Quantitative, Mechanical Net-Net Investing: Foolish or Fantastic?

[DoddDisciple]

This is probably odd for a first post, but I am trying to elicit feedback and hopefully this is a good place for it. I know this may end up being a case of "too long, didn't read," but if not, I really appreciate any comments and thank everyone who posts in advance :)

 

I am seriously considering quantitative, mechanical basket net-net investing as a long-term strategy. I have a general idea of how to accomplish this, and perhaps I am overthinking things trying to find more information since a lot of papers, posts, and what have you make it look like just a systematic accumulation of net-nets can lead to at least decent results.

 

In particular, there is one article by Geoff Gannon that presents a decision framework that is compelling: www.gurufocus.com/news/121824/how-to-pick-netnets. ValuePrax also has a summary of the "net-net magic formula" that Geoff develops in the article: http://valueprax.wordpress.com/2012/05/16/notes-how-to-pick-net-nets-two-philosophies-from-geoff-gannon-and-gurpreet-nurang-geoffgannon-net-nets/.

 

In brief, net-nets are extreme P/B stocks, so it is clear that we already have the cheapness factor taken care of. However, Geoff presents a quality screen of F-score and insider ownership being used for ranking which appears to lead to outperformance.

 

He presents some portfolio return information using StockScreen123 from March 31st 2001 - February 7th 2011. In a portfolio of 12 US net-nets, sold and replenished annually, 20% per annum is generated. Comparing net-nets less than $25M market cap and greater than $25M market cap, the former generates 21% p.a. and the latter 11% p.a., showing a size bias. From experience, he thought that one particular group, that with the highest insider ownership and below $25M market cap, would generate excellent returns. I think I'd have to describe the calculated 32% p.a. for this group as beyond excellent. For another metric, F-score, the return was 25% p.a., though I am not sure if this is among all net-nets, or with just those below $25M market cap.

 

At this point, I honestly have not seen anything better than the above when it comes to a systematic investing in net-net stocks. StockScreen123 uses Compustat which is an excellent sampling of NYSE, NYSE-AMEX, NASDAQ, and some OTC stocks that would be able to capture most, if not all, non-OTC US net-nets. Picking stocks where the owners have an incentive to unlock value, where the price is below liquidation value, and where there can potentially be a quality filter via the F-score seems like a foolproof way to generate returns. I don't know if anyone has ever done this (probably not, since it isn't all that much fun for those actually interested in security analysis), but I'd like to pose some questions to see in what ways this can be optimized further and how I can further solidfy my gameplan:

 

(1) Sizing - I use Fidelity right now for investing, so I would want to initiate positions so as to keep the round trip commission at 1% or less, so $1,590 with $7.95 per buy/sell. Geoff uses a portfolio of 12 in his screens, but I could see doing baskets of 15 or 20 (for a nice 4% of assets per stock), so at least $23,850 to $31,800 per basket purchase. I could see myself doing several basket purchases a year for the foreseeable future and I have several decades as a compounding runway.

 

(2) Ranking - Combined score for insider ownership (0.0 - 9.0, example 5.0 = 50% ownership) and Piotroski F-score (0.0 - 9.0), rank high to low and buy top X in equal weighting.

 

(3) Pricing - Here's an area where I am a little unsure. I believe Geoff's just looking at stocks where current assets exceed total liabilities, versus Graham's (1.0 x cash) + (0.75 x accounts receivable) + (0.5 x inventory) + (sometimes 0.1 x fixed assets) and also his 66% of the spread between current asset and total liabilities.

 

(4) Selling - Geoff recommends holding for 3 years. This seems to be an area where almost everyone else believes to sell close to NCAV, which could make sense if you were initially selective of the stocks purchased. However, the more I think about it, a 3 year holding makes sense for a variety of reasons. For one, I have read that of all net-net, about 50% are still net-nets one year later, meaning that the typical 1 year holding period could shuffle what is already purchased to replenish the basket. By 3 years, perhaps only 12.5% of the stocks will be net-nets, and the rest will have had some activity, even if that is having gone to zero. Also, the 3 year period allows for semi-market neutral activities, like going private, buyouts, etc. to occur and unlock the value prematurely.

 

(5) Buying - Here, I will use limit orders on the net-net, NCAV, or NNWC figure calculated. I could possibly even try to earn on the bid-ask spread by pricing 0.005% less or more than my calculated value or the last trade price, whichever is less, since there is no rush to accumulate shares if there is a 3 year holding period.

 

Beyond the above, I have wondered about several other aspects:

 

(1) Timing - Are there any "January" effects in net-nets and, even if there is, does it matter with a 3 year holding period? I recall there appeared to be one in one of the papers Greenbackd contributed to, where there were large sell-downs before the end of December and the prices remained depressed through some of January due to tax loss harvesting. Even then, I almost feel a random initiation of a basket of limit orders as enough assets are accumulated to keep the buy/sell at 1% of less makes the most sense.

 

(2) International - Could the same model be applied internationally, possibly even eventually having a global net-net portfolio, where the highest f-score and insider ownership stocks regardless of location are sought? Montier found that a basic net-net strategy worked in all markets, though of course in differing degrees. Would the same methodology Geoff proposes work? He talks about avoiding Chinese reverse merger net-nets (which the F-score should catch these as well as any other similar offerings in the future), but I am not sure if foreign net-nets are to be avoided in general.

 

(3) Data - I know StockScreen123/Portfolio123 is one option, but that is covering only US stocks. Screener.co is supposed to have 40k global covered stocks. Value-investing.eu already has F-score calculations and is supposed to cover 22k US and International stocks. Beyond this, would anyone have anything else to recommend aside from massively expensive Bloomberg, CapIQ, and other packages? I don't have a problem checking and performing the calculations by hand (I actually planned on doing that to ensure there weren't any false net-nets), but my main goal is to get as complete data as I can. If this is between multiple packages, that's fine too. It's probably too much to ask to find a single place that does the F-score and insider holdings calculations for all global stocks.

