Trading System Explanation Complete

So, we're all done.

I haven't given you a trading system - I've explained what a trading system needs to be complete.

Hopefully I also gave you some ideas for some rules to experiment with.

Hopefully if you purchase a trading system from somewhere you can now validate that it's a "complete" system.

Hopefully if you read something and profoundly disagree, you'll post a comment so that I can learn further.

And hopefully, when my software becomes available, you'll download a trial version and give it a shot :-) Fingers crossed for early December.

Equity Curve Analysis

Browse through the Collective2 website and have a look at the differing equity curve charts.

When your equity curve is looking fantastic - starts in bottom left and ends in top right - there is nothing to think about. Keep doing what you're doing.

But, during drawdown time, what to do? All systems have drawdown times. But is it just a drawdown, or has your system stopped working?

I gave the example of this chart on that Collective2 site a few posts ago. That system gained 500% over just a few months. Fantastic. And a few months after that is now in 100% loss territory.

The question is, if you were the person using that system, when do you stop using it? When it fell back to a 400% gain? When it fell back to a 300% gain? 200%? 100%? At zero?

So here again, we need some clearly defined, objective rules. It's not good enough to erase the "I'll just hold on for a bit longer - it'll come back" way of thinking from the trade exit, if you're going to be doing the same thing here.

In fact I think it's more crucial here than anywhere to be clear about when to stop using the system.

Now, there is this thing called Profit Factor. Here's what I said about Profit Factor previously:

"Profit Factor is (Av Win / Av Loss) * (Pct Winners / Pct Losers). If it equals 1, the system has neither lost nor made money. If it's less than one then the system is losing money, more than one and the system is making money. The bigger the number, the "better" it is. But this calculation falls down in that it doesn't take into account the total number of trades, or the total amount won/lost.

(5/4) * (55/45) equals (500000/400000) * (55/45).

It does.

Really."

Anyway, you could use the Profit Factor of the last 5 or 10, or however many, trades to determine if you will continue to use your system. Once the Profit Factor falls below your set level, you continue to track the trades, but do not actually trade them. When (if) the Profit Factor comes back above that level, you can then start making real trades again based on that system.

I dare say that poor "500% gain to 100% loss" chap could have benefited from something like this.

On the other hand, here's something I also said about this method in a previous post:

"...just because the last 5 trades have a Profit Factor of 2, doesn't mean the next trade is going to be any good. In fact, a scenario of events could take place where this rule keeps you out of the good trades and let's you get in on all the bad trades."

In the scenario I'm talking about, you have some bad trades, so the Profit Factor falls below the set level. Then, while you are paper trading, you get some good trades, so the Profit Factor comes back up. You now start real trading again, but have some bad trades. The Profit Factor falls below the set level. Then, while you are paper trading, ... repeat.

So I think this Profit Factor idea is not bad, but it's not perfect. Is there anyone, anyone at all, reading this, who might share some thoughts anonymously or otherwise? Please click the comments link and share.

When Not To Trade

There are some times when you don't want to be in the market. At all. Even if all the signs say "GET IN!". And normally these times are around announcements. "Major" announcements.

When major announcements happen, the price quite often spikes. In fact, it quite often spikes in one direction, then the other, then reverts back to where it was before all that happened and continues on as if nothing ever happened.

Only something did happen - one of the spikes took out your stop. Very frustrating, especially in the times when the price does revert back its previous price and course.

Some people trade the announcements by putting entry orders on either side of the current price action just before the announcement, and probably cancelling the order that doesn't get taken up. And probably they do well out of it. But for me, no broker guarantees your order or your stop loss around announcement time. Slippage is a big risk factor.

Anyway, I don't get into it, so I need rules to tell me what to do when this is about to occur and what to do about it.

First, define "major". Not all announcements move the markets. There are some economic calendars you can use for this. There's the Forex Factory one, which specifies whether an announcement is of the "Volatility Expected", "Volatility Possible" or "Volatility Unlikely" variety. Choose which ones you want to avoid and do it. There are other calendars about also.

Other times to avoid trading might be around public or national holidays, weekends, Christmas, Easter, Thanksgiving in the US. Because we always want to be able to get in and out of a trade when we want to get in and out, liquidity is important. And liquidity might be affected around these days. Might be. Up to you.

Once you've defined the times you don't want to be trading, now you have to answer a couple of other questions.

If you don't have any trades open, it's simple - just don't open any new trades in these times. But you should specify, exactly, when these "holidays" start and end. Is it 2 hours before an announcement where volatility is expected, or is it 1 hour? Or 5 minutes?

Does your trading systems Christmas holiday start when the clock ticks over to Christmas day, or does it start at 1pm on December 23rd? Does it end when the clock ticks over to December 26? Or does it end at 10am on January 4th when the New Years festivities have also died down? This stuffs not hard, just write it out and follow it.

Further, if you have an open trade, what do you do with it? Close it? Move the stop loss to lock in as much profit as possible? Leave it alone if you've already locked in a certain amount of profit? Write it down. Follow it. Easy.

