The Essays of Racemize

[John Hjorth]

During the years, our active fellow CoBF board member Racemize [Joel Stevens] has been so kind to share his essays here on the board on a continuing basis, based on some Dropbox links by Joel.

 

Personally, I have read them all - and, actually, I hold them high.

 

I suppose we all have our favorites and "anchors" with regard to reading. Personally, I define an "anchor" as a "piece" [book, essay, article, whatever] that I go back to, to set myself self straight, to get back on the track with my line of thinking, when my thoughts starts straying, and I end up more or less confused and unable to think straight.

 

To succeed, you first have to finish. You have to have a perception of where you are, and where you are headed going forward, to cross the finish line with a good result - the finish line, that is always moving away from you in the horizon relative to your own position - just like a rainbow.

 

To navigate, you need to have perception of where you are, otherwise you "grab the anchor". It's like turning on the GPS in your car to get it to find your actual location before you tell it where you are headed, after which the GPS is guiding you, if you are on a route not yet known and familiar, or if you have got lost somewhere.

 

Recently, I needed to go back to one of Joels essays, and I found out that I have not downloaded it from here [Dropbox link by Joel]. I managed to find it anyway on the Net. After the worst panic evaporated after getting that particular essay anyway, I did send send a CoBF PM to Joel to express my appreciation for sharing the work.

 

It turned out, that Joel actually really appreciated the positive feedback from me, because Joel actually does not get much feedback on his writings. I also mentioned the dead Dropbox links for Joel.  Joel wrote to me, that his Dropbox account got banned, because some [dumb] Twitter user linked to the Dropbox link [on Twitter] - so much thanks for sharing!

 

Anyway, for all my fellow board members ex. Joel, here is a link to Joels essays, now that the Dropbox links on this board are dead.

 

- - - o 0 o - - -

 

To Joel:

 

A sincere thank you for sharing you work!

 

To all other board members than Joel:

 

Enjoy! - If you do not already have them and have read them. Furthermore, perhaps it would not harm to post a couple of lines in this topic, if you like them ... -Perhaps we then get access to more work by Joel going forward!

 

 

- - - o 0 o - - -

 

 

Disclosure: No skin in the game at the fund Joel is running - I just happen to like and appreciate Joels essays.

[Liberty]

Thanks John, and thanks Joel. I've also appreciated your writings I've the years.

[Jurgis]

I read through "Know Your Graph" piece.

 

I agree with the arguments there that the manager/investor should not be evaluated just based on their results.

In fact, I mostly agree that evaluation based on graph would be better.

 

I disagree that the graph is easy or even possible to create in most cases. In fact, even for a single investment - single stock - I have not seen anyone draw a graph distribution for expected outcomes. (OK, OK, I think Ed Thorp could have done the graph for some of his arbitrage investments. This is still an exception).

 

I'd like to see someone present an investment with a graph of outcomes and argue for the distribution used.

Or if someone is up for harder challenge, take any public portfolio (maybe something simpler than BRK... maybe top 5 positions of some investment manager) and draw a graph.

IMHO, this is extremely hard and likely not possible at any useful accuracy.

But maybe I'm too negative and someone can prove me wrong. ;)

 

Thanks for the article.  8)

[LC]

my favorite was the piece on whether to hold cash or not.

[Guest longinvestor]

Thanks to Joel for the content; John for providing the link. I have also learned from prior posts by Racemize, notably the one about staying fully invested.

[Cigarbutt]

I'm fairly new but somehow got into the article on the relevance of remaining fully invested.

This is work in progress for me.

Racemize, your article has made it to my reference base for this topic. Thank you.

I will look at the rest.

[enoch01]

my favorite was the piece on whether to hold cash or not.

 

I confess to never really understanding it.  Here's the conclusion:

 

I again emphasize that this essay is not focused on remaining fully invested at all times, but instead advocates remaining fully invested when there are still compelling investments to be made."

(my bolding).

 

Isn't it tautological that one should prefer "compelling" investments over cash?  That's just what a compelling investment means.

[Jurgis]

my favorite was the piece on whether to hold cash or not.

 

I confess to never really understanding it.  Here's the conclusion:

 

I again emphasize that this essay is not focused on remaining fully invested at all times, but instead advocates remaining fully invested when there are still compelling investments to be made."

(my bolding).

 

Isn't it tautological that one should prefer "compelling" investments over cash?  That's just what a compelling investment means.

