ForexSpirit

For the past few weeks I have been busily beavering away on some trading homework tasks directly related to my mentoring relationship with Dirk du Toit.

I have been working on three documents. The first of these documents is my trading plan which I wrote again from scratch. Dirk provides a very nice trading blueprint template that can be filled out and customised to be your very own trading plan.

I originally completed one of these blueprint templates in 2006 shortly after I started demo trading the 4×1 methodology. The blueprint has been updated since then, and now includes some major revisions that were not present in the first version that I received.

I completed the latest version of the blueprint template without even looking at what I had written last year. When I compare the newly completed version to the one I wrote last year it is rather striking how similar my responses are to questions and sections that exist in both versions of the template. I see this as a rather good sign; I have not had to radically overhaul any aspect of my trading in that time.

The second document I completed was an evaluation of my trading using the median grid system to date. I have over twenty months of experience using the median grid trading system, so there was an awful lot of trading to cover during this evaluation process. Part of this evaluation was detailing how my trades worked out. This was going to involve extracting some statistics about my live trading over the past seventeen months (the other three months were spent demo trading). The easiest way to accomplish this was to put together a comprehensive spreadsheet detailing all my trades during that period.

To complete the evaluation document I therefore needed to construct a third document: a trading record.

The remainder of this post is going to shine some light into the murky depths of my trading and expose some of the statistics that I have uncovered. The following statistics have been compiled from my live trade records dating from 6 June, 2006 to 31 October, 2007.

Overview statistics

Let’s start off with some numbers that detail the overall trading landscape that I ventured through.

I had 141 profitable trades during the seventeen months, with 31 losing trades. That means I had 81.98% profitable trades, while 18.02% were losers.

I have made 8571 gross positive pips (from winning trades). Gross negative pips amount to 4996. This results in a net gain of 3575 pips overall.

My average pip gain was 52 pips. My average pip amount on a losing trade was -141.

If I break those same statistics down between the two years involved the following is revealed:

2006 (remember this contains seven months of trades)
Profitable trades: 61
Losing trades: 12
% profitable trades: 83.56%
% losing trades: 16.44%
Average pip gain: 43
Average pip loss: -152

2007 (this contains ten months of trading so far)
Profitable trades: 80
Losing trades: 19
% profitable trades: 80.81%
% losing trades: 19.19%
Average pip gain: 58
Average pip loss: -133

Drawdown
Drawdown is an extremely important statistic to keep on eye on. Drawdown measures the decline from an historical peak in some variable. In this case, the variable being measured is my trading capital. Drawdown is an indicator of risk which is why it is important to measure it. The greater the drawdown the riskier the trading method or system.

My maximum drawdown was -12.51%. The longest drawdown period I experienced lasted for 91 days (this is the longest time it took to reach a new equity watermark high).

Let’s once again breakdown the drawdown statistics for 2006 and 2007 so far:

2006
Maximum drawdown: -8.69%
Maximum drawdown time: 59 days

2007
Maximum drawdown: -12.15%
Maximum drawdown time: 91 days

Standard measures
There are some well-known financial measurements that I would like to cover next. These include CAGR%, the Sharpe Ratio and MAR ratio.

CAGR%
CAGR% stands for Compounded Annual Growth Rate and is also known as the geometric average return. It reflects the rate of growth that when compounded equally over the specific period would have resulted in identical ending equity. A weakness in this measurement is that it can be greatly affected by a single period of unusually high or low returns.

My CAGR% is currently 60.64%.

Sharpe ratio
The Sharpe ratio is an excellent measure of risk/reward in comparing stock portfolio management strategies. It is not an entirely relevant measure for something such as forex trading. The Sharpe ratio was specifically designed as a measure for comparing the performance of mutual funds. With that severe limitation in mind I still calculated the Sharpe ratio of my performance as it is an almost universal metric when calculating trading statistics. Just do not forget that using a Sharpe ratio to compare between anything other than mutual funds is all but pointless.

Note: in my Sharpe ratio calculation I am not including the risk-free rate of return value as this value changes with time and a different risk-free instrument can be chosen by everyone. So it just complicates even more the ability to use this ratio meaningfully outside of this explicit context. As long as I am consistent in not using the risk-free rate of return value in future calculations of the Sharpe ratio then I will not skew attempts when comparing past, present and future calculations of this value.

My Sharpe ratio is currently 1.69.

