Why Financial Gambling is Easier than Life Gambling

In my view one of the most important component of alpha generation (financial gambling) is weighting.  How large to make your bets.  Interestingly this is not often discussed in the class room or the board room.  Valuation, timing, and strategy are discussed but when was the last time someone asked you what the geo-metric mean maximizing weight of your strategy was (if ever)?  [An interesting side discussion is, 'is the reasons it is one of the most important components of alpha generation because it is discussed so little'].  Also the formal designs of weighting are much less developed (or maybe it better to say much more complicated; even when it is a single strategy weight and not a strategy in the context of a portfolio) than valuation.  But for the under use and difficulties that the sizing of bets entails it makes financial gambling much much easier than ‘life gambling’.

Consider that for a financial gamble you can scale it to your confidence, without (for most of us) having any impact on the market.  In short the way you scale the bet doesn’t impact the way the bet turns out.  On the other hand in life you can’t just scale the bets to confidence because how you scale the bets (which in this case I mean the effort/commitment that you put into an activity) changes the outcome significantly.  Consider the realm of foreign policy if you think  a foreign excursion (pays off 10:1) is worthwhile even given it only 20% chance of success.  A PM (portfolio manager) would put on a weighted bet based on the payout but the statesman would have to calculate the weight that he was willing to commit (and at the same time) consider what they would be doing to the chances for success.  A more challenging problem.

Distribution of Estimates or Estimates of Distributions

Bernanke should receive the Nobel Peace Prize…what he did for the U.S. economy and there in by the global economy has keep this downturn in hand.  I’m not suggesting that the next great power war would have emerged if he had allowed a deflationary dynamic to gain traction but I find conflict to be feed by the fire of hardship and crisis.  So keep this recession from being a depression is worthy of the prize.  Now why am I saying this.  Honestly, no particularly reason just wanted to put it out there.  Now I’m going to criticise something that I believe he was instrumental in doing: the distributed estimates economic growth by the participants’ at fed meeting.  Here are some of the GDP projection and distribution charts from the minutes release today. 

Ben has for quite some time advocated for transparency (ranging from inflation targeting to new disclosure requirement to the forecast disclosure) with regard to policy making.  In a lot of ways this makes sense (if they have more credability the job become easier, although if they just confuse people with the new information credability is lost).  Now I love the way these chart suggest you should think about the world, that of a distributed or uncertain future.  The idea is we don’t know what the future path of GDP growth will look like but these are our range of guesses and lets work with that range instead of a single estimate (…being a fatalist does make decision making easy though (note: that was a joke)…).  But when I was thinking about these charts I realized that they are really a range of point estimates not a true estimate of the range.  This is a suttle but important difference; I might expect (call this the most likley case) growth in 2011 to be at 4% but if I and everyone else thinks that there is a 25% chance of a L recovery (or 1% growth) in 2011, it will not be captured on this chart even though I suspect it really should be.  This leads the chart to be overly rosy (in that particular example).

Ben – Next time make the chart as an estimate of the distribution instead of distribution of the estimates.

Is More Information Better?

My undergraduate thesis was entitled ‘Is More Information Better.”   It was a study of transparency at the Federal Reserve over time.  Looking at it now…the paper made some obvious conceptual errors but in the context of an undergraduate thesis it wasn’t bad.  What made me think about this was a chart I saw in the FT….

The idea is that as the handicapper recieved more ‘information’ they did not become more accurate although they did become more confident.  People are almost always over confident.  I think I read somewhere (although I can’t find it now) that the profession that is least over confident is weathermen…which means if you are constantly told by everyone that you are wrong you may have the right degree of confidence (I think doctors were the worst). 

Another interpretation of this chart is that data is not information.  The handicappers recieved more data but not more information and they were unable to digest that the new data was not useful.  That is how I concluded my thesis…

More information is better but only when it is valuable and relevant.  If increasing the amount of information clouds and confuses the reader as to what the information means that additional information is not valuable.  Thus the FOMC should, as they have been doing, use language very carefully in order to maintaining their own credibility. 

Transparency is not a simple thing and more information is not always better (even when it is clear..although that is a different story).

