial properties with a business attached!

 

[SilverRanger]

INM has always been a REIT in US and pays quarterly dividend... perhaps it's a trust of industrial properties with a business attached!

Apparently their steel racking structures for record keeping are classified as "real estate", and roughly 70% of their revenue comes from that. Given this is a non-cyclical and stable revenue, plus a 90% payout, you have a synthetic REIT!

 

[SilverRanger]

The goldies had a pretty hard run during the last 2 weeks, that seemed to have created some good PT opportunities:

NST @ 3.62/EVN @2.12
Technical:
Support: 1.7
RSI: 20
Correlation: 97.5



Similar fundamentals from my limited understanding of goldies



Target: 1.8? Will see

 

[edman79]

Hi SKC,
Are you still pair trading? If so in what capacity and how are you finding it, if not what are you concentrating on at the moment?
Personally I am pair trading very little compared to 5 years ago. I have found that either the edge has been arbed away or I need the volatility we had closer to the GFC.

Ed.