 

It's not completely tautological. Or at least not tautological for everyone.  8)

 

E.g. I've held about 20% cash since 2009. Clearly through all this time there were compelling investments available. So I should have not held it, yes?

 

 

 

OTOH if we consider that 20% a portion of portfolio dedicated to "fixed income" then perhaps I should have held it. Though perhaps I should have put it into better fixed income funds than "cash". (Which I partially did, but that's another story).

[Jurgis]

I read through "Know Your Graph" piece.

 

I agree with the arguments there that the manager/investor should not be evaluated just based on their results.

In fact, I mostly agree that evaluation based on graph would be better.

 

I disagree that the graph is easy or even possible to create in most cases. In fact, even for a single investment - single stock - I have not seen anyone draw a graph distribution for expected outcomes. (OK, OK, I think Ed Thorp could have done the graph for some of his arbitrage investments. This is still an exception).

 

I'd like to see someone present an investment with a graph of outcomes and argue for the distribution used.

Or if someone is up for harder challenge, take any public portfolio (maybe something simpler than BRK... maybe top 5 positions of some investment manager) and draw a graph.

IMHO, this is extremely hard and likely not possible at any useful accuracy.

But maybe I'm too negative and someone can prove me wrong. ;)

 

Thanks for the article.  8)

 

I could wait more, but I think I'm gonna conclude that nobody gives a f**k about the investment outcome graphs. Which I expected.  8)

 

OK, to raise the bar a bit, I'm gonna present an 5 year annualized return probability distribution graph for an easy stock.

 

 

 

BRK.

 

 

 

8)

 

I can't draw graphs for shit, so here's a description (someone with some math graphing package can probably draw it from description):

 

It's not symmetrical graph. Maximum probability is at 9%. Pretty symmetrical normal distribution decline to 6.5% and 11.5%. This covers about 80% of the distribution area. Rapid decline towards zero probability as we go over 11.5%. Really close to zero above 15%. Slower decline towards 0%. A shallow hump up with a max around 1% that extends both up and down towards -5% (minus here). This mostly represents Buffett dying and/or mega cat(s). The area covers about 10% or so total probability area. Probability goes to near zero at around -9%.

 

So there.

 

Probably a good illustration why this is fricking hard.

 

Have fun.  8)

[racemize]

Re: Know Your Graph

 

I don't think it is terribly hard to create the general shape of the graph or figure out where the modes are for individual investments.  Your BRK example is one that a lot would generally agree with, I think, but obviously the details would vary from person to person (e.g., Buffett did say he thought intrinsic value could compound at 10% going forward, with some help from interest rates).

 

Let's take another example like CBI/FNMA or the like.  Clearly, way different outcomes, and possibly bimodal ones away from each other (e.g., strong peaks at par and $0 value with wide dispersions in between). 

 

So maybe saying this another way, I think it isn't terribly hard to make a "gross" curve, but an actually accurate one is of course impossible. 

 

From a manager's point of view, the question is the kinds of bets you are making/curves of the investments, and how they would combine statistically.  If one has good sizing and small positions, then lots of binary events with positive spread should give you a decent curve around your expected value.  On the other hand, erratic sizing or large sizing will create bimodal curves.  Holding a lot of BRK/MKL/BAM/FFH would presumably give you a pretty steady graph around 8-10%. 

 

The main issue with all of this is:

1) you have to be incredibly self-aware to describe your own graph and look at your outcomes; and

2) if you are looking for a manager, they need to be telling you 1), which is nearly impossible, as it is hard enough on its own and probably not in their best interest if it isn't good. 

[racemize]

Re: Why Hold Cash?

 

The context for this essay, and I tried to make it clear in both the introduction and the conclusion, is whether cash should be held based on macro concerns/overall stock market valuation concerns/etc. and ignoring the micro.  Since the GFC, this forum has been populated with people who would go into cash and talk about market overvaluation in spurts in fits from time to time.  These same people would also be invested in what they considered good ideas, but would hold cash because of these concerns instead of increasing their holdings of their own good ideas.  So, boiling it down to something extremely simple, I tried to answer:

 

If I have investment(s) I like, but I think the market is overvalued, should I hold some amount of cash?  If so, what is the ideal amount?