Those who are right are wrong

People like things that are clear, ‘If you are right, you are right and if you are wrong your wrong.’  The rhetoric is less true than we would like to admit.  We give too much credit to people who get ‘it right’ and harp on people too much that get ‘it wrong.’  Honestly I don’t think anyone predicted the collapse of the financial infrastructure that has happened over the past year and those who where close – albeit wise and insightful – also were off base.  In that the truth was what happened was one possible outcome (the one that happened) but it did not have to happen that way.  If we hadn’t let Lehman fail, if agency lending had not be curtailed, if we didn’t rescue the agencies by hammering the preferred (owned by small banks throughout the country) the deleveraging that was going on may have continued in a some what orderly way and not in the crash and burn method that happened.  I was thinking about this after reading Brad Setser’s blog post last week about selective interpretations of risk theory and how that lead to a weak system.  I agree with the basic point but when I was reading it I got the feeling as if somehow what happened was an inevitability and thus proving without a doubt that the prior risk taking was foolish; which I don’t think is true.  Put another way, if you never fail/collapse/miss the flight your not pushing your self hard enough/not taking enough risk/leaving to early.  So failing – albeit bad – is not proof (although it is evidence) that something is wrong.  My point is that just as you should consider a probabilistic future when considering decisions; when evaluating past decisions (and don’t get me wrong mistakes were made) you need to evaluate them in the context of probabilistic outcome not simply what happened. 

Consider the following extreme example.  Imagine you saw the following: person A offer person B 100 dollars to play Russian roulette with six shooter (with one bullet), and person B took the offer and ‘won’ (didn’t die) and was paid the 100 dollars.  You would say to person B, ‘That was stupid.  Why did you do that?’  If they responded ‘look at the evidence, I was right’  and then walked away.  You’d would not think to yourself that the person had made a good decision (had been right) but that they were crazy.  They were ‘right’ but clearly wrong.

Why you want her to say no.

Now I’ve use a matrix framework (game theory) to describe how to think about epidemics as a policy maker and implied that I would illustrate a personal application.  How about asking someone to marry you (or asking someone on a date – although this works much less well because the commitment (duration and exit costs) are much smaller)? 

Consider the following matrix:

Now there are some underlying assumptions in this matrix, such as the true desire to say is connected to the relationship working.  If you think people are incapable of knowing what they want then just stop now..I think this is a reasonable, most of the time, assumption so let jump through this as the ask-er (the male in this case).  Unlike the policy maker he doesn’t get to pick either of the rows but but he can create an environment in which certain combinations are more or less likely.  For example asking in public will (likely) encourage a yes answer, asking after a romantic dinner will (likely) encourage a yes answer, asking after a weekend away will (likely) encourage a yes answer.  Now realize that if she wants to say yes it is very unlikely that she needs to be encouraged (as I note: the answer being no and her wanting to say yes is odd…update: although it was pointed out to me that parently issue make this not uncommon).  But if she doesn’t know or wants to say no these romantic moments/actions may inspire her to break from that underlying true desire and say yes (which mean you live in the ‘your life is going to suck’ box). 

So what am I saying….strategically you want to optimize the chances that she say no.  If there is any chance she is going to say no make it really easy for her to do so.  Now you don’t have to start a fight and you don’t have to ruin the romance but you may be better off not to increase the pressure to say yes.  Or consider talking about it as you would with any other real question, then after you know she wants it you are free to do some thing romantic.

Epidemics and You

Epidemics are great case studies to think about probabilistic thinking.  Deadly flu epidemics are expensive and destabilizing.  They are also very hard to predict.  Given this, pretend for a moment you are the head of the CDC and need to respond to a possible pandemic.  You have to make a choice before you know if it is a real pandemic because by the time you know it will be to late to respond.  So look at the following matrix.

 

You have to pick A or B and then you’ll find out if the world is in a C or D state.  This is a tough call and you have to figure out what the huge costs really are and remember the world isn’t nearly this discrete (there are many other half solutions).  There should be a bias towards guessing there would be an epidemic which is why the CDC comes out with press releases and warning when the death rate for swine flu is much lower than car accidents (as illustrated in the following chart).  (Think…they are consider the explosive potential path…no need to worry about car accidents they are not really contagious). 