 

(4) Compliment - What would be the best complement to a pure net-net portfolio? Perhaps funds that invest in mergers and arbitrage (or possibly following something like Fat Pitch Financials)? Or, if the money invested doesn't need to be accessed and is locked up for 3 years, is going full-on net-net acceptable? What about in plans (like 401k, 403b) where investment options are limited? Would investing in small cap value domestic and international funds (which are really nothing like net-nets when the average market cap of holdings is $1B) be a poor complement and should something else should be sought?

 

(5) Alternatives - I know the research in this area is extremely slim, but is there anything better than Geoff's idea? The underlying logic of a net-net portfolio seem sound, and even mechanical strategies where rebalancing is increased dramatically still perform decently, as shown by Jae Jun's OSV NNWC screen and backtest (18.28% p.a.). Is there are way to expand market coverage (aside from Nate's unlistedstocks.net) and is that even necessary? We are already in the CSRP 10 category in terms of market capitalization here.

 

I know this has been probably one of the longer opening posts on here, so thanks to anyway who stuck with me on it :D Really looking forward to your all's thoughts!

[west]

Yes, a very long post.

 

Imho, after collecting data from the magic formula screen for about 3 years, learning *how* to invest, and investing in a way where you *understand* what's going on, is a much more rewarding approach than the blind screener/basket approach.

 

For one, you will avoid torpedo stocks that show up occasionally in screeners that will absolutely decimate your portfolio's returns.  I've seen this happen a few times in the magic formula screener.

 

For two, you will actually understand what's going on.  When you're running a quantitative strategy blind, and you know there's some probability of getting a whole portfolio of torpedo stocks, fear sets in.  It's one thing to pooh-pooh this sentiment now.  It's another thing to live through a market downturn or two and know actual fear (and sell out lower than you should have).  When the market's going down, it's good to know that you've got a bungie cord, and to know how it works, versus not knowing if you do and having the fear of turning into a big, warm splat keep you up at night.

 

For three, although I've found that some of the magic formula portfolios reliably outperform the market, the transaction costs kill the deal, unless you're willing to allocate at least $150k to the strategy.

 

I haven't updated the data recently, but here's this if you want to play around:

 

http://www.dusthimer.net/MF_DATA/

 

The above $5000m market cap, 50 stocks selected seems to perform the best, as far as reliability of returns goes, from the data I've collected.

 

I'd still recommend actually learning how to invest tho.  It's not too tricky.  Especially if you follow the small cap/Oddball Stocks/OTC Adventures approach.

[DoddDisciple]

Thanks, West,

 

I honestly don't know if I should just delete this post and make a shorter one - maybe that would prompt more results in the long-run.

 

What I am really looking for is an investment style that fits my personality. Net-nets appeal to me since from Graham on up, people have had success with the method and it looks like it is an area institutional investors are aware of, but can't necessarily enter and invest to a level that returns are removed or diminished (versus, say, Dogs of the Dow). Net-nets' success is predicated on mean reversion and buying below conservative estimates of real-world value. Even on the question of total loss, I remember Greenbackd or Montier looked at net-nets versus the total market, and I believe 3% of the total market blew up, while 6% of net-nets did, which could definitely be insulated by buying in baskets and buying the "higher quality" ones.

 

I have done some research on Greenblatt's magic formula as well. Some people said that his use of EV/EBITDA was the main driver of returns, and the ROC measure actually reduced returns while averaging out the volatility somewhat. Overall, I don't doubt that it works, but I like the 3 year holding period with net-nets that Geoff proposes (and I don't think you could do that with magic formula stocks necessarily) since I would have a harder time following a strategy if I have to rebalance my entire portfolio once a year versus rebalancing a lot of mini portfolios as their 3 years are up.

 

I am planning to keep transaction costs at 1% of assets per basket purchase and then sit on my hands until the 3 years is up. Any estimated returns from net-net portfolios I have seen act as a ceiling on expected returns, not a floor, and I believe extended meddling from me would do nothing but detract from that.

 

I really enjoy all those blogs you mentioned :) It just seems that, the more I look into it, a purely statistical strategy similar to what Graham suggested late in his life is most beneficial for those willing to deviate outside of the norm when it comes to investing, but still remaining somewhat reticent when it comes to day-to-day action.

[ubuy2wron]

The only nthing that will ruin a well thought out mechanistic value approach will be you. Twenty names should give you a large enough sample size letting MR. Mkt do his magic for a long enough period of time without interfering is the hard part. Good luck.

[Packer16]

I think one thing you are missing with the basket approach is the ability to weight by how undervalued the security is.  This I think is a key to many posters out performance on this board.  If you equally weight a mechanical set of firms and have 20 or so you are implicitly using diversification (an I don't know strategy) when if you examined the economics of the businesses you maybe able to apply more of the I know strategy.  The other issue with some of these net-nets is control.  If the firm is controlled by a family that gets outsized benefits from ownership then the net-net may become a non net-net (by the insiders getting funneled cash) or the payout maybe delayed to the point where a large discount to net-net value may make sense.  For example if you or an outsider cannot force either a distribution of cash or assets for 5 or 10 years then the value today is diminished significantly.  I don't know how you screen for these issues unless you dive into the details of what is going on.  I think what these screen show is that out performance can happen even for an I don't know strategy (index).  However, if you apply some knowledge you should be able to do better than the I don't know index.  Just my 2 cents.

 

Packer 

[NormR]

Well, I just ran a Net Net screen and there aren't many of these beasties these days.  You've got to get down in the weeds to find them or go internationally.

