Currency Secrets

If this is the ONLY post you are viewing upon clicking on the Top Currency Secrets category then you either do not have permission or have not logged in to your account.

If you have not registered please do so by reading about it here.

If you have registered please log in with your nickname and password here.

One of the last activities that need be done when system testing is removing the sample constraints. This will allow you to test your system over the entire period of your data to see whether it is still profitable and that the drawdowns aren’t increased.

Here is the resulting code without the sample space constraints: CLICK HERE

And for the last time let’s run a watchlist scan as well as Simulation to see how the numbers have panned out.

Watchlist Rankings = 254 trades
6% risk per trade…
Starting Capital = $10,000
Ending Capital = $77,136.58
# Trades = 241 (13 neglected or 5.1%)
Strike Rate = 43.15%
Avg Win = $2,499.55
Avg Loss = -$1,407.42
Avg W/L Ratio = 1.78
Drawdown = -53.75%
Recovery Factor = 0.98

Unfortunately, as discussed in our previous post, this system has an increase in drawdown, but now the drawdown is WAY outside what we’d be comfortable with.

So, it’s back to the tweaking board.

And that’s pretty much it.

I will not have the time to go into detail with reversal based systems, but with these types of systems you are mainly look for price extremes and then looking to take advantage of price moving back to normal.

Here are some entry conditions that you may want to test:

• Bollinger bands on price burst
• High volatility
• Bollinger bands on range bound indicators
• Bollinger bands on non-range bound indicators
• Reversal indicators - stochastics, RSI, Williams %R
• Reversal price patterns - engulfing patterns, piercing patterns, dojis

Those should be enough to wet your appetite in your design of reversal based systems.

Well I hope this FREE series has helped you understand the process of system designing from an area that really isn’t discussed in much detail anywhere else. I hope that with the knowledge you have gained that it has helped you design your systems better and quicker and more robust than what you previously were accustomed at doing. Even if this series has helped you to see another side to your system that you weren’t previously aware of then I would have achieved my result.

In ending, I have tested many forex systems and many fall short of my goals and targets for the system I personally prefer to trade, and this is the very reason why I trade discretionary, however this does not mean that I do not stop testing new methods and techniques searching for a profitable mechanical system because in the end the best way to trade is automatically (without thinking). In fact it would even be better if the entire process is done automatically… the only thing needed done on my behalf is withdrawing the profits!!

So, dear friends, I wish you all the best in your testing of new ideas and hope that one day you will find that system that complements you perfectly.

Ryan Sheehy
Currency Secrets.com

On the home stretch of this series we will explore money management today. This is by no means the LEAST important of the series - in fact it’s up there as being one of *the* most important.

So what is money management?

Money management answers the question: “How much?“. This not only applies to the quantity of contracts you wish to trade, but also at the ending stats.

First we will begin by looking at how much we are willing to trade.

Okay, as I mentioned before it’s easy to make a system look better or worse than it actually is by modifying your position sizing. The reason why we use a fixed quantity in our initial tests is that we do not want to weight any trade with more contracts than another. If we apply a percentage based position size (whether it be a percentage of capital or a percentage of risk capital) we will be skewing the results in favor to how the system performs at the end of the sample.

So, to ensure that we objectively see how our system is faring in the initial phase of our testing we use equal position sizing.

The important thing now, once we have found our system, is to test our system according to the actual position sizing rules we wish to use with our system: in the case of the systems I implement I prefer to use Maximum Risk Pct.

This style of position sizing ensures that I never lose any more than whatever percentage I am willing to lose. If I am designing an intraday forex system I am generally not willing to risk any more than 1% of capital per trade, if I am designing and end of day forex system I don’t mind expanding this risk to 5% of capital per trade (end of day systems will generally have wider stops, therefore you may want to consider risking more just to make the position worth it).

First thing we need to do is change our code so that it now can accept risk percentage type position sizes, here’s the new code: CLICK HERE

So let’s see what Simulation results we get:
2% Maximum Risk Pct produces…
Starting Capital = $10,000
Ending Capital = $11,547
# Trades = 5 (68 neglected or 93.2%)
Strike Rate = 80%
Avg Profit = $1,549.10
Avg Loss = -$2.10
Avg P/L Ratio = 737.67
Drawdown = -4.35%
Recovery Factor = 3.21

Why should we ignore this data straight away?