MAR ratio
The MAR ratio was developed by Managed Accounts Reports, LLC, and is normally used to report on the performance of hedge funds. The MAR ratio divides the annual return by the largest drawdown. It serves as a quick and dirty risk/reward measurement. The same measurement seems to be named the SOL Quotient by Rob Booker.

My MAR ratio/SOL Quotient is 6.44.

2006
Annual return: 14.05%
Sharpe ratio: 1.81
MAR ratio: 3.16

2007
Annual return: 126.26%
Sharpe ratio: 3.39
MAR ratio: 13.40

Robust measures
I’m now going to cover a group of measurements that I came across in Way of the Turtle by Curtis Faith. They all aim to eliminate the weaknesses inherent in the previous three measurements, which can all be impacted terribly by the start and end dates of the trade data used in their calculation. In effect, the measurements for CAGR% and the Sharpe ratio can often be skewed substantially by even small changes in the start and end points of the data set.

The real culprit is CAGR%, as this measurement is used in both the Sharpe ratio and the MAR ratio. The reason CAGR% is sensitive to changes in start and end dates is that it represents the slope of the smooth line that goes from the start of the test to the end of the test period on a logarithmic graph. If you change the start and end points then the slope of that line can be dramatically altered.

Regressed Annual Return (RAR%)
A better measure of the slope is a simple linear regression of all the points in that line. A linear regression line is otherwise known as a best fit line. Using a linear regression on my equity returns results in what Curtis Faith calls the regressed annual return, or RAR%. This measurement is much less sensitive to changes in the data at either end of the test period.

My RAR% is 79.76%.

Robust Sharpe ratio
The Robust Sharpe ratio is calculated in the same way as the original Sharpe ratio but for one small change: the CAGR% value in the original calculation is substituted for the newly calculated RAR% value instead. This makes the Robust Sharpe ratio much less sensitive to changes in the data set for the same reason that the RAR% measurement is less sensitive than CAGR%.

My Robust Sharpe ratio is 2.22.

R-Cubed
The final new robust measurement is called the robust risk/reward ratio (RRRR). It is a more robust version of the original MAR ratio. R-cubed uses RAR% in the numerator and a new measure that Curtis Faith calls the length-adjusted average maximum drawdown in the denominator (phew!).

Let’s take a look at this new denominator in more detail. It is composed of two components: the average maximum drawdown and the average drawdown period.

Curtis Faith recommends calculating the averages for both the drawdown amount and period by using five values and dividing by five. I do not have enough medium to large drawdown periods in my trade history to make that viable, so I have used three instead.

I calculated the average drawdown amount by taking my three largest drawdowns for the period being calculated and averaging them.

I did the same thing for the drawdown period: I took the three longest drawdown periods and averaged them. You then take this drawdown period (which is in days) and divide it by 365 so as to express it as a ratio.

The final calculation looks like this: RAR% / ( (average drawdown) x (average drawdown period / 365) )

The R-cubed measure is sensitive to the addition or removal of large drawdowns but less sensitive than is the MAR ratio.

My R-cubed is 39.97.

2006
RAR%: 18.69%
Robust Sharpe ratio: 2.40
R-cubed: 20.85

2007
RAR%: 65.42%
Robust Sharpe ratio: 1.75
R-cubed: 55.15

Making sense of the measurements
These measurements, both the standard and robust ones, are not that interesting or relevant in and of themselves. They are best used when compared to something. Since I don’t have statistics for anyone else to compare myself with, I am left with comparing against myself. This means that I will be comparing my performance over time.

Am I getting better or worse?

I’ll be able to track the things that are having the most impact on my trading performance. Are my returns getting better? Are my drawdowns improved in both size and length?

It’ll probably take another couple of years of data to be able to have a meaningful history to wade through but in the mean time I’ll be able to keep track of things as they occur.

Even more statistics
As well as all these standard and robust measurements I have also calculated various statistics that directly pertain to the grid system that I use. I have broken down which grid each trade entry was made in and what my resulting profit or loss was for that trade.

Using this data I’m able to get a snapshot of where the majority of entries have been made. I’m also able to see from which quadrant most of my losing trades originate, as well as my small, medium and large profitable trades. It’s something that I’ve been curious about and have been meaning to undertake for quite a while, so I’m glad that I actually got around to doing it and now have some hard numbers to look at.

I think that’s enough statistics for now!