 

Take care with backtests in the field because the quality of data tends to be poor on the micro-cap end of many databases.

 

I was interested to run across a study that suggested that Net Nets with negative earnings had outperformed those with positive earnings.  I'm not quite sure what to make of it.  (See prior comment.) 

 

It is an interesting area of patient investors with small amounts of capital to deploy.  But be ready for a bumpy ride.

[no_free_lunch]

Well I think I made it through most of your post.  Not too long, just thorough. 

 

There are a couple of things that would scare me about investing in the way you describe.

 

1) At a minimum I would want diversification across sectors.  I would want to make sure I don't have 85% in financial stocks if I'm investing blindly.  With this type of investing that could happen.  To get around this I would be tempted to set a maximum allocation for a sector.

 

2) I would buy an account with one of these backtesting companies and run longer and more recent tests.  My experience with mechanical investing is you have a strategy that is up 20% pa, but when you break out the results it might not have beaten the S&P since 2007.  It always make we wonder if I'm looking at a mirage, or if things have changed.  Has the widespread availability of backtesting software fundamentally changed the market such that these strategies don't work as well as they once did?  Why were the results for so many mechanical strategies so consistent, with rarely more than 2 years of underperformance, and now you have 5-6 years of underperformance? 

 

I would look for something that continues to perform well over the past 5 years.  You then should test the effect of changing parameters.  What if your pool size is 10 companies, 40 companies, 5 companies?  What if you restrict it to specific sectors?  Just keep changing the parameters and see if it continues to outperform in the majority of situations and with the most recent data.  If it does, you might have something. 

 

My personal opinion is that it might just work due to the sub $25M market cap.  I can see how with low volume stocks there might be very little appetite for this type of strategy.  Personally, it's not for me as I just like fundamental investing, I feel that if the market changes I should be able to see it based on the valuations that are offered.    Never forget that the stock market has a feedback loop in it.

[craigatk]

You describe some very thoughtful rules around how to mechanize a net-net strategy.

 

Back when I had a bloomberg and was based in Japan, I studied quite a bit more about Japanese stocks than today. In fact, I completely avoid Japan now based upon my experiences living and working there. Anyways, at the time I was selling services to hedge funds and kept particularly close eyes on the public short-lists and large holders lists, since many of our customers were in there. Always good to appear knowledgeable about their business... Over time I became convinced that Renaissance Technologies was running Graham NCAV net-net style portfolios on a global scale. Why? It's a proven strategy that has worked over a long period of time and (I believe) continues to work today. With some of the smartest quants on the street, why wouldn't they be running it??? They would show up holding some unusually small securities in sizes that couldn't have a direct meaningful contribution to their overall return based upon their AUM. Their approach uses multiple models overlayed with other models, thus one strategy might go long while another is short, creating a net position. However with some of the small Japanese names I saw them in, only explanation (to me) would be some kind of net-net strategy.

 

 

So, I believe diversification across your strategy is the key. Graham was very mechanical in his approach also, buying only on discount to value, without considering hardly anything else about the company. It was very much a basket and numbers game to him. Buying at discount and selling at fair value. At the time, he had quite a market of net-nets to select from. I don't think you change his approach much to adapt to the current US equities market, but rather expand your universe of selection to include a global equities market and you might find that you can still source enough highly discounted from value equities to fill a large enough diversified basket.

 

This is however not easy. How many small-cap net-nets in Japan do you think release English language annual reports? My guess is zero. Their financial statements are public, but it's in Japanese. This accessibility issue with global markets is probably where your biggest advantage would fall.

 

So gaining access to the right data in a format that you could quantifiably and mechanically understand is your hardest problem to solve. If you are a programmer and have enough resources behind you, it's a solveable problem in my opinion. Just like Bloomberg and other data services, you could parse Japanese financial statements into your own model... It's just not an easy problem to solve. So if you are just a small (relative to the big guys) personal account holder with more limited means, it may be something you want to leave to Renaissance and DE Shaw.

 

Hope this gives you something to think about.

[blainehodder]

  4.)  Selling - Geoff recommends holding for 3 years. This seems to be an area where almost everyone else believes to sell close to NCAV, which could make sense if you were initially selective of the stocks purchased. However, the more I think about it, a 3 year holding makes sense for a variety of reasons. For one, I have read that of all net-net, about 50% are still net-nets one year later, meaning that the typical 1 year holding period could shuffle what is already purchased to replenish the basket. By 3 years, perhaps only 12.5% of the stocks will be net-nets, and the rest will have had some activity, even if that is having gone to zero. Also, the 3 year period allows for semi-market neutral activities, like going private, buyouts, etc. to occur and unlock the value prematurely.

 

 

if you are going to be investing in net-nets on a mechanical basis, DO NOT SELL AT NCAV.  Obviously using your judgement would be best, but the 2-3 year shuffling seems appropriate, with an annual double check on the numbers.  The thing with net-net investing is that the winners really rip upward offsetting all the ones that sit there flat or decay for long periods of time.  Very often when a net-net finally starts to get juiced they go way past NCAV, which in turn leads to much of the outperformance.  By selling all of your winners at NCAV you will almost surely hurt your returns.  I searched for a post where Geoff mentioned this, but I couldn't find it.  Anyhow, a stock at NCAV is obviously still cheap, so if good things are happening, why pull the plug so early?  I have a bunch of pdfs from various sources I'll dig through to see if I can find you a source.

[vinod1]

The book quantitative value would be perfect for you if you have not already read it. It goes into quite some detail on many of the issues you are talking about. Also fool.com has a very good mechanical investing board.

 

I think the Quant "bug" is like the Value Investing "bug", you either take to it immediately or you can never get yourself to fully believe in it.

 

Vinod

[oddballstocks]

Lots of interesting thoughts in this thread.

