Generalisations – not all bad?
EFN505 Financial Risk Management Essay and self-reflection due 6.1.2019
By Adam Atkins, student, QUT Master of Business (Applied Finance) 2019
- “To Generalize is to be an Idiot to Particularize is the Alone Distinction of Merit – General Knowledges are those Knowledges that Idiots possess.” William Blake, On Reynolds (1798).
- ” Generalizations and perceptions as life hacks of risk management. Ruin!”
- “Nothing can please many, and please long, but just representations of general nature. Particular manners can be known to few, and therefore few only can judge how nearly they are copied. The irregular combinations of fanciful invention may delight awhile, by that novelty of which the common satiety of life sends us all in quest: but the pleasures of sudden wonder are soon exhausted, and the mind can only repose on the stability of truth.” Samuel Johnson, Preface to Shakespeare (1755).
Arousal: The genie and three wishes:
Generalizations and perceptions as life hacks of risk management. Ruin!
The wishing person canvasses the genie for whatever he might wish for if he had perfect knowledge, and in particular of values he should hold. The genie creates a fourth rule to exclude this possible wish. Why is that? Perfect knowledge is not possible with our two-kilogram salt-electric brains. Even advanced machines cannot have perfect knowledge. It uses heuristics to approximate things to perhaps micro gradients. A generalisation is “a life hack” or general expression of things in context and does not need to be perfect, it just needs to work. We can be particular, and narrow the issues down with more detailed research, analysis and review. But as a general rule, a generalisation can do just fine in the short term, in life and in finance. However when generalisations are based on low quality information or beliefs there is a higher risk of ruin. This essay will argue that in the short term a generalisation may be better than nothing but with sufficient time deeper analysis is required.
Algorithmic trading and generalisations
In capital markets and financial trading, algorithmic trading is now so common it no longer attracts newspaper headlines unless there is a flash crash. An algorithm is a program that follows a series of rigid steps, it complies with a set of rules and responds to stimuli as given in the rules. This means that computer programs are making decisions to buy and sell financial instruments, in split seconds with potentially huge sums of money.
Algorithms can replace a human that would have performed the same task but do it slower and are prone to mistakes. Algorithms can be successful in making money in financial markets, otherwise these would not be used. Algorithms follow rules set by programmers based on generalisations given to them by financial advisors. Computers are getting “smarter” learning as they go and particularising decisions. However algorithms must be monitored and tweaked constantly to meet the needs of the current market.Good algo-trading works often enough and follow the rules rigidly, unlike a human that may make a mistake or stray from the rules, get tired, are not vulnerable to time degradation or make emotional or biased decisions.
Machines or bots like algorithms do not need to be perfect, they only need to be better than humans, and sufficiently better than competitor bots. Like in computer simulations like strategy games, the winning bot need only be about 1% better and the effects compound quickly. So, is a trading algorithm perfect? No. We see flash crashes from time to time, and when the market moves in apparently irrational and huge swings these are often blamed on “algo-traders.” The algo traders are in many instances following general rules given to them. An example momentum algo-trader may have rule that if the market is going down, sell, and if going up, buy. So, if the market takes a sudden turn downwards this is accelerated by algo-traders selling, and stop losses being broken and positions closed out.
No one has perfect knowledge of the markets, of the past and future and no one can make perfect decisions. What we as financial managers and advisors can do in bringing our experiences and knowledge, including generalisations and particular information and maximise prospect in normal conditions. Of course abnormal conditions occur from time to time and judgement calls need to be made by decision makers with incomplete knowledge. Human knowledge can be transferred into algorithms or may be used to educate other humans with tasks done analogue by a human. Either way decisions are being made with imperfect information and analysis. It may be possible for a computer to access all the available information and have as close to perfect data of a financial market, but it would still need to account for the epsilon unknown factors and its models could never be perfect.
A generalisation on time is better than perfect too late
During my military career I was required to formulate complex plans with many moving parts in a friction filled environment under adverse circumstances. These plans involved making decisions that potentially held life and death consequences. Further, these plans are always subject to change with no notice and as every military planner knows, no plan survives contact with the enemy. To assist military decision makers there is a maxim is given, attributed to World War Two general George Patton, that “it is better to deliver an 80% plan on time, than a 100% plan from the graveyard.” In other words, if you waited for perfect information, such as if you delayed making a decision until your analysis was a perfect as possible then it would be too late. Leaders just have to make reasonable plans and decisions that work on time and this includes the use of prior knowledge and generalisation.