 

What I was not trying to answer, and hence my comments on "compelling investments" (perhaps compelling wasn't the best word choice, but I'm not sure what would have been better--maybe investments meeting minimum hurdle requirements, but that's wordy as hell), was: Should I be fully invested even if I run out of ideas?  I think the answer to that is a firm no as a value investor, which I believe most everyone agrees with.  Many people seem to be reading this essay to mean you should be fully invested at all times, which is not what I'm saying.

 

Anyway, the question of holding cash is mostly a function of volatility, so I wanted to find out what the answer was.  E.g., if you are hugely volatile in results, then holding cash just makes sense, as you get huge savings on the way down and huge gains on the way up with cash.  On the other hand, holding cash just sucks if you progress linearly or anything close to it.

[Guest cherzeca]

cash % seems to me to be very dependent on stage of life and wealth.  if you want to get rich, stay invested and concentrated. if you want to stay rich, be cautious and diversified. and if you are no longer working view that cash as a buffer between a happy and unhappy retirement. being cautious in a low interest rate environment is cash since interest rate risk seems to me to be too great. and long/cash seems easier than long/short.

[racemize]

cash % seems to me to be very dependent on stage of life and wealth.  if you want to get rich, stay invested and concentrated. if you want to stay rich, be cautious and diversified. and if you are no longer working view that cash as a buffer between a happy and unhappy retirement. being cautious in a low interest rate environment is cash since interest rate risk seems to me to be too great. and long/cash seems easier than long/short.

 

The essay was really only directed to the portion of your portfolio you want invested long-term in stocks.  Asset allocation based on life situation is just an entirely different question than I was after.

 

Really, this all boils down to, it's really hard to time the market.  In retrospect, not a surprising outcome, just one that I wanted to verify.

[John Hjorth]

cash % seems to me to be very dependent on stage of life and wealth.  if you want to get rich, stay invested and concentrated. if you want to stay rich, be cautious and diversified. and if you are no longer working view that cash as a buffer between a happy and unhappy retirement. being cautious in a low interest rate environment is cash since interest rate risk seems to me to be too great. and long/cash seems easier than long/short.

 

The essay was really only directed to the portion of your portfolio you want invested long-term in stocks.  Asset allocation based on life situation is just an entirely different question than I was after.

 

Really, this all boils down to, it's really hard to time the market.  In retrospect, not a surprising outcome, just one that I wanted to verify.

 

Exactly. As Joel already has posted, it is about being sharp focused on the framework used in the essay, mentioned both in the beginning and the conclusion of the essay.

[John Hjorth]

... Since the GFC, this forum has been populated with people who would go into cash and talk about market overvaluation in spurts in fits from time to time.  These same people would also be invested in what they considered good ideas, but would hold cash because of these concerns instead of increasing their holdings of their own good ideas. ...

 

Personally, I think this is true. This is what makes this essay valueable and worth reading , and rereading  - at least to me.

[Guest cherzeca]

"The essay was really only directed to the portion of your portfolio you want invested long-term in stocks."

 

i get that.  now if you are 90 and want to have funds till your 100, then your views may be different than my views over a 10 year horizon.  but history shows two things:  first, equities outperform anything else long term, so in theory you should be 100% invested in stocks for that portion of your portfolio you manage for long term. second, there are periods (say just before 29 crash, but not to same extent before 09 crash) where if you invest in equities you have an extraordinarily long period to get even.

 

problem is, you dont know if you are ever in that just before crash time.  so you either ignore what you cant know, or act as if the worst of what you dont know will happen...and that is a personal decision based upon age, wealth etc

 

i am not arguing here but simply trying to point out that imo investing is absolutely positively situational, and no general statements can be made, even about your long term portfolio allocation.

[Jurgis]

Re: Know Your Graph

 

I don't think it is terribly hard to create the general shape of the graph or figure out where the modes are for individual investments.  Your BRK example is one that a lot would generally agree with, I think, but obviously the details would vary from person to person (e.g., Buffett did say he thought intrinsic value could compound at 10% going forward, with some help from interest rates).

 

Let's take another example like CBI/FNMA or the like.  Clearly, way different outcomes, and possibly bimodal ones away from each other (e.g., strong peaks at par and $0 value with wide dispersions in between). 

 

So maybe saying this another way, I think it isn't terribly hard to make a "gross" curve, but an actually accurate one is of course impossible. 

