I know it’s geeky, but I love to collect data.
I keep spreadsheet journals for each of my trading systems … and I keep data for any tweaks or amendments I’m looking at making, so I can judge performance before I try anything out with real money.
I also do this for any systems I’m developing, or just for trading ideas I have.
The result is a lot of spreadsheets, a lot of numbers … and sometimes I worry about the amount of time I spend poring over these figures trying to find the ‘perfect’ trading method.
That’s the problem with us data geeks. We get tempted into thinking that if we collect enough data, we can filter it down into some simple, perfect equation.
I want to tell you two stories today, about what we can learn from collecting data, both of which can give some real insight into how that data can be used successfully in live future market conditions.
The perfect woman
Back in 1940s America, a gynecologist called Robert Dickinson and a sculptor called Abram Belski collaborated to create two statues – Normman and Norma. These were Mr and Mrs Average – based on the average measurements of 15,000 young men and women.
The country became quite obsessed with Norma … an image of the perfect woman, and a competition was launched to find the real ‘Norma’ – a woman who perfectly matched the average.
Of the 4,000 women who entered the competition, not one woman came close to average on the nine different measurements taken. In fact, just 40 women came in the average range for just five of the nine dimensions.
The average woman did not exist.
The reaction of many doctors and scientists to this revelation was to deduce that post-war American women were out of shape and needed to exercise more.
Just a few years later, the American military ran into the same issue
In a bid to make the cockpits of their planes more ergonomic and safer for pilots, the US Air Force undertook a study to ensure that the cockpits were built to ‘fit’ their pilots. So, they tasked young researcher Lt Gilbert Samuels with measuring 10 key body dimensions across 4,000 airmen, to find the average, so they could build their cockpit to fit.
Daniels generously defined ‘average’ as falling in the mid 30% range of dimensions, but when looking through his stats, not one single man fitted in the average range for all 10 dimensions. In fact, only 3.5% of pilots fell in the average range for just three of the 10 dimensions.
Again, the average did not exist.
In fact, the ‘nearly average’ didn’t even exist.
It sounds like the same story again … but this is what’s key …
The way the Air Force responded to this data was completely different. Instead of expecting airmen to conform to the average, they designed their cockpits to fit a wide range of differently sized men – straps, seats, pedals were made adjustable, and crash rates declined.
It was a success.
How we misuse data
In the second example above, the methods were adapted to fit the outside world … while in the first example, they tried to change the data to match their methods (or ‘rule’).
It seems clear that (no matter how much data we collect in the creation of our perfect trading rules), we’re never going to be able to tell the markets what they should do. Prices will just go ahead and do their own thing.
(Although, it’s not unheard of that traders – I plead guilty myself – scream at their charts that the price MUST go down … SHOULD bounce of this level …. But the markets are deaf to us!)
It’s vital that we keep data and listen to it … but we mustn’t expect future trades to perfectly fit that data.
Rather than just looking at the profit that’s averaged out at the end of our data collection, we need to look at the ranges, and the extremes.
There are no ‘normal’ market conditions, and no ‘normal’ trades. Trading is all about adapting.
So, how helpful our trading rules if no market conditions will actually fit them?
I have two pet hates in trading systems, we would seem to contradict each other …
- I hate it when people aren’t clear about their trading rules and keep shifting the guidelines (normally to back-fit better profits!)
- And, I hate fully automated systems that have rules and trade without human intervention.
I know, on the one hand I’m saying that we should be trading like robots … and on the other hand, I’m saying that we shouldn’t allow robots to trade for us!
But there is a logic (even if it is a fuzzy one) in my madness.
The reason I like clear trading rules is because they keep us out of bad habits and ensure we’re disciplined. They stop us cheating ourselves (and others) about what our results would have been, if only we’d made some subtle adjustment to our rules as the crucial moment …
However, the reality is that strict rules need to fit with the individual … we all have different risk tolerances, trading funds, time constraints, and goals, so the one-size-fits-all trading system can be too much of a compromise.
It’s about finding a balance – something that’s rules-based enough to keep us disciplined, but with enough flexibility to adapt to the fact that we are all individuals, and market conditions are never ‘average’.