This rational can be applied to finance. The famous Bridgewater and Associates CEO Ray Dalio notes in his book “Principles.” At Principle 5.7 “prioritize by weighing the value of additional information against the cost of not deciding.” Generalisations don’t need to be perfect; they just need to be better than the time it takes to get particular or even perfection and this includes the risk of the costs of not making a decision at all. There will be times when rapid need to be made and that by delaying a decision then an opportunity can be missed or an avoidable loss eventuates. Dalio may attribute this to intuition or deep thinking from his experience. He also remarks that he measures his staff recommendation against their believability and track record, a success bias.
Longer term planning in both finance and the military can allow time for deeper analysis, however in the short term, generalisations and deep knowledge can suffice in the absence of a better alternative.
Heuristics as generalisations in psychology and cyber security
Heuristics are used in both psychology and cyber security, among other fields. Heuristics is an approach to problem solving that is a practical method but is not guaranteed to be optimal. A heuristic is sufficient for reaching an immediate goal in the given time. An example is that if you set your alarm to wake up in the morning, you expect it to work as it has (almost?) always done. If you need to so a simple task, then you may use common sense in the absence of specialised knowledge.
In cyber security malicious threats are constantly evolving. When working as an Information Technology procurement specialist, a cyber security expert said that you can never be always safe, only as safe as you can be at that point in time. He used a helpful analogy of cyber security and contraception. For example, you may use condoms with 96% efficacy, or the contraceptive pill at 98% efficacy, but it is best to use both to prevent a catastrophic cyber security incident.
Cyber security involves heuristics and prevention. If a piece of code looks like a virus, acts like a virus, then it might be a virus and must be isolated pending further review. A catastrophe can never be guaranteed to not occur because cyber threats evolve. So cyber security software uses heuristics that attempt to recognise malicious activity and respond to it, before there is time for a human to make more detailed analysis and respond deliberately.
Generalisations with legal hunches and traditional proverbs
When working as a lawyer, I would be told a story by a client and then asked my opinion with very limited information. The typical lawyer response to any question is “it depends” and “I’ll ask you to make a deposit into my trust account before I devote the time to research and prepare a detailed advice.” I may be able to give a general view, but it won’t be particular until all the issues are examined, potential outcomes weighed against competing interpretations and possibilities, and then factor in the unknown friction of the other side and what a court may decide. Having a general hunch perhaps based on deep thinking and experience or perhaps based on educated guesses on a set of circumstances is useful in forming a preliminary view and then directing research efforts and analysis to answering the issues at hand.
In one case I had a hunch of the likely outcome. I did not know for certain since all the information was not yet available. However, I knew enough facts and law, and was able to provide a preliminary view. This meant generalising and speaking at a high level with the different parties with the intent, and outcome to avert escalation of conflict. This set of skills and experience has been useful in other professional endeavours such as consulting where a client wants a solution yesterday and a general view is needed until more detailed analysis can be performed.
The English proverb “a penny saved is a pound earned” is a finance heuristic. This means that if you save money, then you will be better off in the long run. This does not need detailed analysis since we know it is right often enough and is based on both good tradition and deep thinking through the ages. This can also be tempered with the counter-expression “penny wise and pound foolish” Cautioning a person to not to only make good decisions on small matters but to also make good decisions on larger matters. In the popular personal finance book “Barefoot Investor” Scott Pape implores readers to make better decisions by automating their finances and reducing the risk of losing money. It is far easier to avoid losing money than it is to make money, especially after a loss. Pape advises his audience to avoid debt and build wealth with investing, only after all debts are paid off. Paper general tips rely on generalisations, but going by his audience and case studies these actually work and so assist his readership to avoid ruin. The readership do no need to become financial experts in particular. The audience need to avoid making big mistakes, focus generally on getting the little things right and in the long run be better off.
Generalisations are not bad and don’t always lead to ruin – they can be quite useful In certain circumstances. However in other cases not so helpful such as a deliberate court proceeding or long term planning.
Generalisations in life and in finance are not bad in and of themselves. If the generalisation used is rooted in fact, usually right enough of the time, considering that life and markets are unpredictable and fit the conditions of the situation given the time available, then it can be fine. Generalisations need to be simple enough to be useful and adaptive to the circumstances. When more time is available and there is the opportunity to conduct deep analysis then particularising is beneficial. The essential question I assert is on time. Sure, mis-used generalisations can cause people problems. Like many other things, this is a mental tool for making sense of the world and finance. Generalisations do not necessarily lead to ruin, “it depends.”