 

I am sort of labeled as a net-net investor, I write the blog that has the same name as my handle.  I cover a lot of net-nets on the blog, but don't dedicate my portfolio to them.  I like buying things cheap, some companies at extremely low earnings, asset bargains, two pillar stocks.  I hold about 50 names, maybe 15 are net-nets.  I have another 15-20 that were net-nets at one point that I'm still holding.  As blaine mentioned it's key to know what you're buying.

 

A stock mentioned on this board a few times has been George Risk.  I purchased them back in 2010 for $4.25 a share, I believe NCAV was about $5.25, but they had a decent little business.  If I would have sold at NCAV I would have missed the run to $7 where they're at now, they're still somewhat cheap.

 

There are net-nets I would never purchase for issues mentioned in this thread.  Sometimes management has incentives to enrich themselves at the benefit of shareholders, stay away unless you can buy at a spectacular bargain.

 

I've purchased net-nets around the world.  They seem to perform the same regardless of market.  Right now I have about eight names in a Japan net-net strategy where I'm buying on a very limited criteria at 2/3 of NCAV and selling at 1x NCAV.  It's done well, when I can read the annual reports I will continue to hold above NCAV is something changes.

 

There are investor communities investing and discussing UK net-nets at ADVFN, French investors discussing "daubasses" etc.

 

This is an area of the market without as many professionals.  A little self education will go a long way.

 

As for craigatk's suggestion about Japan companies, he's right, information is hard to come by.  If you can program you have an advantage, all Japanese companies release statements in xbrl.  If you can write an xbrl parser you can parse all Japanese statements and put them into an English language DB.  Sounds complicated, but it's a lot easier than you'd think, the xml is in English as well.  If you can do this email me, I have a parser in Perl written.

 

Packer's comment about weighting is dead on, this is where gains are made.  Know the market and know it well.  I've sort of fallen into knowing the unlisted/pink market pretty well.  When I see a company drop 30-40% on no news with a sudden spike in volume I'm usually buying with two hands.  Days like this appear when a seller will just irrationally dump an illiquid stock forcing down the price.  Buyers who know their companies are able to act quickly.  Positions can become larger quickly, but big gains come from these buys.

 

Likewise always be ready to sell when the chance happens.  I know Gannon likes the 3 year holding or whatever, but be nimble.  I've had net-nets that went from 50% of NCAV to above NCAV in three months, I sell.  If my thesis is they're worth NCAV or slightly more why am I holding for an aribitrary period of time?  These are cigar butts with a last puff for a reason, when volume materializes and someone is offering to buy high sell that instant.  The opportunity might not come along again.

[Hielko]

There are net-nets I would never purchase for issues mentioned in this thread.  Sometimes management has incentives to enrich themselves at the benefit of shareholders, stay away unless you can buy at a spectacular bargain.

I don't buy these companies either, but to play devils advocate: shouldn't this be a reason why this group of companies will be underpriced and should be expected to outperform? If there is an asset that no-one likes to own I smell a potential opportunity. Mechanical net-net strategies might outperform because they don't care about this and just buy an ugly cheap stock. Or they might outperform despite this...

Likewise always be ready to sell when the chance happens.  I know Gannon likes the 3 year holding or whatever, but be nimble.  I've had net-nets that went from 50% of NCAV to above NCAV in three months, I sell.  If my thesis is they're worth NCAV or slightly more why am I holding for an aribitrary period of time?  These are cigar butts with a last puff for a reason, when volume materializes and someone is offering to buy high sell that instant.  The opportunity might not come along again.

I agree, but I don't think this mixes well with a mechanical approach: that implies that you have little knowledge on what it's intrinsically worth, and it would be my guess that selling low and holding for a fixed period is a better approach than buying low and selling at some arbitrary value like 1x NCAV because you are not only selling companies that are maybe worth 1x NCAV, but also the successful turnaround that might be worth 10x NCAV after a few years. I don't know if someone has backtested this, but I think the outperformance of mechanical net-net strategies is often the result of a few stocks that produce great returns. Selling them early could be quite bad for the end result.

[DoddDisciple]

Wow, I have to thank everyone who's posted so far! I've tried to respond to everyone below.

 

Re: ubuy2wron

 

This is my thought as well, and part of the reason why I want to figure out a gameplan I can stick with before undertaking it. I'm looking at each 20 names or so being its own portfolio, and letting positions work out, for better or worse, naturally and for what's left, closing them out after the 3 years. Conceivably, I could end up with having 100+ stocks as a complete portfolio, and there may be some overlap in each basket purchase as well.

 

Re: Packer16

 

At least when it comes to net-nets, studies have shown underperformance when value weighting versus equal weighting. Nate/Oddball had a link to a European student's thesis who discussed this. In Oppenheimer's older net-net paper, he noticed some odd things like how the firms with negative earnings as a group earned better returns than those with positive earnings.

 

For a strategy like this, I'm not looking for outperformance relative to deep value investors who actively hunt for deals at all times, but I am looking to beat the general market (S&P 500) and more small size/value focused indices in the long-term.

 

On the comment of ownership, I understand that my money can get stuck so-to-speak with suboptimal firms. However, would looking at f-score (and possibly z-score) as quality factors along with aggregate insider ownership help diminish the problems I would personally experience, along with utilizing a basket approach? If I were trying to actively force a liquidation, I can understand the complication, but as a portion of a larger hands-off portfolio, and with selling after 3 years versus holding onto the stocks for a decade or so, should I worry about this?

 

Re: NormR

 

It's funny that you mention the quantity of net-nets at the moment. One thing I was thinking long-term, though it may involve too much tinkering with a mechanical strategy, is looking at the number of net-nets as an indicator of current market under/overvaluation, and how that should influence, say, the portion of new money that should be directed to funds like Merger (MERFX) and the Arbitrage Funds, along with other special situation and workout stocks/funds. This is inspired by Graham who suggested a general 50/50 split between bonds/supercash and stocks and that the max to push any one area was 75%-80%. Sonkin had a similar structure with his Hummingbird fund, though Tarsier was just raw deep value.

