Jump to content

Super Awesome Paper on Quant Value Strategies in Japan


west

Recommended Posts

Hey folks!  Over the course of the last few years, I've been collecting financial data for Japanese companies and looking at how different quantitative value strategies have performed in Japan.

 

I recently wrote a paper on how eight different quantitative value strategies performed in Japan during 2013.  Suffice it to say, they did very well.  The top three strategies returned over triple-digit returns.  In just one year's time.  (Although, to be frank, it was an exceptional year for the market.  Triple-digit returns are still pretty cool in my book though.)

 

You can find my paper here:

 

http://www.dusthimer.net/PDF/Quantitative-Value-in-Japan.pdf

 

Skip to the top of page eight (to Table 1) if you just want to see how the strategies performed.

 

Enjoy!

 

PS-

For some quick self-promotion (which was OK-ed by Sanjeev), I'm currently looking for a job doing equity research.  Check out the rest of www.dusthimer.net if you want to see some of the other stuff I've done.

 

Shoot me an email at brandt.dusthimer@gmail.com if you want to talk.  I'll be out the rest of today, but I'll get back to you as soon as I can.

Link to comment
Share on other sites

I've interacted with Brandt via this board, but also offline via email a number of times.  He's intelligent, savvy, and well spoken, I'd recommend him for a job.  Heck, if I had the ability to hire him I would try, unfortunately I'm not hiring right now.

 

I read the Japanese strategies paper a few months back and shifted my approach as a result.  I am investing in more low P/B with growth companies rather than just cheap bottom of the barrel net-nets. 

Link to comment
Share on other sites

Ahem . . . Brandt, you've done a grave disservice to those of us with an affinity for these strategies.  If they become too popular, they stop working.  You may have just inaugurated a long period of underperformance for us with this post.  Reflexivity, you know.  Please take it down.

 

And best of wishes on the job search!

Link to comment
Share on other sites

Great paper, West. Thanks.

 

One additional data source is http://www.kaijinet.com/jpexpress/. This site has English translations of the financials for all Japanese companies that file their statements in XBRL format. Also, the site is scrapable.

 

MSN Money, Financial Times etc probably pull from the raw XBRL data in some form or other. However, these sites simplify the financial statements by aggregating (and dropping) accounts. For example, check out the non-current asset accounts for Isamu on MSN Money — they don’t even add up! In comparison, the kaijinet site breaks down the accounts, which can help you arrive at better estimates of excess cash etc.

 

Good luck with your job search.

Link to comment
Share on other sites

Great paper, West. Thanks.

 

One additional data source is http://www.kaijinet.com/jpexpress/. This site has English translations of the financials for all Japanese companies that file their statements in XBRL format. Also, the site is scrapable.

 

MSN Money, Financial Times etc probably pull from the raw XBRL data in some form or other. However, these sites simplify the financial statements by aggregating (and dropping) accounts. For example, check out the non-current asset accounts for Isamu on MSN Money — they don’t even add up! In comparison, the kaijinet site breaks down the accounts, which can help you arrive at better estimates of excess cash etc.

 

Good luck with your job search.

 

WOW......I wish you would have registered and posted this three years ago.. This is an excellent resource, looks like most statements go back to 2009.

 

Very cool! Thanks

Link to comment
Share on other sites

Great paper, West. Thanks.

 

One additional data source is http://www.kaijinet.com/jpexpress/. This site has English translations of the financials for all Japanese companies that file their statements in XBRL format. Also, the site is scrapable.

 

MSN Money, Financial Times etc probably pull from the raw XBRL data in some form or other. However, these sites simplify the financial statements by aggregating (and dropping) accounts. For example, check out the non-current asset accounts for Isamu on MSN Money — they don’t even add up! In comparison, the kaijinet site breaks down the accounts, which can help you arrive at better estimates of excess cash etc.

 

Good luck with your job search.

is there a version of this for Chinese companies ?

 

Also Good luck West!

I wish i had your Quant skills.

Link to comment
Share on other sites

Hey man, sounds like we come from similar backgrounds.  I was a systems/software engineer for a long time in SV.  I'm now working as a financial analyst at a small investment firm.  Just started this week.  Congrats on passing the CFA's all on the first go.  Just curious, how long did it take you to finish that?  Also, why the jump from software? 

Link to comment
Share on other sites

Great paper, West. Thanks.

 

One additional data source is http://www.kaijinet.com/jpexpress/. This site has English translations of the financials for all Japanese companies that file their statements in XBRL format. Also, the site is scrapable.

 

MSN Money, Financial Times etc probably pull from the raw XBRL data in some form or other. However, these sites simplify the financial statements by aggregating (and dropping) accounts. For example, check out the non-current asset accounts for Isamu on MSN Money — they don’t even add up! In comparison, the kaijinet site breaks down the accounts, which can help you arrive at better estimates of excess cash etc.

 

Good luck with your job search.

 

I second Nate's wow!  (And that I wish I knew about this earlier.)  Many thanks for this!

