Author Topic: Finding Japanese Net-nets  (Read 38596 times)

west

  • Sr. Member
  • ****
  • Posts: 448
Re: Finding Japanese Net-nets
« Reply #50 on: August 06, 2014, 02:34:03 PM »
All of the studies I've read say that a blind holding period of 3 years generally outperforms a blind holding period of 1 year despite the fact you may no longer be in the cheapest stocks in the market.  Why?  The winners run?  Who knows.

I have never verified this myself though, so I don't know how much I trust it.  There's a lot of backtesting strategy lore thrown around.  I've tested very little of it so it's still postulation in my mind.  For example, Greenblatt says Magic Formula picks beat the market even more so if blindly held for three years.  He also says that normalizing EBITDA over 3-5 years for the EV/EBITDA calculation improves the Magic Formula results.  I have tested this one, and this hasn't been true from my tests of it.

So, yeah, lots of quant strategy lore...



west

  • Sr. Member
  • ****
  • Posts: 448
Re: Finding Japanese Net-nets
« Reply #51 on: August 06, 2014, 02:36:05 PM »
I assume you don't want to share before buying into a full position? Fair enough :) . I really like your ideas so far, appreciated!

Nope!  I'm greedy :D

(I've also spent too much time on this to see prices for my picks go up without me.)

This being said, I could have the rest of my positions established by next Tuesday if I really worked on it.  And I've got 15 pretty good ones, some of which are close to Fujimak in awesomeness.  However, I'm dragging my feet...

TwoCitiesCapital

  • Hero Member
  • *****
  • Posts: 2200
Re: Finding Japanese Net-nets
« Reply #52 on: August 06, 2014, 03:13:00 PM »
All of the studies I've read say that a blind holding period of 3 years generally outperforms a blind holding period of 1 year despite the fact you may no longer be in the cheapest stocks in the market.  Why?  The winners run?  Who knows.

I have never verified this myself though, so I don't know how much I trust it.  There's a lot of backtesting strategy lore thrown around.  I've tested very little of it so it's still postulation in my mind.  For example, Greenblatt says Magic Formula picks beat the market even more so if blindly held for three years.  He also says that normalizing EBITDA over 3-5 years for the EV/EBITDA calculation improves the Magic Formula results.  I have tested this one, and this hasn't been true from my tests of it.

So, yeah, lots of quant strategy lore...

I'm reading a book by James Montier - he cites research that suggests that value managers have an average holding period of 4-5 years and that relative out performance for a position (as compared to an index) was the most in years 3-5. This suggests there is something to holding on BUT....

He also citrs research that splits the markets into quintiles by P/E and the bottom one nearly always out performed the second which out performed the third and so on. This suggests there is something to recycling to the cheapest every year.

So there is two possibilities:
1) the years that value doesn't outperform absolutely kill returns enough to ruin years of out performance relative to slightly lesser value stocks
2) or there's something about the persistence/momentum of a stock that moves from 2x P/E to 4x P/E that's not shared by other 4x P/E stocks.

My guess is on the later.



randomep

  • Hero Member
  • *****
  • Posts: 1062
    • Bull Bear and Value Blog
Re: Finding Japanese Net-nets
« Reply #53 on: August 06, 2014, 03:37:12 PM »
Agreed with Nate: in most cases I'm fine with an imperfect translation. I uploaded the Fujimak annual a while ago: http://www.writser.nl/fujimak/translated.htm , translated in English. I'd say it is quite understandable. Good enough for me in any case.

Great job thanks!

topofeaturellc

  • Sr. Member
  • ****
  • Posts: 318
Re: Finding Japanese Net-nets
« Reply #54 on: August 06, 2014, 04:59:20 PM »
All of the studies I've read say that a blind holding period of 3 years generally outperforms a blind holding period of 1 year despite the fact you may no longer be in the cheapest stocks in the market.  Why?  The winners run?  Who knows.

I have never verified this myself though, so I don't know how much I trust it.  There's a lot of backtesting strategy lore thrown around.  I've tested very little of it so it's still postulation in my mind.  For example, Greenblatt says Magic Formula picks beat the market even more so if blindly held for three years.  He also says that normalizing EBITDA over 3-5 years for the EV/EBITDA calculation improves the Magic Formula results.  I have tested this one, and this hasn't been true from my tests of it.