 

The lack of net-nets could be one indicator that domestic markets are overvalued. Greebackd has been doing a series of different studies that all show we are due for a correction. Part of me says to just develop my plan and start now if I am going to follow a mechanical strategy.

 

Would you mind telling where you'd recommend running the screen? I don't have a problem with a subscription service, but would like as much of what I need in one place:

 

(1) global net-net screen in majority of publicly traded markets - probably can't find anything to screen for a lot of unlisted/OTC stocks aside from by hand or with something like Nate's unlistedstocks.net

(2) f-score, z-score maybe, insider ownership

(3) maybe net-net, NCAV, NNWC values - again will still verify numbers are accurate and not the result of a data misread

 

I've seen GuruFocus, Value-investing.eu, StockScreen123, Screener.co, and the Graham Investor screener as possible options so far. I worry about data quality on some, like Graham Investor, of course. 

 

I think the benefit, at least with the backtests Geoff did, is that he used StockScreen123, which uses Compustat and is supposed to be a database that keeps records for liquidations and bankruptcies, as well firms leaving for the OTC index, reducing survivorship biases. You can find a stock like Solitron in CompuStat since it was once in NASDAQ, but for something that has always been OTC, it's not included.

 

Re: no_free_lunch

 

(1) Sector diversification could make sense. However, if you look at other mechanical strategies, like Greenblatt's Magic Formula, there are concentrations in certain sectors at times, which may be a component of the strategy's performance. For-profit education showed up on a lot MF screens lately: perhaps a storm of negativity over a sector sinks all ships, and while some may never come back up, if bought in a basket, the winners should make up for the losers.

 

Also, I'm looking at net-nets with quality filters provided by F-score and insider ownership. Using the example of financial stocks, if I could buy a basket of them as net-nets and they had a quality F-score, a no-nothing approach should lead to reasonable results.

 

(2) I agree that knowledge of a formula as well as access to more ways of screening/backtesting can cause outperformance that was there to disappear. Oddly enough, I feel more comfortable with a net-net strategy vs something like the MF or the numerous other strategies out there in that it is an area where it is unlikely for institutional and other sophisticated investors to enter in a large enough degree to dissipate returns. The existence and outperformance of net-nets has long been known in the investment community, but it remains one of the simplest methods of basket investment and even with the most no-nothing strategies, such as Oppenheimer's buy all the net-nets available each year, it has shown potential returns of up to 30% p.a.

 

There was talk of John Simons' Renaissance Technologies making some investments in net-nets, which was odd for a quant hedge fund. Maybe that's a sign of experienced traders finding a way to profit from the price discrepancies of smaller stocks and future returns in the area may be impaired.

 

I don't know how valid the papers in the net-net area are, but Tobias from Greenbackd contributed to a recent net-net research paper where they ran a net-net strategy through a multifactor model, and outperformance from net-nets could not be fully explained by current factors, like size, value, momentum, etc.

 

Re: craigatk

 

Really good on-the-ground information here! I know that at lot of value bloggers have notice Renaissance Tech engaging in net-net investing as of late. I don't know what this means for the strategy in the future.

 

Japan seems to be a hot area now for people looking to apply basket net-net or extreme low p/b strategies. I have seen several value bloggers who have initiated basket purchases in the area, and research as shown that of global net-nets, at times more than 50% have been located in Japan. Jim Grant had a Japanese net-net fund that was unable to unlock value and was shuttered, but I don't know if that presents a similar warning for individual investors.

 

Expansion to a global level is ideal, but it doesn't look like small-guy screeners have caught up yet. I've read that Nate from Oddball goes through foreign markets by hand, which I am not against, I just want to ensure I am covering the largest pool once I enter those markets.

 

Would you think that the basic idea, which is ranking net-nets by insider ownership and F-score, would still make sense in global markets? Would a stronger form of net-net calculation, like the 1.00-0.75-0.50 method or 66% method, be recommended?

 

Re: blainehodder

 

Not selling at NCAV seemed cornerstone to the success of the basket method, so I would want to deviate as little as possible from that. I almost think the 2-3 year component could be a stronger driver of returns since it is a time frame that runs so contrary to other strategies as well as uncertainty tolerances. I know Geoff recommend up to 5 years, but absent at least some research in the area, I don't know if it is wise to hold onto something that long. 2-3 I can see due to the 50% of net-nets remaining net-nets the next year statistic, so only around 10% of the stocks wouldn't have had a movement by then. Do you have recommendation on holding period or a way, aside from by hand, of looking at longer than 1 year holding periods?

 

An annual check wouldn't hurt, but what would I do at that stage? I would check to ensure the stocks had passed as a net-net and had the correct valuation along with F-scoring before initiating the basket purchase. If the NCAV spread had increased, maybe it would still be included in a basket purchase (and analysis would be covered when performing that purchase). If it's above NCAV, I of course wouldn't want to fool with it.

 

One thing I considered was a floor for loss on any one stock, but even for large cap stocks the price can swing as much as 50% over the course of one year. Where do I set the floor? 70%? 80%? Or does doing the quality screen at the start make this step unnecessary since I could just chalk a company bankruptcy up as just one of those things that happens with a mechanical strategy.

 

Re: vinod1

 

I was really excited for "Quantitative Value" and actually pre-ordered it which is super weird for me. However, after reading it, I was a little let down at the information relative to what is found in the earlier posts of Tobias' Greenbackd blog. Nate went into some of the book's complications for small investors at Oddball, such as how the authors elected to focus moreso on large cap stocks.

