Tuesday, December 8, 2015

The Learning Process and the broken feedback loop of Investing

There's an idea that's been jostling in my head for a while related to how we typically learn, and how investing simply doesn't fit our normal learning process.  

The basic idea is this:  It's difficult to learn what works in investing because the feedback loop is - or can be -  deceptively removed and jarringly out-of-sync with out normal learning methods.  Investing happens at a speed far slower than the speed of thought, and this poses problems for our learning process.  And to complicate matters further, results are noisy.  It is possible to be wrong and make money, and right and lose money.  And even then, the timing of an idea might make all the difference.

The key issue is this:  With most things, you get feedback within a reasonable amount of time that lets you know whether you are doing something properly or not.  You can then make adjustments, try again.  Incrementally you get better at the task.

For example:  Take shooting a basketball.  The first time you shoot a basketball you're probably small and uncoordinated, but have a general idea that tossing the ball in the direction of the hoop is the right thing to do.  When the ball misses (unless you're extremely lucky on your first try) you get immediate feedback that lets you know your error in any number of ways - direction, force, height, touch, etc.  So you get the ball again, and repeat, making adjustments.  Gradually you learn the actions that help improve your basketball shot - how to position your hands, elbow, flip the wrist, arc on the ball, timing the jump, etc.  It's all incremental - but the with the key point being that you get feedback that informs improvement on the next try, and the next try, and the next try.  In essence, shooting a basketball becomes a local optimization problem in a complex mechanical process.

Now in contrast, if we got feedback from shooting a basketball like we got feedback from investing investing, you'd take your first few shots, and then wait about 5 years with the balls in the air to see how you did.  Was it an airball, did you hit the rim, clunk of the backboard, string music?  Crucially, by the time you'd get feedback, you'd likely have forgotten what you were doing when you made those early shots - perhaps even forgetting why you took the shot in the first place.

To complicate matters with investing - with randomness being a strong component - it is entirely possible to in effect invest/throw a ball up in the air completely wrong and have it go in - on a somewhat regular basis.  Similarly, it is also possible to have the proper process and shoot an airball - on a somewhat regular basis.  So as I see it, the feedback process with investment is broken in ways that we typically don't encounter in life.  Or at minimum, the feedback loop is very loose in that you can be regularly rewarded for bad behavior and regularly punished for good behavior.  It's only with sufficient time does a better picture emerge, and as mentioned above - after so much time it's difficult to understand and be able to learn what is "working" and what is "not working" about the investment process given the confusingly reinforced feedback effects.

I guess I'm looking at 4 particular outcomes, and how they inform the learning process.

1. Right thinking, Profitable Result.  Ideally this is what we want.  It's shooting the ball properly, and having the ball go in the basket.

2. Right thinking, Unprofitable Result.  This type of outcome in investing happens all of the time.  Here I'm first reminded of Peter Schiff who awesomely called the housing bubble and financial crisis in 2008, but those who invested with him still lost from 40-70%.  But on a more pedestrian level, it's easy to think you understand what is going on (and largely be correct), but still lose money because of other factors that come into play and interact unpredictably or in ways that are unappreciated.  I currently have an investment in CF Industries, a nitrogen fertilizer company that I put in this category of a pretty simple story that I view largely as "right thinking" but "not-so-profitable" result.  The main input into nitrogen fertilizer is natural gas, and due to the fracking boom natural gas prices in the US are multi-year lows, making CF one of the low cost producers domestically and internationally.  That part has held up.  What hurts though, is that commodity prices are cratering, so things like corn that use alot of fertilizer, are seeing lower prices, and as a result farmers are spending less on fertilizer for the now less valuable crops.  Additionally, the company is going on an acquisition spree buying up capacity globally - something I was not anticipating that ups the risk level of the investment.  And gradually more capacity keeps coming on line even as prices drop, as new plants are being built by CF and competitors who weren't expecting price collapse when construction started years ago.

But ultimately what underlies these type of situations, is that it's entirely possible to have a good idea turn out badly - making it difficult to learn properly and distinguish between that good idea and a bad idea. 