So, yeah, lots of quant strategy lore...

I'm reading a book by James Montier - he cites research that suggests that value managers have an average holding period of 4-5 years and that relative out performance for a position (as compared to an index) was the most in years 3-5. This suggests there is something to holding on BUT....

He also citrs research that splits the markets into quintiles by P/E and the bottom one nearly always out performed the second which out performed the third and so on. This suggests there is something to recycling to the cheapest every year.

So there is two possibilities:
1) the years that value doesn't outperform absolutely kill returns enough to ruin years of out performance relative to slightly lesser value stocks
2) or there's something about the persistence/momentum of a stock that moves from 2x P/E to 4x P/E that's not shared by other 4x P/E stocks.

My guess is on the later.
Yeah. Also has to do with the dispersion of returns in the cheapest quintile.  Fun fact - the worst 60% of performers in the cheapest quintile of P/B or EV/EBIT underperforms every other quintile. But what happens is that over time your portfolio becomes more focused on the 40% that work and that persists for several years.

topofeaturellc

  • Sr. Member
  • ****
  • Posts: 318
Re: Finding Japanese Net-nets
« Reply #55 on: August 06, 2014, 05:07:50 PM »
also I assume you guys know that just using EV/EBIT and throwing out the ROIC factor makes magic formula work better.

Fun fact number 2 - People who self-select on the MF website tend to reject names that score low on the ROIC factor but high on the EV/EBIT and accept names where the opposite is true. And therefore in aggregate the self-selecting people underperform the model.


I suspect the QMJ work that AQR does (which is great and you should read it) - if you disaggregated returns you see that its really giving you something that mostly loads on value but does a decent enough job of screen out a few of those cheapest quintile dogs (and it doesn't need to be many) to outperform a pure value factor.  I.e. I suspect its not buying the absolute worst businesses that adds value, not buying the better businesses.  But I don't have the underlying data to see if that's really true. 

cobafdek

  • Sr. Member
  • ****
  • Posts: 455
Re: Finding Japanese Net-nets
« Reply #56 on: August 06, 2014, 06:01:09 PM »

(I've also spent too much time on this to see prices for my picks go up without me.)


That's funny - I've spent only a little time on this last week, bought a few, only to see prices for my picks go down with me.

Must be some curse on us value investors (short-term only, I hope).  Sorry for the cold comfort.  The gods are laughing at us right now.
« Last Edit: August 06, 2014, 06:05:40 PM by cobafdek »

frank

  • Newbie
  • *
  • Posts: 15
    • Always Invert
Re: Finding Japanese Net-nets
« Reply #57 on: August 06, 2014, 09:14:56 PM »
No wonder why my jpy net-net holdings gained these days :P

I do own or used to own several of the names you mentioned but Hakuseisha has been a real dissapointment with no catalyst in sight.  However besides being a cash cow and controlled by its major customer this co. do have very attractive real estate holding in Tokyo and Osaka that alone should be valued at close to its current market cap:

Owned Property Value:  ~2,068 mil yen (estimate)
1) 東京都千代田区岩本町1-3-9 (ハクセイビル7F) 2,968 m2
Value estimated at ~1.86mil yen/m2
Mkt Value: 1484 mil yen
Carrying Value: 746 mil yen
2) 大阪市中央区西心斎橋1-4-5 (御堂筋ビル7F) 656 m2
Value estimated 0.89 mil yen/m2
Mkt Value: 584 mil yen
Carrying Value: 330 mil yen
3) 中央管财(子会社)新宿营业所 1741 M2  (valuation ignored for now)

Btw, for you JPY net-net investors out there this dog 2665 Mitsui Knowledge Industries is going to be acquired by its parent for 255 yen/share.  Good job for those with the patience to wait out on the deal.



Time for a bump. 'West' brought up a couple of very nice Japanese companies the past few weeks. In a couple of his threads there was some interesting discussion going on about investing in Japan in general, currency hedging and what approach to take when building a Japan basket. I was initially going to reply in a couple of different topics but maybe it's easier if we start dumping all our thoughts in a single topic again.