 

Realistically, they presented several ways of computing value and reducing risk which in aggregate are too involved for a purely one-man, mechanical strategy. Near the start of the book they touched on one of Graham's strategies he presented late in his life (it was the one from his Medical Economics interview). For effort involved, the strategy performed just as well in the long-run as the model they developed over the course of the book. Of course it looked more risky since they were looking at a smaller pool of stocks, so there were times when Graham's strategy would have involved being fully invested in one stock.

 

I'll poke around Fool's Mechanical Investing board some more. I did see some sample net-net performance portfolios on there, and the one year returns were sometimes mind-boggling for a quant strategy, like 200%. Of course, I think that was for 2009. Maybe I'll make a post on there. 

 

Re: oddballstocks

 

Hi, Nate. I love your blog! You are probably one of the most prolific investment writers within a one hour radius and definitely the most generous with your ideas (weird thing to say, I know, but I'm within an hour of you and you're competing against several R1 institutions in that radius).

 

I actually wondered a little about "What would Nate think?" when writing this post and after seeing your thoughts on "Quantitative Value." I had many of the same problems you did with the book. I know you cover a lot more area than just net-nets, but for a pure know-nothing strategy, basket net-nets are one of the few areas where this could really work since there is a clear buying indicator. The selling is a little unclear, but the value spread is made at the buying, and selling is less important.

 

George Risk seems pretty popular. As far as I know, it along with Ark are the only 2 domestic stocks Geoff owns. I recall a post of yours about how you should seriously consider selling at NCAV on some stocks since they can be pretty ugly and if you can get your value out of them quickly, that's all for the better. I'm sure you still think about it on a case-by-case basis.

 

It's interesting that you are buying stocks around the world and are going very old school about it, going straight through the list of names on that country's exchange. You don't even subscribe to any screeners, do you? You just use FT.com, maybe Bloomberg's site if you have a name, and a lot of by hand research.

 

I'm not a programmer, but do work with some on online projects, so I understand what you're talking about and appreciate the parser offer. I'll definitely investigate those investment communities.

 

One interesting thing is that Geoff started the net-net newsletter on GuruFocus, but then it looks like you took it over. He had a hold for one year and then sell strategy, even though he said he would prefer a multiyear holding period. Now, you've adopted buy/sell/hold recommendations. Geoff had one article about how to put $1M to work in net-nets, which essentially meant buying the same stocks the newsletter recommended. I honestly don't know which strategy will do better long-term.

 

Re: hielko

 

Couldn't the same problem regarding enrichment happen in any company? I would think that prioritizing the net-nets and seeking those with high insider ownership would do a good job of aligning interests, even if the interest of management is to buy out the rest of the firm and take it private, possibly at a suboptimal price relative to estimates of value.

 

I agree that a value-based, instead of time-based, selling approach works better with a mechanical strategy. All I would want to know going in is that, in aggregate, most of the stocks are being purchased below a very pessimistic liquidation value at a point in time. Of course, some firms can continue to burn through cash, but in total, the goal of the businesses should be to generate shareholder wealth now and in the future.

 

Investing in net-nets in a mechanical way is almost a leap of faith, in that we see the value now has a margin of safety, but we are unsure what will happen to unlock that value. We can try to analyze what will happen, but the market always has ways of surprising us, both good and bad. Could this mean low risk, high uncertainty in the long run?

 

[DoddDisciple]

Shoot, I wrote all that and forget another question for Nate, though anyone else is welcome to chime in.

 

From your blog, it looks like you have an employer-sponsored 401k through Fidelity. Since you've said they restrict you to funds, that means you either have a list of overpriced Fidelity funds or the BrokerageLink option that allows you to invest only in funds, not individual stocks.

 

I almost feel like doing a 40/60 split between domestic and international in the plan, but the funds (196 of them) have atrocious value representation, and even worse international small offerings. FISMX I believe has the lowest average market cap for international funds at $1.4B, but it is heavily weighted to growth. For domestic, FCPVX has a little more value weighting and is at $1.6B average market cap. The fees on each of these is around 1%, which feels like highway robbery since I know ETFs that would get me the same exposure I am trying to get with these funds, but better and at lower cost. I've tried to get BrokerageLink to be added (funds only or full), but I don't think it will ever happen.

 

I'd appreciate if you or anyone else has advice, especially on what funds go well in combination with a deep value/net-net strategy.

[NormR]

It's funny that you mention the quantity of net-nets at the moment. One thing I was thinking long-term, though it may involve too much tinkering with a mechanical strategy, is looking at the number of net-nets as an indicator of current market under/overvaluation, and how that should influence, say, the portion of new money that should be directed to funds like Merger (MERFX) and the Arbitrage Funds, along with other special situation and workout stocks/funds. This is inspired by Graham who suggested a general 50/50 split between bonds/supercash and stocks and that the max to push any one area was 75%-80%. Sonkin had a similar structure with his Hummingbird fund, though Tarsier was just raw deep value.

 

The lack of net-nets could be one indicator that domestic markets are overvalued. Greebackd has been doing a series of different studies that all show we are due for a correction. Part of me says to just develop my plan and start now if I am going to follow a mechanical strategy.

 

Would you mind telling where you'd recommend running the screen? I don't have a problem with a subscription service, but would like as much of what I need in one place:

 

(1) global net-net screen in majority of publicly traded markets - probably can't find anything to screen for a lot of unlisted/OTC stocks aside from by hand or with something like Nate's unlistedstocks.net

(2) f-score, z-score maybe, insider ownership

(3) maybe net-net, NCAV, NNWC values - again will still verify numbers are accurate and not the result of a data misread

 

I've seen GuruFocus, Value-investing.eu, StockScreen123, Screener.co, and the Graham Investor screener as possible options so far. I worry about data quality on some, like Graham Investor, of course. 