The pink area in the above chart matches up with the Red text in the matrix.  This is what I’m calling the unacceptable solution.  The world in which civilization as we know it is at risk.  To avoid this the policy maker must treat any possible epidemic as if it will become an epidemic.  This means they will be wrong alot but we actually want that. 

One challenge for the policy maker is to maintain their credibility in a world where they are (and should be) often wrong.  It is hard to imagine that the general population will simply understand the chaotic dynamics and more likely that they will respond to some degree of fear mongering.  If this is true consider how should the policy maker use that tool.  Admittedly it is a dangerous tool but is very powerful (Shame is a less powerful but less dangerous tool).  While considering this remember that from the individuals perspective until the epidemic is already an epidemic is probably doesn’t make sense to change their behavior (except for the very cheap things like washing their hands); just like voting is not ‘rational’ because the likeli-hood the vote matter is so small that it is not worth the time (even for a minimum wage earner) thus policy makers have use shame (in the form of duty) to ensure that democracy keeps working (and that people keep voting).   They use disinformation (albeit not complete lies) to inform the audience. 

Now this matrix framework (game theory) is a useful way to think but you may say, ‘Why would I ever use it’…consider the following …

Be Confident that we don’t know…

In an era of fairly dire economic predictions it is, in a way, comforting to reflect that these prediction are by no means set in stone.  They are guesses, it could be better (although it could be worse).  Looking at some predictions about today from a decade ago illustrate how off we can be.  Look at the following IMF report (based on CBO projections) that the U.S. would have negative marketable debt (or have excess cash of debt) starting in 2006.  Now this projection was before the Bush tax cuts which would have altered these projections but even with that in mind the change in circumstances is incredible.  What is the take-away…be skeptical and operate knowing that we live in an uncertain world. 

Crowds

The economist highlighted a statistics that this weeks crowds at the G-20 meeting in London where relatively small and peaceful compared to protests over the past decade.  Trying to anticipate what will spark a crowd and how a crowd once sparked will move and develop is next to impossible (in my view).  

That being said I was thinking about the following: many (if not most) participants in protests are there for the protest itself and are fairly agnostic to the cause (they are just there to watch).  Look at the pictures from London, there are more people watching the protestors confront the police than people confronting the police.  Consider the Hungarian Revolution in 1956.  Although the Hungarian Freedom Fighter was ‘man of the year’ in 1956 it would have been hard to find someone who participated in the violence because as a percentage of Hungarians on the street it was a low percentage.  From a personal very limited/micro perspective the proportion of people watching the crowds after a NCAA basketball national championship as opposed to actually participating in was (or seemed like) 100:1.  So if the protest crowds actors are a tiny minority why so many police? Are large police force unwarranted. 

No.  They are a major deterrence from the spark becoming a blaze.  Think of the typical crowd as having a 50:1 watcher:participant ratio, this gives the true believer (the participants) a great opportunity.  If they can get the other 50 involved they can give a showing of their power which is disproportionate to their actual power and – if they are pursuing a forced change in policy – greatly increase their chance of success.  [For a descent fictional account of this type of thing read: Steinbeck's In Dubious Battle].  From their perspective  they want to create the perception of power.  This mean make your event/protest/cause the ‘it’ thing and increase the amount of powder in the store house.  Of course this can backfire if you overreach and are perceived as a threat to the established power core priorities (ala why did Russia at first give the Hungarians some autonomy and then crush the revolt). 

These channels to portray power create a challenge because when the fire of a large protest is triggered the ‘owners’ of the protest can have a very hard time controlling it and communicating and thus accomplishing their ultimate goals.  I think Jefferson overestimated the ability of crowd driven revolts accomplishing good result when he said, ‘god forbid we should ever be 20 years without such a rebellion, what country can preserve its liberties if its rulers are not warned from time to time that their people preserve the spirit of resistance let them take arms.  What signify a few lives lost in a century or two.  The tree of liberty must be refreshed from time to time with the blood of patriots and tyrants it is its natural manure.’  He said this in support of the French Revolution which in hindsight was a very costly method of accomplishing what it did. 

So having police appearance decreases the risk of a small minority being able to hijack a passive majority to accomplish its goal by increasing the perceived costs of participation. 

