Jump to content

Magic Formula Data

west

By west
in General Discussion

 Share

Recommended Posts

NormR Newbie

NormR

Posted
Posted

any data on the maximum drawdown for the magic formula strategy? Can't seem to find any data on the standard deviation..

 

For larger stocks, 1974-2011 Quantitative Value pegs SD at 16.9%, worst drawdown -37%.

 

 

Link to comment
Share on other sites
  • Replies 92
  • Created
  • Last Reply

Top Posters In This Topic

Popular Days

Top Posters In This Topic

Popular Days

netnet Newbie

netnet

Posted
Posted

 

This is 3 year data, hardly statistically valid. In fact given the tumult, I would expect something like this during this period.  That said, to me MF is a beginning, it gives companies to analyze.

Link to comment
Share on other sites
luck Newbie

luck

Posted
Posted

i decided to give fnsax a spin awhile back.  i took profits in part due to the 1.35% expense fee.  in general, i also haven't been too impressed with mutual funds outside of those run by bruce berkowitz - fairx and faafx. 

 

i agree with twa, i think best to cherry pick a few fundamentally sound companies off the list for further review or use it as another confirmation or data point in overall analysis.

Link to comment
Share on other sites
oddballstocks Newbie

oddballstocks

Posted
Posted

 

This is 3 year data, hardly statistically valid. In fact given the tumult, I would expect something like this during this period.  That said, to me MF is a beginning, it gives companies to analyze.

 

Whenever I hear this response I wonder how long is needed for something to be statistically significant?  I've heard people claim 120 years of market data isn't enough to show stocks outperform bonds at a statistically significant level.

 

Link to comment
Share on other sites
blainehodder Newbie

blainehodder

Posted
Posted

 

This is 3 year data, hardly statistically valid. In fact given the tumult, I would expect something like this during this period.  That said, to me MF is a beginning, it gives companies to analyze.

 

Whenever I hear this response I wonder how long is needed for something to be statistically significant?  I've heard people claim 120 years of market data isn't enough to show stocks outperform bonds at a statistically significant level.

 

 

I agree Nate, but in the case of a long term quantitative investing strategy,  wouldn't it be best to look at the performance over a full market cycle at least?  If we only look at the run up since 09, you could conclude that buying "high beta" glamour stocks is a great strategy, netnets are horrible, etc.

 

Given the rationale behind the MF, I believe it will outperform a comparable index over a full market cycle as backtesting seems to indicate, even if real world performance falls short of the backtests .  Almost anything with a cheapness metric in its ranking should outperform over time I think given you buy enough stocks for general diversification.  I don't see how adding a quality criteria to that could be a bad thing.  Personally, I do believe the MF is one of the best strategies out there if you aren't a stock picker.  Adding a z-score, and/or f-score to the ranking might help avoid some real traps as well.

 

If you are a stock picker (I am) I think you have to ask yourself if the marginal time spent is worth any out performance you might achieve over a strategy like this (instead of just the index)...  or maybe you just enjoy it!  I do :)

 

I still can't believe no super value ETFs exist... straight low fee (sub 1%) 30 stock MF ETF, every netnet under 66% NCAV with acceptable z-score ETF over 25 mill, all P/B below 1 f-score/z-score over a hurdle (not Chinese) ETF etc.  I would consider putting the majority of my money in those if they existed, and only pick a couple stocks outside of that.  Maybe it isn't feasible for the really small type strategies above due to liquidity, but certainly a cheapest decile ev/ebit, or p/b S&P type index could exist.

 

Link to comment
Share on other sites
Kraven Newbie

Kraven

Posted
Posted

 

This is 3 year data, hardly statistically valid. In fact given the tumult, I would expect something like this during this period.  That said, to me MF is a beginning, it gives companies to analyze.

 

Whenever I hear this response I wonder how long is needed for something to be statistically significant?  I've heard people claim 120 years of market data isn't enough to show stocks outperform bonds at a statistically significant level.

 

Usually it's the amount the person saying it's statistically insignificant feels it would need to be to make them personally comfortable.  On the other hand, if refuting someone's point any amount is statistically insignificant or questionable. 

Link to comment
Share on other sites
NormR Newbie

NormR

Posted
Posted
I still can't believe no super value ETFs exist... straight low fee (sub 1%) 30 stock MF ETF, every netnet under 66% NCAV with acceptable z-score ETF over 25 mill, all P/B below 1 f-score/z-score over a hurdle (not Chinese) ETF etc.  I would consider putting the majority of my money in those if they existed, and only pick a couple stocks outside of that.  Maybe it isn't feasible for the really small type strategies above due to liquidity, but certainly a cheapest decile ev/ebit, or p/b S&P type index could exist.

