Measuring Your Trading System's Edge

I first read about the Edge Ratio (E-Ratio) in the book 'Way of the Turtle' by Curtis M. Faith, it remains one of my favourite books on trend following to this day. The concept of the Edge Ratio is quantifying the statistical edge of your trading system by measuring the market's price movement.

Measuring the edge of your trading system's entry,

The edge that the trading system's entry provides can be measured by seeing how the price behaves over a certain period of time, starting from the moment the trade is taken until a fixed point in time in the future. This involves breaking the price movement into two parts - the good (or favourable) part and the bad (or unfavourable) part. In Way of the Turtle, Mr Faith calls these good and bad price movements the maximum favourable excursion (MFE) and the maximum adverse excursion (MAE). The maximum favourable excursion (the good part of the price movement) is simply the maximum number of pips the price moves in favour of your trade from the moment you entered it until a fixed point in the future. Likewise, the maximum adverse excursion (the bad part of the price movement) is simply the maximum number of pips the price moves against your trade from the moment you entered it until a fixed point in the future. The Edge Ratio is therefore the maximum favourable excursion divided by the maximum adverse excursion. An Edge Ratio of greater than 1 indicates that your entry signal has positive expectancy and an Edge Ratio of less than 1 indicates that your entry signal has a negative expectancy. We would expect a truly random entry to have an Edge Ratio of around exactly 1.

When measuring an entry's Edge Ratio it is important to pick a suitable time frame to measure the maximum favourable and adverse excursions over. For example, say we were trying to test a 20-day Donchian channel breakout and we found that from the moment of the breakout until 5 days in the future there was a negative expectancy, but from the moment of the breakout until 50 days in the future the breakout had a positive expectancy. This would tell us that within the first 5 days of a breakout there was more chance of the price moving against the breakout than with it, but over a period of 50 days the price was more likely than not to move in the direction of the breakout. So which time frame should we use, the 5 day time frame or the 50 day time frame? The answer is that it depends upon the trading system and that you should use a time frame that is similar to the trading system's time frame. In the case of the 20-day Donchian channel breakout, measuring the edge of such a trading system's entry with a time frame of 50 days would certainly be preferable to 5 days as 20-day Donchian channel systems are medium to long term trading systems where trades are far more likely to last around 50 days than they are to last around 5 days.


I am going to test the following common entry strategies on the EUR/USD currency pair and measure the entry's edge over periods of 5, 10, 20, 40 and 80 days. These systems are -

1. Entering long when the price closes higher than it has ever closed in x number of days; and entering short when the price closes lower than it has ever closed in x number of days. A modification of a typical Donchian channel breakout.
2. Entering long as soon as the price breaks higher than it's x number of days high; and entering short as soon as the price breaks lower than it's x number of days low. A typical Donchian channel breakout.
3. Entering long when the smaller simple moving average crosses above the bigger simple moving average; and entering short when the smaller simple moving average crosses below the bigger simple moving average. A typical duel moving average crossover.
4. Entering long at the close of each day whilst the smaller simple moving average remains above the bigger simple moving average; and entering short at the close of each day whilst the smaller simple moving average remains below the bigger simple moving average. A modification of a typical duel moving average crossover.

1. Entering on a new high or low closing price,

Time-Based Exit	New 5 Day High Or Low Closing Price	New 10 Day High Or Low Closing Price	New 20 Day High Or Low Closing Price	New 40 Day High Or Low Closing Price	New 80 Day High Or Low Closing Price
5 Days	1.06 to 1	1.08 to 1	1.07 to 1	1.09 to 1	1.19 to 1
10 Days	1.06 to 1	1.07 to 1	1.10 to 1	1.12 to 1	1.27 to 1
20 Days	1.11 to 1	1.13 to 1	1.19 to 1	1.21 to 1	1.40 to 1
40 Days	1.12 to 1	1.13 to 1	1.20 to 1	1.24 to 1	1.42 to 1
80 Days	1.07 to 1	1.08 to 1	1.13 to 1	1.17 to 1	1.29 to 1

2. Entering immediately on a new high or low,

Time-Based Exit	A New 5 Day High Or Low	A New 10 Day High Or Low	A New 20 Day High Or Low	A New 40 Day High Or Low	A New 80 Day High Or Low
5 Days	1.04 to 1	1.04 to 1	1.03 to 1	1.05 to 1	1.12 to 1
10 Days	1.04 to 1	1.04 to 1	1.06 to 1	1.06 to 1	1.17 to 1
20 Days	1.13 to 1	1.15 to 1	1.18 to 1	1.19 to 1	1.37 to 1
40 Days	1.12 to 1	1.14 to 1	1.20 to 1	1.22 to 1	1.40 to 1
80 Days	1.08 to 1	1.08 to 1	1.13 to 1	1.15 to 1	1.28 to 1