Money Management

Money Management rules tell you exactly how much money to risk on each trade. They also tell you when to add more capital to your trading bank.

In short, these money management rules keep you trading even if your trading rules stop working (or didn't work in the first place).

For example, let's say your trading system used to make a profit 80% of the time. So, because of the great strike-rate, you start using more and more money in each trade. You decide to use a fixed amount of $10K on each trade, even though you only have a $50K trading bank.

Well, 5 losers in a row and you are dead in the water. Even systems with 80% strike-rates can have 5 losers in a row. Very quickly. That's not good money management.

So let's change it and use 20% of your trading bank on each trade. First losing trade you risk and lose 20% of $50K, which is $10K. Now you have $40K.

Second trade you risk and lose 20% of $40K, which is $8K. Now you have $32K.

Third trade you risk and lose 20% of $32K, which $6400. Now you have $25,600.

Fourth trade you risk and lose 20% of $25,600, which is $5120. Now you have $20,480.

Fifth trade you risk and lose 20% of $20,480, which is $4096. Now you have $16,384.

So, you can see that using a fixed amount of money lost us everything in 5 trades. Using a percentage of your trading bank adjusts itself each time, so that 5 losses in a row still leaves us with money.

And there aren't too many people that suggest you risk 20%. Most people seem to say between 1% and 3% of your trading bank is all you should have at risk on any one trade. Even if you had a string of losers, risking just 3% of your trading bank is going to keep you trading for a long, long time.

The downside, of course, is that by risking so little, your profits aren't going to be so big. Frustrating if you're just starting out. But let's not forget the oft-quoted 80% (or 90%, or 95%, whatever) of traders who quit or lose their money. Perhaps, expecting only to profit, they risked too much of their bank.

So that's Money Management. Do you use a fixed amount of money on each trade? Do you use a percentage? Easy.

There's some other stuff about when to add money to your trading capital. I remember "pyramids" and other things in a book I read by Daryl Guppy (his books started me off on the technical analysis journey). For me, I just use 2% of my trading bank, and the trading bank fluctuates as I make and lose money. Essentially it means I add the money to the bank straight away.

For currency trades, the calculation is:
A. Get total trading bank.
B. Get 2% of A. This is how much money we want to risk.
C. Get pips at risk in the trade to be opened.
D. Get $ per pip of the currency of the trade ($10/pip for currencies ending in "USD", variable for others).
E. Multiply C with D. This gives us how many dollars would be at risk if we traded 1 contract (1 contract = 100,000 units).
F. Divide B by E. By dividing the amount of money we have to risk, by the amount of money risked if we traded 1 contract, we can get the number of contracts that we are able to trade.

We now have a figure like 0.869 or 1.342. It's up to you and your broker what you do now. You should always round down so that you are actually risking less than the maximum amount you said you wanted to risk.

If you use Oanda, then you can actually trade 0.869 or 1.342 contracts, which equate to 86,900 and 134,200 units respectively.

If you use FXCM's mini account, you need to round it to the nearest lot of 10,000 - 80,000 and 130,000 units.

If you use a broker that allows only whole-numbers of contracts, then 0.869 means you can't trade until you get more money from somewhere, or you find a trade risking less pips. 1.342 means you can trade with 1 contract.

Geez I hope these numbers are right... :-)

Nearly Done

So now your system has:

• Entry Rules & Value - specify when and where to enter.


• Initial Stop Value - specifies where to place the initial stop loss.


• Initial Take Profit Value - where to place the initial take profit (if you want to use a take profit).


• Trade Management Rules & Value - when and where to move the stop loss. (I include exiting at market as a Trade Management Value).

You can also have Take Profit Rules and Values, if you use a take profit and you want to move it from its original position.

Anyway, that's what you need for a trading system. Pretty much. Not really rocket science. On the other hand, far better than buying because the newspaper said so and having no defined point of exit.

But there's a couple of bits left. One is Money Management - how much money to risk on a particular trade.

There is also When Not To Trade Rules, which I refer to as holidays. Even if your Entry Rules are signalling a trade, there might be specific times of the day, week, month, or year that you shouldn't get in.

And then there's equity curve analysis - is this just a normal drawdown the system is having, as all systems do, or has the system stopped working altogether? If it's still there, have a look at this chart on the Collective2 website. Gained 500% in a couple of months. But now? Is that a drawdown or failure? Stop using the system and bank the 100%-odd profit, or stick with it? Get this right and it won't really matter what rules you use for everything else.

[Edit: That system had a 100% profit at time of writing the article. As of now (26-Oct-05) it has gone to a roughly 50% loss.]

Trade Management Rules & Values

As I said in a previous post, to obtain profit from a trade, you either have a take profit set, or you gradually move your stop loss to a position of break-even (if you want), and then to positions above your initial entry.

When you are relying on the stop loss being moved above the initial entry point, that's where Trade Management Rules come in. As with everything else, I believe they need to be explicit and objective.

