We needed more trade data, so Nick taught himself to code. In other words, computer programming to test trade conditions. He has become extremely good at it.
Okay, he started from a high baseline. A spreadsheet wizard—and a master’s degree in engineering—he has some mathematical skills!
As a trader himself, he already knows what’s practical. We both spent a couple of weeks of the festive break poring hours into coded trade development.
What I’m talking about here are algorithms that give trade entry and exit conditions.
Did it work?
The surprise was what didn’t work—off-the-shelf indicators for one. We tweaked and adjusted the conditions and continually run the strategy tester. But the results were poor.
That was when we knew that we needed to develop original code, based on an edge.
We went with a (structured and systematic) trial that tested intraday price conditions against the previous day’s daily price levels.
With a trend-criterion agreed it then took dozens of pages of code to run the what-if conditions. Some worked, many didn’t.
Each test looked at the past two years of hourly price movement. We studied the results of net profit, trades closed (total), per cent profitable, profit factor, maximum drawdown, the average per cent per entry and the average number of bars in a trade.
Without a similar test of strategy—carefully coded and adjusted trade parameters—how would anyone know what they’re doing works?
In our testing, each change or addition to a strategy often improved some figures—or increased drawdown slightly —but significantly reduced the number of trades taken and, therefore, the overall profit.
To make a small correction in condition improves one trade but taken over multiple entries is often a loss-maker.
The trial confirmed that the change in profit factor (reward/risk) has more effect on profit (per improvement condition) than an increase in the percentage of wins.
However, our limitation with intraday is the end-of-day close. We elected not to take a trade overnight as brokers’ do not provide an automated exit condition that met our needs.
A trailing stop was not profitable, and other solutions increased the maximum drawdown unacceptably.
However, we achieved some 60% profitability and with wins being twice that of loses, and with acceptable drawdown. We were jolly pleased too.