 

I think the benefit, at least with the backtests Geoff did, is that he used StockScreen123, which uses Compustat and is supposed to be a database that keeps records for liquidations and bankruptcies, as well firms leaving for the OTC index, reducing survivorship biases. You can find a stock like Solitron in CompuStat since it was once in NASDAQ, but for something that has always been OTC, it's not included.

 

I used a Bloomberg terminal which is a little $$$ for most investors.  Also, even Bloomberg isn't perfect when it comes to very small stocks - particularly in Canada.

 

Do you have a study that shows F-Score works for Net Nets?  If -ve earnings helps (still trying to get my head around this one) then F-Score might hurt.

 

Oh, if you've links to Net-Net studies, please share  :)

 

Btw, I've been following Graham's Simple Way for years.  It is a high volatility method.  So high, I'm not sure many would be able to follow it.

 

Oh, and the value premium seems to persist for ~5 years at least but I've not seen a study on it for Net Nets.

 

I'd also opt for an equal weighting method (at least approximately).  Ideally, lots of small bets.

 

[oddballstocks]

It's funny that you mention the quantity of net-nets at the moment. One thing I was thinking long-term, though it may involve too much tinkering with a mechanical strategy, is looking at the number of net-nets as an indicator of current market under/overvaluation, and how that should influence, say, the portion of new money that should be directed to funds like Merger (MERFX) and the Arbitrage Funds, along with other special situation and workout stocks/funds. This is inspired by Graham who suggested a general 50/50 split between bonds/supercash and stocks and that the max to push any one area was 75%-80%. Sonkin had a similar structure with his Hummingbird fund, though Tarsier was just raw deep value.

 

The lack of net-nets could be one indicator that domestic markets are overvalued. Greebackd has been doing a series of different studies that all show we are due for a correction. Part of me says to just develop my plan and start now if I am going to follow a mechanical strategy.

 

Would you mind telling where you'd recommend running the screen? I don't have a problem with a subscription service, but would like as much of what I need in one place:

 

(1) global net-net screen in majority of publicly traded markets - probably can't find anything to screen for a lot of unlisted/OTC stocks aside from by hand or with something like Nate's unlistedstocks.net

(2) f-score, z-score maybe, insider ownership

(3) maybe net-net, NCAV, NNWC values - again will still verify numbers are accurate and not the result of a data misread

 

I've seen GuruFocus, Value-investing.eu, StockScreen123, Screener.co, and the Graham Investor screener as possible options so far. I worry about data quality on some, like Graham Investor, of course. 

 

I think the benefit, at least with the backtests Geoff did, is that he used StockScreen123, which uses Compustat and is supposed to be a database that keeps records for liquidations and bankruptcies, as well firms leaving for the OTC index, reducing survivorship biases. You can find a stock like Solitron in CompuStat since it was once in NASDAQ, but for something that has always been OTC, it's not included.

 

I used a Bloomberg terminal which is a little $$$ for most investors.  Also, even Bloomberg isn't perfect when it comes to very small stocks - particularly in Canada.

 

Do you have a study that shows F-Score works for Net Nets?  If -ve earnings helps (still trying to get my head around this one) then F-Score might hurt.

 

Oh, if you've links to Net-Net studies, please share  :)

 

Btw, I've been following Graham's Simple Way for years.  It is a high volatility method.  So high, I'm not sure many would be able to follow it.

 

Oh, and the value premium seems to persist for ~5 years at least but I've not seen a study on it for Net Nets.

 

I'd also opt for an equal weighting method (at least approximately).  Ideally, lots of small bets.

 

NormR,

 

In terms of F-Score, this was the move valuable item in Quantative Value.  I seem to remember they had a study in there where if you took all the stocks below 1x book value and divided them in half based on F-Score, shorted the low F-score stocks, and purchased the high F-Score stocks that strategy would outperform the market by 22% a year, that number is incredible.

 

The problem is having the capital to do such a strategy, I don't know how much it would take, but it would certainly seem worth it.

 

I have the Oppenheimer study, as well as a more recent study (from 1974-2008) on net-nets, shoot me an email at oddballstocks at gmail dot com and I'll send it your way.  The study is fascinating in that it shows a solid outperformance, the problem is at times the portfolio is small.  The paper studied buying strickly at 2/3 NCAV and selling at 1x NCAV.  The outperformance was giant, but in the mid-1980s a portfolio would consist of one or two stocks.

 

 

[NormR]

In terms of F-Score, this was the move valuable item in Quantative Value.  I seem to remember they had a study in there where if you took all the stocks below 1x book value and divided them in half based on F-Score, shorted the low F-score stocks, and purchased the high F-Score stocks that strategy would outperform the market by 22% a year, that number is incredible.

 

Ya, but wasn't that in mid-large stocks?  The -ve earnings finding threw me for a loop and has me questioning other factors when it comes to Net Nets. 

 

Btw, the March/April FAJ has an interesting article on M-Scores (fraud checks) which also seems to help quite a bit.

 

 

[txitxo]

My recommendations:

 

1. As somebody has mentioned before, net-nets are a good example of a "rockets" screen. Some companies will tank, most will stay put, and all the returns come from a few stocks. The problem is, they often are the ugliest girls in the ballroom, and no self-respecting classical value investor would get near them. So by all means make sure you avoid funny accounting and robber managers, but if you are investing with screens, you'll have to eat the whole animal from nose to tail. That pretty much applies to any mechanical strategy.

 

2. Therefore you have to spread yourself thin, avoid worrying too much about individual companies, and let the law of large numbers do its work. You'll have to use more than one screen, net-nets will not provide you with enough companies to do that except at market bottoms.

 

3. Create your own screens, by introducing small variations on established ones. You don't want to be competing with Renaissance Technologies, but with other individuals who are not as rational as you are.

