Last night I was having a relaxing bath thinking about a complex mathematical problem called the Multi-Armed Bandit Problem. When I say I was thinking about it, let me set the record straight I wasn’t thinking how to solve it as that is way beyond my pay grade and stumped some of the best mathematicians in the world for decades until John Gittin solved the problem in 1979 with his Gittins Index, instead I was thinking how I could use some of the insights to explain certain problems associated with allocating capital and trading.
Indulge me for a minute. Ten days ago I placed a bearish trade on the South African Rand, it almost immediately turned in my favour with the autocratic lunatic President Zuma firing his finance minister sending the Rand into free fall. Now with the biggest winner in my accounts 1yr track record I developed itchy fingers and cashed in. Of course the Rand has continued to fall even further since, leaving me angry with myself for cashing in too quick when I had a much larger take profit price level in mind. Simply I got caught up with the money management emotions instead of keeping my eye on the prize.
The story of my Rand trade left me with a sense of regret. Regret avoidance is one my main principles I live my life by. I have this attitude of I would rather fail trying than live with the regret of not trying. Jeff Bezos describes this nicely when he is asked why he left his cushy job to start an online book selling business – Amazon. Hang in there gentle reader we are getting closer to the point as mentioned in the subject line.
With my regret avoidance mindset I said to myself there must be many PsyQuation traders (and traders not yet connected to PsyQuation ?) who suffer from regret aversion, so how can we build a solution into the PsyQuation alert architecture and help everyone, myself included? I have been thinking about very little else this past week and I believe I may have found a solution.
To begin with there is an algorithm developed in game theory called Regret Minimisation, for the nerdy reading click here. I have not yet shared this with the PsyQuation data science team so be on the lookout for a new PsyQuation Regret Alert in the coming weeks; lets hope I don’t regret this letter ? when they tell me to get lost. Would you believe this introduction has very little to do with the subject of our discussion. This is just the journey that I went on to find the insight that the Gittins Index provided me with.
Here is the question in my own words that highlights the point I want to make:
You have $100,000 of capital to allocate to a trader. You are required to choose one over the other, you are faced with the choice of allocating to a slightly above average performing 5yr trading veteran or a rookie shooting the lights out over 6 months. Who do you allocate to?
For those of you who have read one of my previous letters on Optimal Stopping you will realise that there is a mathematical solution playing in the same sandpit of the problem I am raising which Gittin solved in his 1979 paper called Bandit Processes and Dynamic Allocation Indices.
In conclusion, the simple answer to the question I proposed is you always go with the Rookie. Why this insight gave me so much joy is that it explained the human beings natural inclination to be optimistic without enough supporting evidence. Another reason for my joy is that it provided some mathematical support to our business model of building a platform to facilitate capital allocations to emerging trading talent. For those of you who haven’t; you should check out the Qualifying Series.
However, in reality the solution is far more complex that simply choosing the rookie. The PsyQuation Score™ plays an important role in providing an allocator with an optimal solution to making the correct choice to the question described above as there are far more facts you need to consider before effectively answering the Rookie over the Veteran question.