 

From a manager's point of view, the question is the kinds of bets you are making/curves of the investments, and how they would combine statistically.  If one has good sizing and small positions, then lots of binary events with positive spread should give you a decent curve around your expected value.  On the other hand, erratic sizing or large sizing will create bimodal curves.  Holding a lot of BRK/MKL/BAM/FFH would presumably give you a pretty steady graph around 8-10%. 

 

The main issue with all of this is:

1) you have to be incredibly self-aware to describe your own graph and look at your outcomes; and

2) if you are looking for a manager, they need to be telling you 1), which is nearly impossible, as it is hard enough on its own and probably not in their best interest if it isn't good.

 

Below may sound like nitpicking, so please skip if you're not interested.

 

I think it isn't terribly hard to make a "gross" curve, but an actually accurate one is of course impossible. 

 

Right. Although even for "gross" or "approximate" one, it's probably harder than you said.

 

I thought about BRK and likely the curve is much flatter - less certain - than I suggested. I.e. we don't know more than we think we don't know.

 

Also with CBI or Fannie or SHLD, you think that there's binomial distribution, but likely the curve is quite flat too. I.e. everyone who argued that SHLD is zero or huge gainer in the last ?8? years have been wrong. Same mostly with Fannie. So maybe there was a hump towards zero, but likely much flatter than people expected. It's likely SHLD hit a probability spot that people thought was something like <1% chance (sorry this is a bit inaccurate given continuous prob distribution, but you know what I mean).

 

Sure there are steep fall-offs at certain points, e.g. there's probably almost no scenario for BRK to return 20%+ for 5 years annualized, but overall being certain of the humps is hard.

 

I think this relates to the behavioral economics question/test where people are asked to estimate weight of the 747 or distance to the moon with 99% confidence interval and they choose way too narrow range. The illusion of certainty.

 

I find this topic interesting, since I wonder if it's worthwhile for me as investor to try to come up with an approximately-right curve. And how much approximately-right.

 

Anyway, thanks for article and thanks for comments.

[giofranchi]

Thanks John, and thanks Joel. I've also appreciated your writings I've the years.

 

+1

 

Cheers,

 

Gio

[racemize]

Re: Know Your Graph

 

I don't think it is terribly hard to create the general shape of the graph or figure out where the modes are for individual investments.  Your BRK example is one that a lot would generally agree with, I think, but obviously the details would vary from person to person (e.g., Buffett did say he thought intrinsic value could compound at 10% going forward, with some help from interest rates).

 

Let's take another example like CBI/FNMA or the like.  Clearly, way different outcomes, and possibly bimodal ones away from each other (e.g., strong peaks at par and $0 value with wide dispersions in between). 

 

So maybe saying this another way, I think it isn't terribly hard to make a "gross" curve, but an actually accurate one is of course impossible. 

 

From a manager's point of view, the question is the kinds of bets you are making/curves of the investments, and how they would combine statistically.  If one has good sizing and small positions, then lots of binary events with positive spread should give you a decent curve around your expected value.  On the other hand, erratic sizing or large sizing will create bimodal curves.  Holding a lot of BRK/MKL/BAM/FFH would presumably give you a pretty steady graph around 8-10%. 

 

The main issue with all of this is:

1) you have to be incredibly self-aware to describe your own graph and look at your outcomes; and

2) if you are looking for a manager, they need to be telling you 1), which is nearly impossible, as it is hard enough on its own and probably not in their best interest if it isn't good.

 

Below may sound like nitpicking, so please skip if you're not interested.

 

I think it isn't terribly hard to make a "gross" curve, but an actually accurate one is of course impossible. 

 

Right. Although even for "gross" or "approximate" one, it's probably harder than you said.

 

I thought about BRK and likely the curve is much flatter - less certain - than I suggested. I.e. we don't know more than we think we don't know.

 

Also with CBI or Fannie or SHLD, you think that there's binomial distribution, but likely the curve is quite flat too. I.e. everyone who argued that SHLD is zero or huge gainer in the last ?8? years have been wrong. Same mostly with Fannie. So maybe there was a hump towards zero, but likely much flatter than people expected. It's likely SHLD hit a probability spot that people thought was something like <1% chance (sorry this is a bit inaccurate given continuous prob distribution, but you know what I mean).

 

Sure there are steep fall-offs at certain points, e.g. there's probably almost no scenario for BRK to return 20%+ for 5 years annualized, but overall being certain of the humps is hard.