In one of the threads somebody asked when to sell. I have no good answer for that. In general my strategy is to buy stuff very cheap and to try and sell it around book value. But due to terrible corporate governance in Japan I'm not sure if a lot of these companies should be trading at or above book. However, I don't really think it's that big of an issue at current prices and I excel at deferring all difficult decisions :) . Especially if TBV is growing at a nice clip I don't mind holding on for a while if the stock price doesn't go up. Nate pointed out something interesting in another thread: maybe it is better to avoid the cheapest net/nets and to buy slightly more expensive stocks that look good on other metrics too (ROIC, P/E, FCF).

@west and others: what does your current basket look like approximately? How many companies and how big a part of your total portfolio? Mine is still work in progress but I am considering the following companies primarily:

Car Mate - 0.9x NCAV, 0.45x TBV, 6x 5yr avg P/E, CAGR 9% for the past 5 years (topic).
Isamu Paint - 0.7x NCAV, 0.4x TBV, 8x 5yr avg P/E, CAGR 6% for the past 5 years, shares outstanding decreased 5% the past few years.
Fujimak (bought already) - 0.5x TBV, 6.5x 5yr avg P/E, CAGR 10.5% for the past 5 years, expanding in Asia. (topic).
Okayama Paper - 0.7x NCAV, 0.3x TBV, 6.5x 5yr avg P/E, CAGR 7.5% for the past 5 years, shares outstanding decreased 14% (!) the past years.
Murakami - 0.46x TBV, 5.8x 5yr avg P/E, CAGR 13% for the past 5 years, 0.7x EV/EBITDA. Looks actually like a decent company (topic).

This is a second tranche of companies that look slightly less attractive at first glance:

Sugimoto & Co - .07x NCAV, 20x 5yr avg P/E. Not too enthousiast about this one at first glance. Has run up quite a bit already.
KSK - ~1x NCAV, 0.6x TBV, 11x 5yr avg P/E, buying back small amounts of shares.
Namura Shipbuilding - Still have to look at. Already some shipbuilding exposure, also have some preliminary doubts about the competitive position of Japan in this industry (topic).
Global Food Creators - Still have to look at. Looks like it ran up a bit already.
KG Intelligence - Still have to look at. Already have some magazine exposure. Not sure I need more.
Maruzen - Still have to look at. Kitchen equipment, Fujimak looks cheaper at first glance.
Hakuseisha - Still have to look at.
Maezawa Kasei - Still have to look at.
Riken Keiki - Doesn't look extremely cheap at first glance anymore.

Any interesting additions?

Each position will be around ~1 to 2%. I am looking to allocate approximately 10% of my portfolio to Japan, around 8 different companies. I'm splitting between some traditional net-nets and a couple of companies that are slightly more expensive but (I hope) make up for that in quality. By the way, I would like to do something similar with Korean preferreds. Packer had some great ideas but I have no decent broker for these as of yet.

I'm mostly jotting this down for myself. Feel free to ignore. Probably contains horrible errors.

peter1234

  • Hero Member
  • *****
  • Posts: 614
Re: Finding Japanese Net-nets
« Reply #58 on: August 07, 2014, 01:13:01 AM »
This is an awesome thread!

I was wondering if someone understands why Pabrais venture into Japanese net net land was not overly successful?

Are these net nets different from Pabrai or a similar bet at a different point in time?

Is it just that the more attractive ones are too small for an instituitional manager?

 ;)

NeverLoseMoney

  • Full Member
  • ***
  • Posts: 141
Re: Finding Japanese Net-nets
« Reply #59 on: August 07, 2014, 04:28:21 AM »
This is an awesome thread!

I was wondering if someone understands why Pabrais venture into Japanese net net land was not overly successful?

Are these net nets different from Pabrai or a similar bet at a different point in time?

Is it just that the more attractive ones are too small for an instituitional manager?

 ;)
I'm guessing he was limited to the larger net nets, given the size of his funds and that these larger net nets underperformed. I remember when I read about his results in Japan and I was shocked about his results there. That was not a typical result of an investor in Japanese net nets during that period. Other investors who blogged about their picks, like Nate and Geoff Gannon, did fine I believe.