[Side note: If you think this is really interesting take a look at MIT's OCM class '21H.001 How to Stage a Revolution']

Quantity vs. Quality

People are prone to over weight information.  You should note that when someone say that we must do X because the science or the data said so that it can’t be true.  First off data is only an picture of reality and often a very fuzzy picture; it doesn’t ever tell you what to do.  It helps present the consequences of actions but doesn’t tell you which path to take.  Unfortunately the fuzziness of the picture doesn’t come through point estimates.  Thus reasonable skepticism is beneficial to helping you become successful.  With this warning put in place I’d like highlight a data set that I’m highly cynical about (and doesn’t help you make any decisions) but is interesting. 

The reason I’m cycnical about this data set is it about sexual patterns (If I have to pick two things, of the top of my head, that people lie about (explicitly to others but not the top two implicitly to themselves) it would be sex and drugs).  That being said a survey of high net worth people was taken about sex.  The sampling is interesting because I wonder if this group of people (average net worth of 89 million) are more ‘natura’ because they are not limited by typical life constraints and thus are more flexible to ‘do what they want.’ 

Both men and women thought sex was important (81% of women and 76% of men) although 72% of women had an extramarital affair (‘only’ 54% of men had).  [Note these high numbers suggest to me that this survey is better than I originally thought...who would lie to say they had an affair when they didn't].  These points may have suggest that your stereotype might be off but there was some affirmative evidence of standard mindset: Men though the benefit of being wealth was more quantity of sex while women though the benefit was high quality. 

Confirmation Bias

Voltaire said ‘doubt is not a pleasant condition but certainty is absurd.’ 

Although many, if not most, people implicitly operate based on probabilistic thinking.  It is uncommon for that thinking to be pervasively explicit (systemically intentional).  Because of its implicit nature, far too often,people will behave as if the after an event has occured it was a certainty.  Now I don’t plan to argue about an all encompassing pre-destination either from a theological or Newtonian perspective; I simply don’t need to counter the argument that events are certainties.  I only have to prove that you – as an actor who knows his desires- cannot know with certainty the future and thus the best way for you to operate within this uncertain world is to act with an explicit recognition of the likelihood of events and the distributions of outcomes.   (I know there are some paths of this debate I’m not addressing but you’ll have to forgive me because for the moment that is not the point). 

With the background out of the way let’s look at one benefit of this modus operandi: an enhanced ability to change (for the better).  Beside the reality that you will be able to more quickly recognize opportunities and risks with explicitly probabilistic thinking but you will also not be embedded to a singular mindset.  You don’t have to commit inextricably to some tenant and thus feel as if you are a ‘flip-flopper’ or betraying your first love when you change your mind.  This is not a trivial problem.  Consider that fact that most people go into debates knowing which side they will choose well before anything is said. 

People will start pulling out evidence that supports their position and ignore evidence that counters it.  If that person was thinking probabilistically about the issue they have an easier time taking in new information because any information is not counter to your view but simply causes slight shifts in the estimation the probabilities or payouts (and your confidence in them). Another way to say this is, there is no emotional angst because everything is a confirmation. 

This self confirming bias is most problematic when little is known about the subject.  This failure comes from people not treating their actions around a probabilistic world with levels of confidence.  People feel uncomfortable saying I don’t know (or ideally this is what I think, but with an extremely low level of confidence). 

An example of these failures in logic is being displayed at the G-20 meeting with the reception of Barrack Obama.  He stumbled over Dmitry Medvedev and he was given the benefit of the doubt that he is not foolish (as he should be, really who cares if you can pronounce names), but I’m suspicious George Bush would have received the same treatment.  According to a graffiti artist in London, Obama has ‘good morals.’  I’m suspicious that he has (1) the right level of confidence in that statement and (2) the probabilistic nature of if that view is true.   For the U.S. this bias is good because we may be able to achieve goals that we may not have been able to achieve before and the greater the bias the longer we will be able to get away with over achievement. 

[Notes: (1) Obama popularity with street people really does matters to foreign leader because working with him will affect their domestic political standing and (2) foreign policy issues where there is a real conflict will be determined by national interests not by personalities].