You are not alone if the title hurt your head. I think my eyes are bleeding looking at it, and I had to pop two painkillers. I was flipping through a book at Barnes & Noble (shout out) and I ran into some math regarding probability through time.
This article drifts all over the place. Holy crap. I am adding this paragraph after writing the whole thing. I get seriously out there about other stuff and bring in some other concepts. I think its useful, but slightly confusing. Proceed at your own peril. I do apologize.
Math is not a strong subject of mine, but statistics and particularly probability theory were my best subjects due to their relevance in gambling. Specifically, poker and blackjack. I got pretty good at 7-card stud in college but most everyone played Texas Hold’em, which I was rubbish at. I did acceptably well in blackjack.
This post will probably be split between this site and The Chaos Vault, because I have not figured out how exactly to tie this principle to investment analysis except when it comes to the validity of negative information and the importance (and manipulative effects) of timelines. The post on my “for fun” site will probably get into a lot more tangential information from a math or novelty perspective.
Flipping Through a Book Caused This Pain
The table I saw was a bit beyond me at this point. I plan on starting that book in the next month or so. The jist of it was whether the non-occurrence of an expected event in regards to a theory, is strong evidence that the theory is wrong. Theory here means almost anything including a series of events or otherwise. For example, the market crashes every five years is a theory. Real estate will go up in the long-term is a theory. Real estate will collapse once every 50 years is a theory. That you will get a tails after flipping 4 heads in a row is a theory.
Let us take real estate. Assume my argument is that real estate will go up in the long-term. It is simple and overbroad. Now for the theory to be true real estate prices will have to increase over time to be accurate. That does not mean straight line it just means that any sufficiently long period of time will net out to a positive gain for real estate prices. Now we said long period of time so let’s go with 10 years for the sake of ease.
I am not capable of doing the math yet, but lets tackle it qualitatively. If over the course of 10 years the price of real estate does not increase, is my theory wrong? From what I understand the likelihood that my theory can be proved by 10 years of non-occurrence is lower than you might expect. Even on an extreme level, assuming that real estate does not even move in the 10 years, the theory might still have a pretty good chance of being valid. Now, I know that is not very satisfying, but imagine you were betting on the theory. The theory is not established, you are collecting data.
If you told someone to assign odds about the validity of the theory they would probably rate the chance the theory is false very high. In reality your chances are pretty good, and betting a small amount on the validity of the theory could give you a lot of money if it is proven accurate. I say that without getting into the giant headache that is trying to determine what evidence you need to prove your theory. For this example go with, God said it was true.
The chance that your theory is true if the price of real estate went down in that 10 years would affect the probability a bit more, but not as much as you might think. Now I am going outside the scope of the book I was flipping through. This is more about how we as humans perceive probability. This delves into the whole Black Swan thing. That concept works on the other side though. It says that a theory can be disproved by one negative. I am saying that one negative does not disprove a theory, but the difference is that “real estate goes up with time” is conditional on the last word. Time is key. The black swan thing was an absolute. “All swans are white” is not going to change with time. It is going to change once you find one black swan. So absolutes are destroyed by one contrary piece of evidence.
When it comes to time conditional theories contrary information is not dispositive depending on time. Mathematically, I think that even if you had a theory that has evidence from the beginning of the universe that runs contrary to the expected outcome if the theory is true, you would still have a small probability that the theory is eventually proven true.
Now most of this depends on the initial likelihood that something is true. The real estate will always gain theory probably has a high probability of being true. However, after 10 years of contrary evidence, we as people would probably disagree with the theory.
Most of that was pointless, and I have probably just confused all of you. However, you can see that how we perceive events and non-events differs from the reality (okay you can’t see it, because I have not proven it, but take my word for it?). We humans are prey to our perceptions and the power of heuristics. Our brains conserve processing power. We hold what is likely to be true based on minimal evidence, and we do not consider things thoughtfully. Sometimes (almost always) some (all) people (people) use that to manipulate people. It is not always intentional, but it happens. It like logical fallacies. Some people who use ad hominem do not realize that their argument is a pile of crap, because attacking the individual does not invalidate their position. Tell that to any politician or voter. For something called a fallacy, we as people are really in love with them. Another logical fallacy is overgeneralization, like “people are not to be trusted due to their believing ad hominem almost en masse.”