Two reasons:

1. It ignores more than 90% of trades; and
2. It has less than 50 trades

Obviously we don’t have enough capital to enter into the trades, therefore we have two options available to us:

1. We can increase our risk percentage; or
2. We can increase our capital

Depending upon what you have you can either test both of the above changes or only one of them. I will assume that you can only increase you risk percentage, here are the next set of results when we do this…

3% Maximum Risk Pct…
Starting Capital = $10,000
Ending Capital = $16,020.70
# Trades = 27 (46 neglected or 63%)
Strike Rate = 48.15%
Avg Profit = $648.50
Avg Loss = -$172.13
Avg P/L Ratio = 3.77
Drawdown = -14.76%
Recovery Factor = 2.93

Still not good enough. Onward.

5% Maximum Risk Pct…
Starting Capital = $10,000
Ending Capital = $28,896.19
# Trades = 65 (8 neglected or 10.96%)
Strike Rate = 44.62%
Avg Profit = $1,096.40
Avg Loss = -$358.32
Avg P/L Ratio = 3.06
Drawdown = -28.07%
Recovery Factor = 2.61

Much better, but now we are still neglecting 10% of trades.

What if we just went one more percent?

6% Maximum Risk Pct…
Starting Capital = $10,000
Ending Capital = $34,583
# Trades = 70 (3 neglected or 4.3%)
Strike Rate = 47.14%
Avg Profit = $1,329.65
Avg Loss = -$521.49
Avg P/L Ratio = 2.55
Drawdown = -33.48%
Recovery Factor = 2.41

Done.

Now that we’ve satisfied our quantity of trades do we satisfy our own risk tolerance (this is the second how much to our money management)?

First thing I do is check the drawdown figure: -33.48%

This is a tad too high for me… I prefer systems that have drawdowns no greater than 25%… at the most 30%.

But if it were acceptable the next set of how much statistics I would look at would be the profit made.

Is the profit that was made enough to meet whatever need I am looking for in the forex market? If I’m looking to have this as a retirement fund can I retire with these returns? If I’m looking for *any* form of additional income is this sufficient?

Only you can answer these questions.

So is it a case of back to the drawing board?

No, not at all.

All we would need do is go back to our tweaking and add/modify our entry/exit conditions until we end up finding the right match.

This may have seemed like a lengthy process, but once you start doing more of the process you will begin to see what can be modified to improve results (etc).

With our next post we will wrap up the process as well provide some tips for those who wish to trade reversal type systems.

It’s easy to carried away with yourself in this aspect of system design… and that is why it’s important to be sure of your direction and what you’re doing when you begin adding or modifying your entry code.

It was the reason why I tried to have you seek out what sort of trader you are.

Knowing this can go a long way in helping you create your system.

Okay, so where do we begin?

There are many different things you can add to your existing entry rules, here is a very broad scope of what types of entry conditions can be added:

• Volatility
• Sentiment
• Trend
• Time
• Volume (obviously not included in the forex market)

Let’s quickly explore in a little bit more depth each of the following areas:

Volatility

Out of all of the different types of entry gates you could possibly add I ALWAYS try and add a volatility gate - again, this is just my preference and is nothing that is set in stone… in my experience I have found ways to improve some of the systems I have created in the past by adding a volatility rule.

Indicators that are available to help your guage and measure volatility include: Standard Deviation, ATR (Average True Range), and HHV-LLV (highest high value minus lowest low value over a set period).

Sentiment

A sentiment indicator is what is also commonly known as a leading indicator and generally this indicator shows areas where price may be overbought or oversold and likely to reverse.

There are quite a few indicators that fall into this category so I’ll refrain from detailing as many as I know, however some of the more popular indicators include:

• RSI (Relative Strength Index) - range bound between 100 and 0 where readings above 70 indicate overbought (and thereby likely for a drop) and below 30 oversold (up move likely)
• Stochastics (D & K) - range bound between 100 and 0 and generally readings above 80 are considered overbought and below 20 oversold
• Williams R - range bound between 100 and 0 where readings below 10 are OVERBOUGHT and above 90 OVERSOLD (note the difference)
• MACD (Moving Average Convergence Divergence) - this indicator is NOT range bound but the histogram of this indicator generally shows turning points
• Momentum (or Rate-Of-Change) - same as MACD with NOT being range bound… combining this indicator with Bollinger Bands can help identify potential overbought and oversold conditions
• CCI (Commodity Channel Index) - NOT range bound, indicator value above +100 indicates price is overbought, below -100 price is oversold

I could continue with other lesser known indicators, but these are enough to give you an idea of what type of sentiment indicators are available.

Trend

These indicators help identify the current trend… the most popular indicators in this family generally use the moving average or some derivative of a moving average.