 

Microcap ETFs would be a bit nutty due to liquidity issues.  But there are several value ETFs to pick from with P/B being a popular factor.     

Link to comment
Share on other sites
blainehodder Newbie

blainehodder

Posted
Posted

I still can't believe no super value ETFs exist... straight low fee (sub 1%) 30 stock MF ETF, every netnet under 66% NCAV with acceptable z-score ETF over 25 mill, all P/B below 1 f-score/z-score over a hurdle (not Chinese) ETF etc.  I would consider putting the majority of my money in those if they existed, and only pick a couple stocks outside of that.  Maybe it isn't feasible for the really small type strategies above due to liquidity, but certainly a cheapest decile ev/ebit, or p/b S&P type index could exist.

 

Microcap ETFs would be a bit nutty due to liquidity issues.  But there are several value ETFs to pick from with P/B being a popular factor.   

 

Agreed.  I guess I just haven't found any of these ETFs that I really consider cheap enough for my Grahamsian taste.  A quick look at the holdings of some of these ETFs and I see p/b stocks above 3, P/e averaging above 18.  Not exactly a cheap enough stocks for outperformace in my opinion.

 

Do you know any true value ETFs?  Most of the ones I find would be laughable to Graham.

Link to comment
Share on other sites
NormR Newbie

NormR

Posted
Posted

I still can't believe no super value ETFs exist... straight low fee (sub 1%) 30 stock MF ETF, every netnet under 66% NCAV with acceptable z-score ETF over 25 mill, all P/B below 1 f-score/z-score over a hurdle (not Chinese) ETF etc.  I would consider putting the majority of my money in those if they existed, and only pick a couple stocks outside of that.  Maybe it isn't feasible for the really small type strategies above due to liquidity, but certainly a cheapest decile ev/ebit, or p/b S&P type index could exist.

 

Microcap ETFs would be a bit nutty due to liquidity issues.  But there are several value ETFs to pick from with P/B being a popular factor.   

 

Agreed.  I guess I just haven't found any of these ETFs that I really consider cheap enough for my Grahamsian taste.  A quick look at the holdings of some of these ETFs and I see p/b stocks above 3, P/e averaging above 18.  Not exactly a cheap enough stocks for outperformace in my opinion.

 

Do you know any true value ETFs?  Most of the ones I find would be laughable to Graham.

 

Doing another survey is on my list.  The "pure value"series of U.S. ETFs (RPV, etc) seem to be better than most. 

Link to comment
Share on other sites
vinod1 Newbie

vinod1

Posted
Posted

A member on the Fool message board who has done some research on MF. Some very interesting results.

 

http://boards.fool.com/brk-shareholders-mtg-30027775.aspx?sort=whole#30027775

 

 

Note, reputable studies (including my own) generally show market beating

performance, but the advantage is a fraction of what is claimed.

A typical test of mine: buy the 6 highest-ranked stocks each 3 months

and hold for a year, a portfolio of 24 stocks--about what he recommends.

Total return 1989-2011 14.7% versus 9.3% for the S&P, advantage 5.4% without trading costs.

This particular test limits itself to the largest 1000-1500 US stocks

meeting [best guess of] his industry filters, so the big outperformance

if any must lie in very small stocks. Other tests including the very

small stocks also found only small advantages.

 

What can I say?

A large number of people have tested this.

Only one, Mr Greenblatt, got really high returns from it.

The exact reason isn't very important--it's not a foolproof money spinner.

 

Having spent over a decade examining tens of thousands of quantitative

investing methods, I have never seen a single plausible scheme that

showed 30% returns with annual holds while long equities all the time

even in backtest, let alone in real life. Maybe a few systems with

tortuously complex over-fix filter criteria, but I can't even remember one of those.

I don't imagine even Jim Simons could manage it, and he's the Gretzky of quants.

Link to comment
Share on other sites
oddballstocks Newbie

oddballstocks

Posted
Posted

 

This is 3 year data, hardly statistically valid. In fact given the tumult, I would expect something like this during this period.  That said, to me MF is a beginning, it gives companies to analyze.

 

Whenever I hear this response I wonder how long is needed for something to be statistically significant?  I've heard people claim 120 years of market data isn't enough to show stocks outperform bonds at a statistically significant level.