 

4. The Quantitative Value book does a good job at presenting the backbone of any good screen: financial soundness and cheapness. Avoid suspicious companies and buy a few dozen stocks, spred across the year (I do one every week) in the bottom 10-20% by EV/EBITDA,  P/B, or whichever cheapness indicator you like. Keep them for >1 year and you'll beat the index. Tinker with that procedure, and you almost certainly won't.

 

5. Adding momentum is a bit more controversial. It seems to increase returns, but it may amplify volatility (e.g. see the drop in the Tiny Titans screen during 2008-2009).

 

6. Be realistic about slippage, commissions and taxes. They wil eat you alive if you are not careful. Trade as little as possible, and watch out for those stocks who are truly illiquid.

 

7. Mechanical screens will always tend to be more volatile than the market, after all you are investing in a small sample of  micro caps, so to protect my stomach lining I complement them with some timing indicators to avoid statistically  "dangerous" markets (like the US currently). 

 

8. Spend a few hours during the weekend selecting the stock you want to buy and then do something fun during the rest of the week, write poetry, read the Classics, or try to convince governmental agencies that it is essential to understand Dark Energy...

 

[Packer16]

In regards to using an F-score that may help but my primary issue is the issue of hostile management for which the F-score will not help.  A good unlevered cheap business run by an abusive management team would get through your screen but you may have to wait 5, 10 or 20 years to get a reasonable return if management is funneling money to themselves.  This is a not uncommon situation in some of these net-net situation.  That is part of the reason they are a net-net.  I think using the F-score as a screen for stock to further study makes more sense as you can then screen out the hostile management ones. 

 

An interesting article you may want to read is Jeremy Grantham's memo on the potential disadvantage of Graham and Dodd-type investing Part I and II (April 2012 and Jan 2011).  He describes that the value excess return is due to the twice in a century event (think Great Depression) when a number of these firms will go out of business.  The premium appears to be greater after these events and eventually as this strategy gains more followers leads to average performance versus larger more quality names.  It something interesting to think about as the only way to out perform is to do something someone else cannot or will not do.  I am not sure where we are in the net-net cycle but given the lack of a large number of these available today a good portion of the excess return may already be realized.

 

Packer

[rijk]

very interesting thread, i have tried to monitor actual performance of several net net portfolios that are currently running, the ones i am aware of are

-gurufocus ncav bargains - inception mar 2011 - hand picked

-cheap stocks 26 net net index - inception sep 2011 - mechanical

-canadian net net portfolio - jan 2013 - mechanical?

http://www.theglobeandmail.com/globe-investor/net-net-capital-portfolio-not-for-faint-of-heart/article7016571/

 

when reading about net net performance, it seems as if most studies and articles conclude that a net net strategy will handily outperform the market and that a net net portfolio should return around 20%/year on average

 

however, when looking at the actual performance of the three portfolios, it seems that at least for these three examples, actual performance do not beat the market and are far below 20% average

 

i monitor all three portfolios in google finance and just eye balling the performances using equal weights, it looks like all three portfolios have a negative or close to flat performances since inception (since most of the "owners" of these portfolios are on the board, please feel free to correct me)

i noticed that after one year, the 26 net net index portfolio significantly outperformed (36%) the gurufocus portfolio (flat), but when diving into the details, nearly the complete outperformance of the 26 net net index portfolio was due to one net net that had been incorporated into the portfolio at a time when this net net was being taken over by another company, and the outperformance was achieved after the stock merger, so at the time this company was incorporated into the portfolio, you were really buying the acquiring company which was not a net net, so i wonder how many investors would have bought and kept that specific outperformer

 

another example i can share is my own personal experience with jnets (japanese net nets), japan seems to be the ideal situation for a live net net experiment, so i started two baskets of hand picked jnets in jan 2012 and 2013, the actual performance is 20% (local currency) after nearly 18 months, this looks reasonable but is again below the 20% annual benchmark and far below the 62% nikkei performance

 

regards

rijk

 

 

[NormR]

very interesting thread, i have tried to monitor actual performance of several net net portfolios that are currently running, the ones i am aware of are

-gurufocus ncav bargains - inception mar 2011 - hand picked

-cheap stocks 26 net net index - inception sep 2011 - mechanical

-canadian net net portfolio - jan 2013 - mechanical?

http://www.theglobeandmail.com/globe-investor/net-net-capital-portfolio-not-for-faint-of-heart/article7016571/

 

Far too short a period to come to conclusions regarding the method, IMHO.  Most good methods suffer from long periods of underperformance - often many years in length.

[Packer16]

Another example of underperformance was Jim Rea's mutual fund based upon Ben Graham's net-net approach.  Part of the issue may be timing (when the market rewrards these bargains versus when they go broke or remain zombies). 

 

Packer

[NormR]

Another example of underperformance was Jim Rea's mutual fund based upon Ben Graham's net-net approach.  Part of the issue may be timing (when the market rewrards these bargains versus when they go broke or remain zombies).

 

I thought this was based more on an extension of his defensive method?

[Packer16]

You are correct. It was based upon his 10 criteria from the II so it is not directly applicable but does illustrate the effectiveness (or lack thereof) of using a raw screen to manage money.

 

Packer

[NormR]

You are correct. It was based upon his 10 criteria from the II so it is not directly applicable but does illustrate the effectiveness (or lack thereof) of using a raw screen to manage money.

 

I'm more sympathetic to screens/quants than most on this board.  But I acknowledge that they aren't without their issues. 

 

The O'Shaugnessy US Value fund seems to have had some success using a relative value/momentum method. 

 

I think the upside for most such methods is a few percentage points above the market.  You're not likely to get Buffett like numbers from them, but they're a good place for beginning value investors to start.  (Mind you, Net Nets are probably best pursued by more knowledgeable people.)