3. Entering on a duel simple moving average crossover,

Time-Based Exit	3 Day SMA Crossing a 5 Day SMA	5 Day SMA Crossing a 10 Day SMA	10 Day SMA Crossing a 20 Day SMA	20 Day SMA Crossing a 40 Day SMA	40 Day SMA Crossing a 80 Day SMA
5 Days	0.98 to 1	1.08 to 1	0.95 to 1	1.18 to 1	0.95 to 1
10 Days	1.00 to 1	1.01 to 1	1.00 to 1	1.27 to 1	0.83 to 1
20 Days	1.00 to 1	1.03 to 1	1.05 to 1	1.26 to 1	0.82 to 1
40 Days	1.00 to 1	1.03 to 1	1.03 to 1	1.26 to 1	0.87 to 1
80 Days	1.00 to 1	1.02 to 1	1.03 to 1	1.13 to 1	1.05 to 1

4. Entering every day in the direction of the shorter simple moving average,

Time-Based Exit	3 Day SMA Above Or Below a 5 Day SMA	5 Day SMA Above Or Below a 10 Day SMA	10 Day SMA Above Or Below a 20 Day SMA	20 Day SMA Above Or Below a 40 Day SMA	40 Day SMA Above Or Below a 80 Day SMA
5 Days	1.05 to 1	1.03 to 1	1.04 to 1	1.05 to 1	1.01 to 1
10 Days	1.05 to 1	1.03 to 1	1.09 to 1	1.07 to 1	1.04 to 1
20 Days	1.07 to 1	1.10 to 1	1.08 to 1	1.07 to 1	1.09 to 1
40 Days	1.06 to 1	1.10 to 1	1.07 to 1	1.04 to 1	1.15 to 1
80 Days	1.03 to 1	1.07 to 1	1.03 to 1	1.01 to 1	1.17 to 1

Out of the four entries I tested, the 80-day modified Donchian channel entry where we enter in the direction of a new high or low closing price worked the best, followed closely by the standard Donchian channel breakout its self. I suspect that the reason why the modified Donchian channel entry worked slightly better than the standard Donchian channel entry is that using the closing price instead of the extreme price tends to help avoid false breakouts.

Using a duel simple moving average crossover as an entry didn't work very well at all except on one time frame (the medium term time frame with the 20-day SMA crossing a 40-day SMA). I suspect that part of the reason for this is that systems that use simple moving average crossovers for entry signals often don't produce entry signals until the new trend is already well under way and often nearly finished. In short, trading price action is often better than technical trading indicators, which are merely (often useful) price distortions.

And, as I expected, the modification to the duel moving average crossover where we entered every day in the direction of the shorter of the two moving averages produced a slightly positive edge on all time frames. I expected this to produce a positive edge because entries would more likely than not be in the direction of the over all trend, and I also expected this edge to be small as we were entering everyday rather than on much more significant event such as a new high or low price.

Measuring the edge of your trading system's exit,

Measuring an exit's edge is a little more complicated than measuring an entry's edge. An exit cannot be entirely separated from the entry that caused the trade to be opened in the first place, and, unlike with the trading system's entry, the market price movement that happens after the exit is irrelevant - we are only interested in what happens after we have opened a trade until the time we close it, what happens after the trade is closed is of no consequence to us.

The entry that produced the most significant edge was the modification of the typical 80-day Donchian channel entry where we used the closing price as an entry signal. This system had a significant edge when used with a 20-day time based exit. I am going to compare the results of entering after a new 80-day high or low closing price and holding the position for exactly 20-days with entering after a new 80-day high or low closing price and exiting sometime between 20 and 40 days when certain exit conditions have been met. These exit conditions are –

1. 1. The trade has been open for at least 20 days and less than 40 days and the market's closing price moves and closes at least 1 percent in our favour over 3 days. If this condition isn't met between days 20 and 39 then we will close the trade anyway on the 40th day after we entered it.
2. 2. The trade has been open for at least 20 days and less than 40 days and the market makes a new 3-day low closing price when we are short or new 3-day high closing price when were are long. If this condition isn't met between days 20 and 39 then we will close the trade anyway on the 40th day after we entered it.

My conclusion,

When we compare the results of exiting immediately after 20 days, with the results of waiting for 20 days, and then waiting for the market to make another movement in our favour before finally exiting, it does seem to show that an exit can also make use of a statistical edge in technical trading systems. However to those traders who say that 'entries don't matter, only exits do' I would ask them to look carefully at the results of using a time based exit as it seems to me that the statistical edge than can be harnessed from an entry is far greater than the statistical edge than can be harnessed from an exit, and, even if this weren't the case, any edge that you can add to your trading system matters. Clearly a trading system's entry matters.

• DeltaStock
• DeltaStock - Open a Live Account

• Home
• Technical Trading Systems
• Trading Articles
• Misc Articles
• FX Brokers
• Disclaimer
• Links
• Contact

• Forex Trading Systems
• Index Trading Systems
• Commodity Trading Systems
• Forex Trading Robots

• The characteristics of the best Forex trading systems
• Trading Strategies that work

• Curtis Faith