I think I have discussed this before. Everyone has a timeline that proves their theory. If you are not right in 10 years, try 20 years. I always use gold as an example. Obviously now the metal has lost some of its luster, but a few years ago it was all the rage. Always go up they would say, and yet on a 30 year time line it can be considered to be down from its high. On an inflation adjusted basis gold still has not recovered from its high in what was it early 1980s… it was during Carter so early 80s. I’m only 25 so I don’t remember Jimmy Carter as president, but I am just a history buff.
Real estate is down from 1890 to now. That really does depend on geography though. Still the point is that the boundaries of an argument can actually determine its outcome. It reminds me of a couple of classes I took in politics. I knew one person who was at one point a professional pollster and he explained to us how he could get whatever results he wanted in a survey based on how he framed the question. This included the order of questions as well as the wording. It was always funny to see how a yes/no question would elicit a different result from a choice.
It is like asking if you are a conservative, yes or no? Would yield different results if you asked someone whether they were conservative or liberal. When given the option people gravitate to what they most identify with as you’d expect. However in a yes or no instance, I can say no I am not a conservative, and in their head they know they are also not a liberal. Then you ask separately about party affiliation. I am not doing a good job of finding really compelling questions, but it matters how you ask a question. So even representative sample polls are gamed, based on the survey. Scientific polls about science stuff like social science (not politics) can be more trust worthy if it is about data finding. Political polls are goal driven. More people agree with me so you game it.
How you frame things matter, and in finance that is usually timelines. A lot of services and funds will advertise how they have done since 2008. Really I would like to know how they have done from 2006. how did they handle the nice highs in 2006 to now. Instead going from 2008 means they are picking the lowest of the low. It is like if I said how well as the stock market done since the moment of the flash crash? Well that massive decline is not really a good place to start, but it really would prove my point. You could have been stupid into 2009 and made money off 2008 lows. It is not impressive that your newsletter had 70% gains in 2008. It’s an outlier caused by factors beyond your control. You see that crap a lot though. Timelines are used to portray luck as skill.
Tying the Two Together
This affects all of us. Even here I write about stocks, but I understand my limits.
If my argument is that this stock picker or money manager is lucky, and he has 5 years of never getting one wrong… what is the probability that he has skill instead of luck? One big win is definitely not enough to prove prescience. Paulson bet and won with real estate. Luck or skill? Both? Probably both. Considering his fund lost a lot of money after his big real estate play, is he really that much better than anyone else? It was a mix of skill and luck that he managed to see what was happening in real estate. After all, that information was available to everyone. He was smart enough to see it and make the connection. He had the guts to bet on it. Does that mean he will perfectly predict the next crash? Unlikely. He probably won’t even predict the next real estate crash. Unless it looks exactly like this one and he gets déjà vu.
Do not jump to the other end of the spectrum. He is not just an average guy who got lucky. This is not a lottery victory. It was just the right set of events, and he had the right information in front of him, and the right mindset. All of that sounds like skill though. It’s really not. He can’t do that with any set of data. You can’t give him all the data in the world and expect nothing but gold. He made the right connection once and won big. His hedge fund lost money in 2011 and 2012. He bet the wrong way on certain developments. It happens. It is also true that someone else presented the same information he had might not come to the same conclusion, and even if they did they might not have invested the same way. Luck mixed with skill, but notice how his success is not easily replicable even by himself.
Wow this article drifted in topics a lot. The point is that the one event did not prove he was an oracle. Even over the course of 10 years, the argument that a stock guru is good cannot be significantly proven even if you only have positive information. If all your evidence over 10 years points to the validity of a theory, there might still be a significant chance the theory is false. It depends on the initial probability of the theory. I think we tend to draw trends and make conclusions to quickly and easily. Other people frame the data in such a way as to prove their arguments, but do not fall prey to this. They are manipulating you through the use of timelines and specific data that suits them. It might not be with malice, but it is still manipulative.