Time

Using time allows you to delay entry/exit, or possibly include it as a condition whereby you may find a certain day of the month more profitable to trade than another, or certain month(s), or certain year(s)!

Volume

Popular indicators that use volume include OBV (on balance volume) & Accumulation/Distribution… I’ll refrain from discussing these as our forex data does not contain volume and we obviously cannot test something that we do not have!

Alright, now that you have a basic understanding of what classes of indicators are available you can begin adding them one by one into your system and going back through the entire tweaking process.

I must admit, going back through the tweaking process seems like a lot, but when you begin doing this time and time again you’ll begin to instantly zoom through the whole process without batting an eyelid.

So which entry gate do we add?

Well, as discussed above, I love the volatility indicators so I’ll begin by using this indicator first. Considering that we are trading a breakout style system I would use a volatility gate by finding low volatility zones as a breakout from congestion is one of my favored trading strategies. Here’s how I’d begin coding it… CLICK HERE

What results do we get?
Watchlist Rankings = 76 trades
$500 position size…
Starting Capital = $10,000
Ending Capital = $48,582.96
# Trades = 73 (3 neglected or ~4%)
Strike Rate = 45.21%
Avg Profit = $2,382.82
Avg Loss = -$1,001.25
Avg P/L Ratio = 2.38
Drawdown = -61.43%
Recovery Factor = 1.96

Here were our results from the original AtStop system…
Starting Capital = $10,000
Ending Capital = $37,275.36
Number of Trades = 113 (2 neglected)
Strike Rate = 46.02%
Avg Profit = $1,217.12
Avg Loss = -$590.14
Avg P/L Ratio = 2.06
Drawdown = -64.50%
Recovery Factor = 2.36

Do we have any major changes?

Well, several things I like about the new system is the increase in profits (due to increase in position size) with less quantity of trades. An improved Avg P/L ratio, a better drawdown (even though it isn’t to be too highly entered into consideration) and a lower recovery factor (not good).

Will I choose this system over the other?

Yes… I DO prefer less work for more pay, but the changes aren’t anything fantastic.

Okay, what can we add next?

Let’s move to sentiment indicators…

Out of all of the sentiment indicators the ones I dislike the MOST are the ones that are range bound. In the forex market there can be prolonged periods where an overbought or oversold condition stays in effect for a sustained period, by using indicators that are NOT range bound and using boundaries that change with the volatility of the indicator (by incorporating Bollinger Bands ONTO the sentiment indicator) you can more effectively manage your overbought and oversold conditions.

My favored sentiment indicator is the Momentum indicator - it isn’t range bound.

Therefore my new code would now look a little like this: CLICK HERE

Here are the results:
Watchlist Rankings = 52 trades (oooo… getting close to my limit of a MINIMUM of 50 trades for the sample)
$1,200 position sizes…
Starting Capital = $10,000
Ending Capital = $106,908.36 (969.08% net profit!!)
# Trades = 50 (2 neglected or 3.8%)
Strike Rate = 52.00%
Avg Profit = $5,886.78
Avg Loss = -$2,339.50
Avg P/L Ratio = 2.52
Drawdown = -65.22%
Recovery Factor = 2.38

Nice!

But did you notice with this system that we’ve suddenly had a BREAKOUT of results? What I’d quickly test now is the NEW conditions that were placed and delete the previous additions.

Here are the results:
Watchlist Rankings = 73 trades
$1,100 position sizes…
Starting Capital = $10,000
Ending Capital = $93,012.53 (830.13%!)
# Trades = 71 (2 neglected or 2.8%)
Avg Profit = $4,809.86
Avg Loss = -$2,370.35
Avg P/L Ratio = 2.03
Drawdown = -63.46%
Recovery Factor = 2.20

I think this was MUCH better than our previous addition, in fact I LIKE IT and will discard the previous volatility additions.

Onwards.

What if we now added a trend indicator? I personally don’t prefer adding moving averages into my system - especially in breakout type systems such as mine considering that the entry order already incorporates trend into its order price.

I also don’t use any TIME type indicators in my system, but this shouldn’t prevent you from testing it.

So what’s the new system?

Here’s the code: CLICK HERE

And what does it mean in English?