 

I think this relates to the behavioral economics question/test where people are asked to estimate weight of the 747 or distance to the moon with 99% confidence interval and they choose way too narrow range. The illusion of certainty.

 

I find this topic interesting, since I wonder if it's worthwhile for me as investor to try to come up with an approximately-right curve. And how much approximately-right.

 

Anyway, thanks for article and thanks for comments.

 

Well, I'm not going to argue with the illusion of certainty, as that was a lot of what I was trying to point out in the essay.  I guess what I'm trying to say is that I also can't use the illusion of certainty as an excuse for not at least attempting to understand what is going on and trying to improve.  Anyway a few thoughts in with no particular coherence in structure:

 

Regarding approximate graphs: I do think those two have binomial properties, but there is the possibility of it being quite flat also.  FNMA though, I do think we can argue for spike at 0 and a spike at par, with potentially a large, very flat distribution around par, probably skewed in the towards $0 direction.  CBI certainly has a lot more uncertainty around it, so it is almost all speculation. 

I think perhaps we are just talking past each other on how approximate "approximate" is.  I think I'm probably arguing for something slightly more than a bucket approach that gets more refined the more certain it is.  For Berkshire, I think a lot of people would have some kind of normal distribution around 8% for IV and then you can slap on whatever market filter you want for what can happen macro wise.  If we are at a 5 year horizon, that latter filter is more important than the IV, but the longer we stretch out the horizon, the more the IV part matters.  For something like CBI, it's just very very hard because it is a high uncertainty stock at the present moment.  FNMAS should be easier than that because we know it could be worth nothing and it is easy to argue that par should have a decent high chance of happening too.  Certainly there are other outcomes that are not easy to know.  Mostly, I'm just saying that it isn't too hard to put investments in one of a set of buckets (e.g., higher certainty, low return; high uncertainty, larger dispersion of returns; or the ever-elusive high uncertainty, high but uncertain returns that we all believe we can find), then do some refinements, and then understand that even that isn't all that accurate.  But the buckets can do a lot of work for you, I hope anyway.

 

Regarding usefulness of graphs: I think mostly I'm taking the stance that the more I think about how uncertain things are and try to create graphs, the better, in both idea evaluation and actual investment.  E.g., it is another mechanism to cause me to analyze downside scenarios.  Also, to the extent that I think a graph looks one way at investment and then the outcome wasn't even in the graph, the more I can pound into my head how uncertain things are and hopefully get better at initial analysis.

 

The possibility of measuring graphs: Personally, I'm intrigued with the idea of looking at outcomes of investments as data points to generate an investment graph for me.  For example, I currently track every investment's individual IRR/performance.  I currently have something like 40 data points.  I can create a distribution from those data points and/or run statistics on them.  Obviously, this has some issues as the individual outcomes aren't representative of the risks being taken, but if I'm using a consistent judgement pattern, it can tell me something about my graph.  So far, I have something like 80% of investments being positive, but there's also a few blow-ups that I'm not too happy about (ZINC/Intralot/CBI (we'll see how the last two actually end up)).  Anyway, these are random ideas that have not been fully thought out.

 

I think this is a hard topic, but one that is important to think about.  Really the essay was my first attempt to get some thoughts down and an attempt to make it useful/actionable.

[Jurgis]

I agree with the rest you wrote, so I cut it out.

 

The possibility of measuring graphs: Personally, I'm intrigued with the idea of looking at outcomes of investments as data points to generate an investment graph for me.  For example, I currently track every investment's individual IRR/performance.  I currently have something like 40 data points.  I can create a distribution from those data points and/or run statistics on them.  Obviously, this has some issues as the individual outcomes aren't representative of the risks being taken, but if I'm using a consistent judgement pattern, it can tell me something about my graph.  So far, I have something like 80% of investments being positive, but there's also a few blow-ups that I'm not too happy about (ZINC/Intralot/CBI (we'll see how the last two actually end up)).  Anyway, these are random ideas that have not been fully thought out.

 

I think it's a good idea. I have to look back at Quicken and see if their per-security IRR has gotten correct. Might be a good thing to study for me sometime when I have time. Yeah, as you say, there are issues in terms of risks taken, but maybe looking at this will give some ideas how to deal with the issues.

 

Thanks

[John Hjorth]

New Peter Lynch compilation [not an essay though, ref. the CoBF topic title ...] in the works from Joel here.