Finding Alphas. Igor Tulchinsky

been asserted in the world of finance, including option models, only to be shown to be wrong and assumption laden in a meltdown.

      And so you see the paradox: the only rule that works is the rule that no rule always works. Rules are just little specks of dust, pointing their way towards reality, but never quite reaching it.

      This is the very reason why it is of utmost importance to cut losses when riding the changing sea of UnRules. How can this be mastered? What is the correct way to deal with the millions of shifting rules, all of them imperfect and often conflicting, based on different sets of circumstances and assumptions?

      Trading is a microcosm of reality, a dynamic environment of immense complexity in which billions of participants act based on billions of rules and beliefs, which, in turn, affect the environment. The challenge in trading is, of course, to derive the rules that describe the markets, and then use them successfully to earn profits without changing those markets in such a way that the rule itself is destroyed.

      Rules in trading are called alphas, and alphas are really little algorithms for predicting the future of securities returns. Managing millions of rules, where each rule is really a hypothesis, is a complicated matter and a subject in itself. When dealing with millions of rules in the realm of trading, certain regularities become apparent. This brings us back to cutting losses. The best, most universal way of dealing with this complexity (and the fact that all rules eventually break down) is by mastering the long-known trading principles behind cutting losses.

      In the world of trading, the concept of cutting losses has been around for a long time. It originated in what is possibly the oldest type of trading known as trend-following, in which a bet is made that a rising security will keep rising. In such a scenario, trades are typically entered at a new high and exited when accumulated profits drop more than a certain amount from an all-time high.

      In today’s trading world, strategies or alphas are seldom as simple as that. Instead of following a particular security, trend following is applied to the accumulated P&L of the strategy.

      To put it in plain English and in more general terms: cutting losses simply means letting go of rules that no longer work.

      Although the efficacy of cutting losses is very easy to see in the microcosm of trading, its principles hold in many areas of life, including business, entrepreneurship, and relationships.

      Cutting losses requires discipline and subjugation of one’s ego. Typically, any kind of thinking and decision making, including following rules, can be laden with and distracted by emotion. Brain scientists have seen that people with damage to the brain in the emotional center are unable to make simple decisions like deciding what shirt to put on in the morning. In our work using alphas to make trading decisions, the typical state of mind is driven by an emotional state of confidence. When coming up with a particular strategy, the process starts with, “I understand how the world works, I believe in my rule, here is my rule.” Since ego and pride are involved with this confidence, it is hard to let go of the rule one has come up with, even in the face of evidence to the contrary.

      Perhaps it is for ego reasons that the principles of cutting losses are not followed more widely. The other reason is lack of knowledge of alternative rules that might be implemented. The high cost of changing one’s strategy also contributes to resistance to letting go of rules that no longer work.

      It is wise to refrain from believing exclusively in any particular theory or any particular rule. Believe them all. And don’t believe any of them completely. Sometimes they work, sometimes they don’t.

      The best indicator of whether a rule is good is how well it is working at the moment. The rest is speculation. If a rule works, we invest in it; if it doesn’t, we don’t.

      We collect all ideas and let time and performance show what works and what doesn’t, and when.

      A new idea, rule, or alpha is postulated based on history, and through statistical analysis (with sometimes a touch of fundamental wisdom), it goes into our knowledge base.

      From this vast universe of ideas we construct the closest thing possible to a depiction of reality. To do what we do, you have to be comfortable with the fact that you will never know everything there is to know.

      They say in the land of the blind, the one-eyed man is king.

      And when it comes to trading and financial markets, even having one eye is an accomplishment.

      HOW DO WE APPLY THE PRINCIPLE OF THE UNRULE AND OF CUTTING LOSSES?

      We acknowledge that the number of imperfect ideas is unbounded, and reality is unknown and unknowable. Each imperfect idea does succeed in describing reality a little bit. So the more alphas we have, the better we can describe an aspect of reality, and the closer we can come to having “one eye” with which we can increase profits.

      Since no rule is perfect, a combination of ALL rules comes as close to it as one can.

      Applying all rules simultaneously is the key to success. For example, to cross the street, one might have the following rules in mind:

      1. Look left, look right, then left again, then it is safe to cross.

      2. If you hear a loud noise, turn in the direction of the noise.

      3. If you see a car headed towards you, run!

      You may start out crossing the street believing in and comforted by Rule 1 when you hear a horn honking, triggering Rule 2. Rule 1 should be cut immediately because the safety conclusion has been challenged by the noise. Then Rule 3 is applied.

      So we have the following implications:

      ● It is necessary to come up with as many good rules as possible.

      ● No single rule can ever be relied upon completely.

      ● It is necessary to come up with a strategy for using rules simultaneously.

      How do you identify when a strategy is not working? If it performs outside its expected historical returns, signaled when:

      ● drawdown exceeds what’s normal;

      ● its Sharpe falls;

      ● it otherwise goes out of the historical box, defying the rules that were initially observed.

      It is important to pursue different strategies simultaneously, and to shift one’s efforts into working strategies. As a simplified example, suppose one has a theory for describing when gold prices rise. The theory works 50 % of the time (years). Suppose one has 10 such equally good theories. A combination of the theories will describe reality better than any one of them. And the best way to manage which one of them is most accurate is by observing which ones are working now.

      Then comes the application of cutting losses.

      When a strategy stops working, determine the belief that motivated the activity. If the belief was obviously false, you are playing dice here. Best to terminate the activity and engage in more productive ones.

      For example, let’s say you hire someone to renovate your house. They promise to do the job for $50,000, but less than halfway through the job, they’ve already spent $45,000. At this point, if switching to a new builder can be done cheaply, cut the old one.

      Suppose we are engaged in an activity – let’s call it X – which starts to lose money.

      The activity can be anything, perhaps a trading strategy or a business. The questions to be asked are:

      ● Am I losing money in activity X?

      ● What is the loss amount? Call the loss Z.

      Before starting activity X, what was the anticipated amount of the maximum loss? If Z exceeds this amount and the exit cost is not so high, cut the loss.

      SUMMARY

      Examine each potential action prior to embarking on it.

      Determine:

      ● What’s the objective?

      ● What are the normal, expected difficulties?

      Plan