 

It will definitely come in handy, especially since MSN Money just changed their website design.  So I needed to write a new scraper anyways :)

Link to comment
Share on other sites

Hey man, sounds like we come from similar backgrounds.  I was a systems/software engineer for a long time in SV.  I'm now working as a financial analyst at a small investment firm.  Just started this week.  Congrats on passing the CFA's all on the first go.  Just curious, how long did it take you to finish that?  Also, why the jump from software?

 

Hey opihiman.  On the CFA stuff, I don't remember.  I didn't have a job for (most of) the time I was doing Level II and Level III, so I think it was about eight to ten weeks?  However, I'm a little "special" in the fact that I work practically all the time.  When I did those eight to ten weeks, most days I was doing more than ten hours of study a day, and I know I clocked in at least one fourteen hour day once a week.  Sooo... Assuming eleven hours a day on average, seven days a week... Maybe 600 to 800 hours?

 

Sorry if this isn't a great answer...

 

On why I'm switching out of software, I like learning the 10% that matters 90% of the time in all subject areas.  You name it.  Computers...  Bikes...  How specific industries like, why not?, the fertilizer industry works.  When you're doing investment research, you're doing nothing but learning about new stuff.  All the time!  I like that.

 

In software I felt like I learned everything there was to know in my area (minus the extreme nuances) and that I was just repeating the same stuff over and over.  It became very boring.  Plus, I was subject to a lot of forces outside of my control, which I didn't really care for.

 

(I still like software.  A lot.  I just don't want to do systems coding 40+ hours a week any more.)

 

Just curious, but is that a similar reason as to why you switched?

Link to comment
Share on other sites

Hey man, sounds like we come from similar backgrounds.  I was a systems/software engineer for a long time in SV.  I'm now working as a financial analyst at a small investment firm.  Just started this week.  Congrats on passing the CFA's all on the first go.  Just curious, how long did it take you to finish that?  Also, why the jump from software?

 

Hey opihiman.  On the CFA stuff, I don't remember.  I didn't have a job for (most of) the time I was doing Level II and Level III, so I think it was about eight to ten weeks?  However, I'm a little "special" in the fact that I work practically all the time.  When I did those eight to ten weeks, most days I was doing more than ten hours of study a day, and I know I clocked in at least one fourteen hour day once a week.  Sooo... Assuming eleven hours a day on average, seven days a week... Maybe 600 to 800 hours?

 

I looked online and I'm seeing estimates that say most candidates study for more than 300 hours per level.  Still, my numbers (if they're even correct... I'm doing an estimate at this point) are double that.  I'm a little slow when it comes to studying stuff, but I would still take my estimated hours of studying with a massive grain of salt.

Link to comment
Share on other sites

Does everyone know about the reports that fidelity gives out for free that have 3 years of data translated?

 

No, I had no idea about that!  I don't have a Fidelity account, but I would happily open one with little bit of money if this service is any good.  Could you attach the reports for Fujimak (JP:5965), Sapporo Clinical Labs (JP:9776), and Nansin Co (JP:7399) to this thread so I can see what they're like?  It would be much appreciated!

Link to comment
Share on other sites

Hey man, sounds like we come from similar backgrounds.  I was a systems/software engineer for a long time in SV.  I'm now working as a financial analyst at a small investment firm.  Just started this week.  Congrats on passing the CFA's all on the first go.  Just curious, how long did it take you to finish that?  Also, why the jump from software?

 

Hey opihiman.  On the CFA stuff, I don't remember.  I didn't have a job for (most of) the time I was doing Level II and Level III, so I think it was about eight to ten weeks?  However, I'm a little "special" in the fact that I work practically all the time.  When I did those eight to ten weeks, most days I was doing more than ten hours of study a day, and I know I clocked in at least one fourteen hour day once a week.  Sooo... Assuming eleven hours a day on average, seven days a week... Maybe 600 to 800 hours?

 

Sorry if this isn't a great answer...

 

On why I'm switching out of software, I like learning the 10% that matters 90% of the time in all subject areas.  You name it.  Computers...  Bikes...  How specific industries like, why not?, the fertilizer industry works.  When you're doing investment research, you're doing nothing but learning about new stuff.  All the time!  I like that.

 

In software I felt like I learned everything there was to know in my area (minus the extreme nuances) and that I was just repeating the same stuff over and over.  It became very boring.  Plus, I was subject to a lot of forces outside of my control, which I didn't really care for.

 

(I still like software.  A lot.  I just don't want to do systems coding 40+ hours a week any more.)

 

Just curious, but is that a similar reason as to why you switched?

 

Hey West,

 

Ah, very cool.  That's awesome, man.  My new boss was telling me things like, "The CFA is no joke, and it's almost equivalent to doing a masters degree."  I don't know about that.  I've heard varying mix things about the difficulty of the CFA.  But, it sounds like it was a walk in the park for you.  That's cool.