 

 

I agree Nate, but in the case of a long term quantitative investing strategy,  wouldn't it be best to look at the performance over a full market cycle at least?  If we only look at the run up since 09, you could conclude that buying "high beta" glamour stocks is a great strategy, netnets are horrible, etc.

 

Given the rationale behind the MF, I believe it will outperform a comparable index over a full market cycle as backtesting seems to indicate, even if real world performance falls short of the backtests .  Almost anything with a cheapness metric in its ranking should outperform over time I think given you buy enough stocks for general diversification.  I don't see how adding a quality criteria to that could be a bad thing.  Personally, I do believe the MF is one of the best strategies out there if you aren't a stock picker.  Adding a z-score, and/or f-score to the ranking might help avoid some real traps as well.

 

If you are a stock picker (I am) I think you have to ask yourself if the marginal time spent is worth any out performance you might achieve over a strategy like this (instead of just the index)...  or maybe you just enjoy it!  I do :)

 

I still can't believe no super value ETFs exist... straight low fee (sub 1%) 30 stock MF ETF, every netnet under 66% NCAV with acceptable z-score ETF over 25 mill, all P/B below 1 f-score/z-score over a hurdle (not Chinese) ETF etc.  I would consider putting the majority of my money in those if they existed, and only pick a couple stocks outside of that.  Maybe it isn't feasible for the really small type strategies above due to liquidity, but certainly a cheapest decile ev/ebit, or p/b S&P type index could exist.

 

Thanks for the response, I didn't realize the results were just the mutual funds since 2009.  I've been following the strategy through the Yahoo! Group since 2007, people have had real money invested in this since the 2006 timeframe and their results mirror what others have said on the forum, it outperforms but not by much.  The participants were slaughtered in the 2008/09 downturn, most couldn't stick with the losses.  Even in the recovery the stocks didn't do well.

 

What always troubled me was that out of the 20 or so results given an investor had to pick randomly a few stocks each quarter.  It seemed reading the newsgroup that investors had a tendency to pick the losers even if their picking was random.  A few had devised systems that improved upon the formula by including simple non-intuitive checks.  A lot on the group were (still are) using technical analysis to try to avoid the falling knives.  I know many took losses on the Chinese stocks as well.  Once that sector tanked they mysteriously disappeared from the lists, small consolidation to the investors who lost real money with them.

 

I think the difficulty with a deep value ETF is the liquidity.  Investing in a lot of those types of stocks is feasible as an individual with smaller amounts of capital, but I can't imagine trying to move $100m through the sector.  How do you possibly buy stakes in companies like Northfield Precision (a deep value net-net with a $3m mcap)?  Especially when shares trade by appointment.

 

I have thought about the feasibility of opening simple value funds, buying a bunch of net-nets at 2/3 NCAV or something.  The problem is it's hard to raise money for a fund like that, what investor wants to buy into 100s of Japanese stocks?  Or lots of tiny microcap stocks that barely trade?  There's a reason so many funds have the term "growth" in them, and so many value funds are mid/large cap GARP funds.  It's much easier to buy growth cheap than to actually buy value stocks in size.

Link to comment
Share on other sites
west Newbie

west

Posted
  • Author
Posted

I'm gonna try to get Quantiatve Value from the library. I've heard some great thing about it.

 

Doesn't Greenblat say  that the MF never trailed over a 3 year period (or perhaps it was 5)?

 

From his 11-03-06 Columbia lecture (courtesy of John Chew) the longest period of underperformance vs. the market was 46 months.  The second longest period was 39 months.  He starts talking about it an hour and fourty minutes in if you have the video.  If not, you may want to meander over to csinvesting.org :).

 

I'm still not convinced on using the strategy blind at this point myself.  And I've got about two and a half years worth of data now...

Link to comment
Share on other sites
twacowfca Newbie

twacowfca

Posted
Posted

The problem is that screens that identify extreme values pick up a lot of turkeys or frauds that may not be evident at first glance.  MF as a screen worked to some degree until the slew of Chinese frauds destroyed the results.  What will work using the screen is waiting for the fat pitch: a company you know has honest management and a good business with problems that are over discounted.  In other words : sound investing.  :)

Link to comment
Share on other sites
mvalue Newbie

mvalue

Posted
Posted

I'm gonna try to get Quantiatve Value from the library. I've heard some great thing about it.