3. Wrong thinking, Unprofitable Result. Ideally, this is what you like to see for proper learning.   Bad idea gets punished by the market.  This is that sub-prime mortgage lender I invested in prior to the 2008 crisis.  Yeah, I did that not understanding what garbage those loans would turn out to be.  I'm more careful with financial companies now, and largely stay away from them.

4. Wrong thinking, Profitable Result.  These are the "oops" situations where a stock goes up for unexpected reasons.  For instance, I almost invested in Yahoo a few years ago in the low-to-mid teens thinking they had a valuable franchise, and should be able to figure out how to monetize all the viewers they have.  The stock certainly rallied from that point to over $50/share, and is currently priced at $34.85.  However, the rally had little to do with turning things around, it was mostly due to Yahoo's ownership stake in Alibaba.  Here something totally unexpected drove investing outcomes.  It's like the ball slipping out of your hands, but still going in the hoop.

5.  Timing is Off.   OK, I'm adding a #5 here, because even if you get other things right, just having the timing off can ruin your returns also.   This idea falls in the vein of "being early is indistinguishable from being wrong."  Investing has a time component, and your investments can simply take too long to work out even if your idea is sound.  I currently have an overweight investment in energy/oil stocks, and it is increasingly looking like they are going to take much longer to work out than I had hoped.  Given marginal extraction costs and assuming global demand growth, I'm fairly confident that oil prices will eventually settle much higher than where they are presently, but the "when" is an important part of the overall consideration that helps determine whether an idea is profitable or not.  Opportunity costs.



So the thing I'm left wondering is how can the learning process with investing be improved.  Ideally, I'd be able to analyze a situation, make my choices, and push a button and see what happens in five years to know whether I'm right or wrong, and understand why.  Then repeat the process.  But that isn't possible.  Similarly, I can't go back in time 5 years ago and try to recreate that scenario, because I already know how the world has changed in that time, and what ultimately happened with stocks like AAPL and GOOG on the upside, and what has happened to Coal and Gold stocks on the downside.  And in any event, as Howard Marks points out, more things can happen than will happen, so even if you run the clock forward and get one outcome, if you were to repeat it again, you'd likely see something different happen.

In these types of situations where uncertainty levels are high, it is pointed out from multiple sources that adhering to and focusing on a proper process is critical.  (Focus on process, not the outcome - like a poker player who is happy if he gets chips in the pot when the odds are in his favor, even if he loses on the river).

A key issue in this case is to determine the proper process.  I can take the quant approach and focus on factors like value, momentum, quality, and size to improve my odds and apply a process to that approach.  That is one way to go, and I do that for much of my filtering process.  Essentially it's just a way of saying, "in the past, investments with these types of characteristics have tended to outperform in the past, so going forward I'll pick those types of investments in a systematic way."

However, if I'm trying to learn to select individual securities and learn about the selection process and the errors I make in that process, it seems a much more complicated effort to "learn" than is required by the quant approach.  Maybe I should be happy with a quantitative approach, but I'd also like to be able to successfully apply other qualitative factors in a way that is productive to the overall process.

For example, right now OUTR Outerwall is a company that has been on my radar and it just dropped 25% today to a share price of  $44.  Their business is mostly Redbox movie and game rentals from vending machines.  Despite being a business that is probably in long term decline, from a quantitative standpoint they look to have very high amounts of value and quality.  The business generates tons of free cashflow, and they're probably trading at less than 4X free cashflow currently, so if they just have a "little bit" of life left in the company they should pay back.   Digital streaming (or something similar) will eventually put them out of business, but the question is how much money will they make prior to then?  And will they not waste those profits trying to unsuccessfully grow?  To me it looks like they should earn alot more than their $44 share price over their remaining life, but I'd love to be able to "learn" from the current situation by being able to hit the fast forward button and look forward 5 yrs and see what happens in the interim.  What risks am I seeing right?  What risks am I missing?   I'm aware that I like contrarian plays like this, but buying stocks in strong downtrends is dangerous (and quantitatively not a good idea based on probabilities), but I can't help but be drawn to these type situations.  It just looks so cheap assuming it has some life left in it.

I guess in five years I'll check back and see what I can learn.