 

No.  To be frank, I got burnt out on IT.  I've done it for about 15 years now.  And, quite differently to you, I just got tired of learning new technologies all the time.  Technologies that were changing or going nowhere.  Case in point, about 7 years ago, J2EE and SOA webservices was being heavily pushed by players like IBM, Sun, Oracle, BEA.  I worked in that for a long time, but I always felt that it was way too cludgey, cumbersome, and over-engineered.  I felt that eventually, more simplified frameworks and languages would overtake J2EE.  It turns out I was right!  Haha.  I spent all this time and energy learning something that just went into the dirt--well, mostly.  Anyways, it sounds like you just got bored of it and wanted something new.  I'm sure you will find something.  I'll ask my boss if he needs another analyst.  You have a great background.  And, you're way ahead of the other analysts here.

 

One last thing, since you have a software engineering background and are interested in the quant approach, have you ever tried creating automated trading/investing systems?  I am trying to get ahold of a good, raw database like CompuStat and test my own strategies on it.  I had one idea several years ago that I finally got to try out.  Someone on here, I forgot who, sent me the data I needed by exporting it out of CompuStat.  It was pretty simple.  I just needed end of day price data of every stock that traded on the S&P in the CompuStat history.  I imported the data into a JVM's 2 GB heap space.  Then I ran my my backtest and walk forward tests via Java classes.  It spits out basic statistics like draw downs, stocks that passed the filter, etc...  Anyways, so far, that one idea is seriously outperforming.  Like, I'm talking 40%+ CAGR.  The only problem though with my strategy is the lack of oppty's.  You pretty much have to put a 100% of your portfolio in one idea.  Sometimes, there are several stocks that pass the filter.  There would be more, but I narrow down the range to over a certain market cap and just the S&P.  The tracking portfolio I set up for testing this idea out in a real world walk forward test is up over four times since 2010.  I put 10% of my portfolio in that tracking portfolio, and now I'm thinking about increasing the overall percentage.  I have other ideas I want to test out.  But, it's not possible without other stock metrics.  Anyways, if you're ever interested in doing some quant strategies, let me know.  I thought my current firm would have access to a raw database like CompuStat.  Apparently, they use online databases, and that's not really going to work for me.

Link to comment
Share on other sites

Whoops, I missed your post opihiman...

 

Ah, very cool.  That's awesome, man.  My new boss was telling me things like, "The CFA is no joke, and it's almost equivalent to doing a masters degree."  I don't know about that.  I've heard varying mix things about the difficulty of the CFA.  But, it sounds like it was a walk in the park for you.  That's cool.

 

Not quite a walk in the park.  Not by any stretch ;).  I was very lucky that I had nothing going on (even employment-wise) at the time.

 

No.  To be frank, I got burnt out on IT.  I've done it for about 15 years now.  And, quite differently to you, I just got tired of learning new technologies all the time.  Technologies that were changing or going nowhere.  Case in point, about 7 years ago, J2EE and SOA webservices was being heavily pushed by players like IBM, Sun, Oracle, BEA.  I worked in that for a long time, but I always felt that it was way too cludgey, cumbersome, and over-engineered.  I felt that eventually, more simplified frameworks and languages would overtake J2EE.  It turns out I was right!  Haha.  I spent all this time and energy learning something that just went into the dirt--well, mostly.  Anyways, it sounds like you just got bored of it and wanted something new.

 

It was a little bit of column A, a little bit of column B for me.  I was getting bored because a lot of it was all the same stuff.  But I was also very frustrated by the whole "having to learn a whole new language to do the exact same thing I was already doing" thing.  It's great that there's constant innovation in tech, but I like to invest in learning stuff that's going to be around for the long haul.  I don't think that's true for most things (outside of theory) in tech...

 

One last thing, since you have a software engineering background and are interested in the quant approach, have you ever tried creating automated trading/investing systems?  I am trying to get ahold of a good, raw database like CompuStat and test my own strategies on it.  I had one idea several years ago that I finally got to try out.  Someone on here, I forgot who, sent me the data I needed by exporting it out of CompuStat.  It was pretty simple.  I just needed end of day price data of every stock that traded on the S&P in the CompuStat history.  I imported the data into a JVM's 2 GB heap space.  Then I ran my my backtest and walk forward tests via Java classes.  It spits out basic statistics like draw downs, stocks that passed the filter, etc...  Anyways, so far, that one idea is seriously outperforming.  Like, I'm talking 40%+ CAGR.  The only problem though with my strategy is the lack of oppty's.  You pretty much have to put a 100% of your portfolio in one idea.  Sometimes, there are several stocks that pass the filter.  There would be more, but I narrow down the range to over a certain market cap and just the S&P.  The tracking portfolio I set up for testing this idea out in a real world walk forward test is up over four times since 2010.  I put 10% of my portfolio in that tracking portfolio, and now I'm thinking about increasing the overall percentage.  I have other ideas I want to test out.  But, it's not possible without other stock metrics.  Anyways, if you're ever interested in doing some quant strategies, let me know.  I thought my current firm would have access to a raw database like CompuStat.  Apparently, they use online databases, and that's not really going to work for me.

 

Trading strategies aren't really my cup of tea.  I'm pretty lazy when it comes down to it.  I like to sit around and read.  I maybe check the market once or twice a day.  Being tied to a desk constantly watching a screen isn't really my style.  Maybe not all trading strategies are like this, but that's my prejudice.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...