 

Doesn't Greenblat say  that the MF never trailed over a 3 year period (or perhaps it was 5)?

 

From his 11-03-06 Columbia lecture (courtesy of John Chew) the longest period of underperformance vs. the market was 46 months.  The second longest period was 39 months.  He starts talking about it an hour and fourty minutes in if you have the video.  If not, you may want to meander over to csinvesting.org :).

 

I'm still not convinced on using the strategy blind at this point myself.  And I've got about two and a half years worth of data now...

 

http://www.formulainvesting.com/Results/US_Strategy/ shows it running behind after 44 months as of Dec 31, and likely almost 47 based on what I've seen of the portfolios YTD that isn't in that data (this strategy is much closer to the Little Book than the mutual funds). So looks like it's going to surpass that model losing period...

Link to comment
Share on other sites
blainehodder Newbie

blainehodder

Posted
Posted

 

I think the difficulty with a deep value ETF is the liquidity.  Investing in a lot of those types of stocks is feasible as an individual with smaller amounts of capital, but I can't imagine trying to move $100m through the sector.  How do you possibly buy stakes in companies like Northfield Precision (a deep value net-net with a $3m mcap)?  Especially when shares trade by appointment.

 

I have thought about the feasibility of opening simple value funds, buying a bunch of net-nets at 2/3 NCAV or something.  The problem is it's hard to raise money for a fund like that, what investor wants to buy into 100s of Japanese stocks?  Or lots of tiny microcap stocks that barely trade?  There's a reason so many funds have the term "growth" in them, and so many value funds are mid/large cap GARP funds.  It's much easier to buy growth cheap than to actually buy value stocks in size.

 

...If you could open such a fund with fees below 1% let us know!  I'd be happy to invest!

 

But you are right.  Let's say you would have to make 100K/year in profit at minimum to make it worth your time... then you would need probably $20M AUM assuming 100K expenses and a 1% fee.  At this point you may have too much cash to deploy into the strategies unless you had multiple funds with different strategies.  global microcap MF, global net-net...and more.  I guess it just might not be feasible (yet)?  More likely, you'd need 2 people running the funds, contract out your accounting and legal expenses etc.  It starts to look daunting without a performance fee, but then that defeats the purpose.  Most people who would invest in a microcap netnet fund "basket" could do it without paying the expenses associated with a performance fee.  Just hit Graham investor, or a Bloomberg or whatever and buy away. Then if you start actively managing it, if you are charging the currently insane management fees of hedge funds, all the outperformance would just go in your jeans, as opposed to customers.

 

Still, not all cheapo startegies that work are in microcaps, and I haven't found many ETFs that are truly cheapo strategies in my opinion.

 

 

 

On a different note, one of the best sources of ideas for me comes from ranking microcaps by EV/EBIT and ROIC on the Bloomberg. If MF no longer works well on huge stocks, it certainly still seems to identify winners like JCTCF in the 20-50 mill category.

Link to comment
Share on other sites
stahleyp Newbie

stahleyp

Posted
Posted

I'm gonna try to get Quantiatve Value from the library. I've heard some great thing about it.

 

Doesn't Greenblat say  that the MF never trailed over a 3 year period (or perhaps it was 5)?

 

From his 11-03-06 Columbia lecture (courtesy of John Chew) the longest period of underperformance vs. the market was 46 months.  The second longest period was 39 months.  He starts talking about it an hour and fourty minutes in if you have the video.  If not, you may want to meander over to csinvesting.org :).

 

I'm still not convinced on using the strategy blind at this point myself.  And I've got about two and a half years worth of data now...

 

thanks, west. I'm gonna look into that.

Link to comment
Share on other sites
netnet Newbie

netnet

Posted
Posted

 

This is 3 year data, hardly statistically valid. In fact given the tumult, I would expect something like this during this period.  That said, to me MF is a beginning, it gives companies to analyze.

 

Whenever I hear this response I wonder how long is needed for something to be statistically significant?  I've heard people claim 120 years of market data isn't enough to show stocks outperform bonds at a statistically significant level.

 

At the minimum, you need to go a whole market cycle.  Also, I would guess, that the Chinese fraud problem really screwed up results.  No experienced investor should have bought into that crap, though. Again, I look at MF  as a starting point.

  • If you scrub for the Chinese frauds, and the results were good, one could still argue that that was cherry picking.
  • You would expect value stocks not to do as well on a big leg up in the market.