Okay, here goes…

• Plot Momentum(Close, 30) indicator on chart
• Plot Bollinger Bands on the Momentum(Close, 30) indicator with time period 30 and standard deviation 1
• When the Momentum(Close, 30) is lower than the UPPER Bollinger Band place a BuyAtStop order at the Highest High of the last 30 daysplus one pip
• If the Momentum(Close, 30) is higher than the LOWER Bollinger Band place a ShortAtStop order at the Lowest Low of the last 30 days minus one pip
• For successful BUY entry orders place a stop at the lowest low of the last 30 days and trail (i.e. only move the stop loss when the lowest low of the last 30 days is GREATER than the current stop price)
• For successful SELL entry orders place a stop at the highest high of the last 30 days and trail (i.e. only move the stop loss when the highest high of the last 30 days is LESS than the current stop price)

Next we’ll explore the Money Management aspect of system development and see how our system stands up.

Out of all the previous tweaking parts of system design these next two are the toughest: adding exits & modyfing/adding entries.

It is SO easy during the tweaking process to get caught up and whisked away in testing a bazillion exit techniques all at once.

I’m probably not going to be saving you much in the way of time with the technique I’m about to share with you, but I’ll assure you that if you do it you’ll advance far more than normal traders & computer geeks!

What is it?

Throughout all my tweaking years the best additions I have made to my systems are the ones where I have physically traded, or physically scanned through all the previous trades and SEEN what would improve results.

Yes this does take time, yes it would be easier to set up some super script that can formulate every conceivable exit known to man and have your computer process it, but the most important computer is the one that resides in between your two ears (and this should give you a hint as to why I became a discretionary trader!).

So… put away the coding screen, run a simulation of your current system and now go and look through the charts and visualize as many trades as you can over the next 30 minutes (yes, give yourself a time limit because all the best ideas don’t do anything unless you test them).

While you are viewing these trades I want you take note of what happens during the trade. Take note especially of trades that go wrong straight away, as well as trades that do extremely well. How could you have traded better? Was there some telltale sign?

By writing down some general ideas that you have about the system you will begin to formulate better exit strategies.

Conditions I want you to be wary of when you are scanning your exits include:

• Your trailing stop
• Volatility (ATR, Bollinger Bands)
• Other popular indicators (RSI, Momentum, MACD, Stochastics)
• Initial stop loss

So… let’s have a look at some of the things I spotted with this system:

• First thing I noticed was how rare it was for systems to reverse sharply back, I think the system can handle a tighter initial stop loss;
• During large volatile swings I didn’t feel as though the system was consistent enough to add an exit of this calibre in (i.e. some trades would have done well with a high volatility exit whereas others would have done worse with it! So… I’ll leave it alone);
• The trailing stop seemed to be at a reasonable distance and was able to keep you in the position of the main trend. No real change or addition came to mind.

Alright, so how does this translate into code?

Well, the only real change I can see so far for this system is to place a better initial stop loss. Here’s one way you could code it: CLICK HERE

What does it produce?
Watchlist Rankings: 116 trades
$300 position size…
Starting Capital = $10,000
Ending Capital = $37,728.96
# Trades = 114 (2 neglected)
Strike Rate = 45.61%
Avg Profit = $1,217.12
Avg Loss = -$573.57
Avg P/L Ratio = 2.12
Drawdown = -63.87%
Recovery Factor = 2.28

Here were our results from the original AtStop system…
Starting Capital = $10,000
Ending Capital = $37,275.36
Number of Trades = 113 (2 neglected)
Strike Rate = 46.02%
Avg Profit = $1,217.12
Avg Loss = -$590.14
Avg P/L Ratio = 2.06
Drawdown = -64.50%
Recovery Factor = 2.36

Do we have any startling difference? No.

We have improved some numbers (slightly), but not enough to effect a change… and this is important: whenever you ADD stuff you want to make sure that the results have at least a better result in comparison to your previous system’s results. If you are looking for an exact measure then you really want to improve ANY of the following stats by at least 20%:

• Decrease in quantity of trades (if I can do less and still make the same amount that’s great!)
• Increase in profits
• Reduction in drawdown
• Increase in Avg P/L ratio

Why isn’t any good change a step in the right direction?

If you begin adding too many conditions into your system you begin to optimize it too much. It will restrict future options and will make your system too optimized for the sample space you are testing. And one other important thing I should mention: you should have about 50 trades in your testing sample. Some people say 20, others 30, but for back testing purposes I aim high and always try to get at least 50 trades into my simulation.

Too few and you are basing your decision on insufficient data (small sample size).

I personally don’t do too much testing on “Other Popular Indicators”. I’ll leave this option open for those who have fancy indicators to include into this portion, but throughout my life I’ve found that most indicators are useless. Again, this is only through my eyes, others may have differing views.

Next we’ll look at ways in which we can begin adding or modifying our entry rules - out of all the changes this one has the greatest effect on the bottom line (and it’s the reason why so many start at this area FIRST and then work their way down).