How Do I Test ... ?

I write this post in a particularly good mood, having grown the account by 20% this week.

Admittedly, the markets have been on the move - but then that is what my trading system is designed to capitalize on - so I reckon I've earned the right to profit.

But this seems like insane growth, right ? I mean, if you can profit this much in a week, you can lose as much or more, right ?

... Well yes, this is absolutely, true - so what gives anyone the confidence, or faith, to trade this way ?

The answer is simple. Backtesting. The painstaking analysis of years worth of price data; hours and hours at your tradestation, building spreadsheets, analyzing them, tinkering with the variables of your system, then starting the process again and again until you are sure your system is optimized to its potential.

In my experience, this is the level of commitment it takes to trade this way.

... there is no short cut.

So, how do I do it ?

Well the whole process starts (and ends) in psychology. You have to understand yourself well enough to nail down what your appetite for risk is, how you want enter trades and how you will react once they are placed.

For me, my risk appetite is relatively high, but I don't have the time or inclination to chart watch. Indeed, when I do I have an annoying (and destructive) habit of attempting to second guess the price action, which risks derailing my own system if I let it.

These observations led me to conclude that I needed to develop a high probability, 'fire and forget' system, where trades can be entered at a specific time in the day - when the markets are quiet - and then be left to trigger and thence either mature at my profit target rapidly or blow up.

I absolutely don't want to follow them once they are in play. That is the discipline.

With this established I open up my testing software and download the most recent 3 years of data for the currency pair I have chosen. I simultaneously open my backtesting spreadsheet in Excel.

The Testing Process

What I am aiming to do with my spreadsheet is record, trade by trade, all the variables which might conceivably influence the probability of a particular trade maturing profitably.

For me this includes the following: trade type; time, day and date of trigger; time day and date of expiry; duration; maximum drawdown in-trade; profit in pips and $ and what the underlying market conditions were when the trade fired.

With respect to my testing software I bring up any indicators I have chosen to guide my trade decisions and start advancing the candles bar by bar.

Here I will only add that I am looking for a near mechanical trigger. I want the absolute minimum of discretionary input.

As trades fire and complete I go back to the spreadsheet, record the variables, then make notes on the chart screen and take a screen-shot. I basically record as much as data as I can think of.

I am not, on this first run through, interested in money management, or testing myself as to whether I would or wouldn't put at trade on when faced with a particular candle or condition.

I simply want to see what trades the system will generate and how they turn out.

When this first test run is complete I go back to Excel to slice and dice my data, looking for any artifacts or trends which merit a tweak in my entry criteria.

Only then do I go back for a second run, applying money management criteria based on probability of trade success.

Its worth pointing out that this stage generally warrants multiple iterations as I seek to match and optimize both probability and profitability.

Finally - and only when I think I have something special - do I commission the construction of an Expert Adviser (EA) that can be used to automatically backtest a far larger data sample.

This sounds exhaustive and it is, but at the end I have a system that I am truly ready to test and trade live...

... and absolutely crucially, the confidence to trust it.