Link to comment
Share on other sites
oddballstocks Newbie

oddballstocks

Posted
Posted

 

I think the difficulty with a deep value ETF is the liquidity.  Investing in a lot of those types of stocks is feasible as an individual with smaller amounts of capital, but I can't imagine trying to move $100m through the sector.  How do you possibly buy stakes in companies like Northfield Precision (a deep value net-net with a $3m mcap)?  Especially when shares trade by appointment.

 

I have thought about the feasibility of opening simple value funds, buying a bunch of net-nets at 2/3 NCAV or something.  The problem is it's hard to raise money for a fund like that, what investor wants to buy into 100s of Japanese stocks?  Or lots of tiny microcap stocks that barely trade?  There's a reason so many funds have the term "growth" in them, and so many value funds are mid/large cap GARP funds.  It's much easier to buy growth cheap than to actually buy value stocks in size.

 

...If you could open such a fund with fees below 1% let us know!  I'd be happy to invest!

 

But you are right.  Let's say you would have to make 100K/year in profit at minimum to make it worth your time... then you would need probably $20M AUM assuming 100K expenses and a 1% fee.  At this point you may have too much cash to deploy into the strategies unless you had multiple funds with different strategies.  global microcap MF, global net-net...and more.  I guess it just might not be feasible (yet)?  More likely, you'd need 2 people running the funds, contract out your accounting and legal expenses etc.  It starts to look daunting without a performance fee, but then that defeats the purpose.  Most people who would invest in a microcap netnet fund "basket" could do it without paying the expenses associated with a performance fee.  Just hit Graham investor, or a Bloomberg or whatever and buy away. Then if you start actively managing it, if you are charging the currently insane management fees of hedge funds, all the outperformance would just go in your jeans, as opposed to customers.

 

Still, not all cheapo startegies that work are in microcaps, and I haven't found many ETFs that are truly cheapo strategies in my opinion.

 

 

 

On a different note, one of the best sources of ideas for me comes from ranking microcaps by EV/EBIT and ROIC on the Bloomberg. If MF no longer works well on huge stocks, it certainly still seems to identify winners like JCTCF in the 20-50 mill category.

 

You basically described why I haven't done this yet, the numbers just don't work well.  To make a reasonable salary after expenses requires an AUM that's higher than could be reasonably deployed in many of these stocks.

 

I've come to realize if one wants to do well on an individual level small and microcap stocks offer the best opportunity.  If one wants to do well as an asset manager they need to slightly outperform the market and raise capital like crazy investing in mid-cap or large-cap stocks.

Link to comment
Share on other sites
Hielko Newbie

Hielko

Posted
Posted

 

This is 3 year data, hardly statistically valid. In fact given the tumult, I would expect something like this during this period.  That said, to me MF is a beginning, it gives companies to analyze.

 

Whenever I hear this response I wonder how long is needed for something to be statistically significant?  I've heard people claim 120 years of market data isn't enough to show stocks outperform bonds at a statistically significant level.

 

At the minimum, you need to go a whole market cycle.  Also, I would guess, that the Chinese fraud problem really screwed up results.  No experienced investor should have bought into that crap, though. Again, I look at MF  as a starting point.

  • If you scrub for the Chinese frauds, and the results were good, one could still argue that that was cherry picking.
  • You would expect value stocks not to do as well on a big leg up in the market.

I think that's certainly a form of cherry picking. Magic formula is supposed to be great because it's buying ugly and cheap companies that humans wouldn't buy because "no experienced investor should have bought into that crap". If it would have bought a few real Chinese companies that went op 1000% no-one would have complained about the bad picks, in that case it would all be part of the strategy.

Link to comment
Share on other sites
  • 5 months later...
ASTA Newbie

ASTA

Posted
Posted

I was just thinking about this today and the magic-formula has worked really good on big companies lately except HPQ(I now from experience :D).

My secrete sauce to the Magic formula is a second screen of guru ownership and to be safe bigger companies.

I might be data-mining but this has worked quite good.

One can go one steep further and reduce the stocks selected by only going by stocks mentioned here and the list is right more then 50% of the time at least of a winner :D

Example

Dell,msft,csco,hlf,aapl, and bby and so on.

 

So if one finds a consensus of COBF, magic-formula and Guru ownership(10% or less above 52 week low) the likely hood of failure is greatly reduced in my opinion just my thoughts.

Of curse I might just be totally wrong just typing some thoughts :D

 

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
 Share
Go to topic listing


×
×
  • Create New...