In the example trading systems I gave, you'll notice the Trade Mgmt Rules I gave were something like:
Long - S/L Mgmt Rules: Every Bar

And the Trade Management Value was (for example):
Long - S/L Mgmt Values: Low Of Previous 50 Bars

If we are trading a system manually (i.e. no computer involvement), then the rule for setting and moving the stop loss would probably be written something like "The stop loss is, and remains at, the low of the previous 50 bars".

But when you're making a computer do this for you, it must be specified exactly what that means. And what it means is "every bar calculate the low of the previous 50 bars. Use that as the new value of the stop loss". So I made a "Every Bar" rule, which of course evaluates to "True" at every bar. Without it, the stop loss would be set initially, but wouldn't move after that.

Some systems have more complicated Trade Management Rules than that. For example, and assuming of course the period and standard deviations for the bollinger setup are specified, "When the price closes above the upper bollinger line, move the stop loss to the value of the upper bollinger line". In this case, the Trade Management Rule is "close is above upper bollinger line", and the Trade Management Value is "the value of the upper bollinger line".

A close above the upper bollinger line isn't going to happen all that often. But when it does, this system is going to move it's stop to the value of the upper bollinger line. Given that there aren't too many bars that have their low above the upper bollinger line, that trade would probably get closed out the following day, but this is just an example.

Now, while I haven't exactly been thorough on Trade Management Rules and Values in this article, because the possible rules to pick from would be a lot like the Entry Rules and the possible values a lot like Entry Values, they are very, very important.

This is, after all, how we decide when to exit the trade. And exiting is when the profit is made, regardless of how we got in the trade in the first place.

There are some people who believe spending a lot of time on the Entry Rules is futile, that even random entries would work, because it's all in the exit. The thing about systems that you buy is that they focus mostly on the entry. Personally I think it all adds to your probability of success, but certainly the exit is where it either fills your wallet, or empties it.

You enter long and the price goes higher. Do you sell now? Is the price going even further, or is it about to turn around?

You enter long and the price goes lower. Do you sell now? Is the price going even further, or is it about to turn around?

Guess-work. It's why you need a solid trading system, proven over a long-ish period (which varies depending on the timeframe you trade), with clear, objective rules.

When you enter a trade and the price goes against you, that's where your stop loss comes in. You've already set it, hopefully, at a spot where you decided that if it went there, you are prepared to admit you got it wrong this time and take the loss.

When you enter a trade and the prices goes in your favour, your Trade Management Rules now specify when to move that stop loss to lock in some profit. When we set the stop loss initially, we gave the price some wiggle room. Because the price wiggles. We have to keep doing the same thing when we move the stop loss. Set it relatively close and yet relatively far.

And therein lies the problem. Exactly when and where to move the stop loss? Here's where experimentation and backtesting comes in. Buy some of those $49 systems to get some ideas. Try them out. Add rules of your own.

By this time you should have everything needed to specify entries. Look on the chart to see what normally happens after your system entered a trade. If the price continues on in your favour, then retraces 20 pips, then goes on, then retraces 20 pips, then you can't move your stop loss closer than 20 pips away from the price.

If the price seems to always retrace through the 0.382 fibonacci retracement level before heading further north, but it hardly ever breaks the 0.618 except when it's a reversal, then it stands to reason that your stop loss should be just below the 0.618 fibonacci level and not the 0.382 one.

And this is why your Entry Rules can't be random. Because probably your Trade Management Rules are going to change based on what normally happens to the price after you've entered. Random entries aren't going to allow you to work this out.

So, backtest. Experiment. Work out when the price is *most likely* just retracing before continuing on, and when it's *most likely* reversing. Difficult to do, but a worthy task. And then tell me :-)

About Stop-And-Reverse Systems
I've completely neglected this kind of trading systems, sorry. That's where your system is always in the market - if it's not going long then it's going short. In this case you wouldn't need either a take profit or (stricly speaking) rules which move your stop loss.

But, I include to "Exit At Market", which is what you do with your long trade in the stop-and-reverse system when you want to go short, as a Trade Management Rule.

Sure, in forex, if you have 100,000 going long, you could purchase 200,000 going short, which would effectively close your 100,000 long and leave you with 100,000 short. But I prefer to keep my trades distinct.

And so if I create a stop-and-reverse system in my software, I set the entry rules for going short to match exactly the Trade Management Rules of the long, and vice versa, with the Trade Management Value set to "Exit At Market". That way the long trade is closed, and the new short trade is opened.

Comments

I'm not sure if anyone realises, but comments are more than welcome. I was hoping for some exchanges of ideas. Unless, of course, everyone agrees with what I'm saying :-) Even the support-resistance stuff??

Comments will help your own understanding, along with anyone else who happens to read it. Sometimes you might even like to write something controversial, something which is even the opposite of what you believe, just so that someone replies and gives you a solid explanation on why what you actually believe is true. (If that makes sense).

Being a bit selfish for a second, comments on this blog have enabled my software to become better. As in less costly.

Anonymous comments are fine. Agree? Disagree? Let me know.

Example Objective Trading Systems

A lot of stuff to read, perhaps some examples will make what I'm trying to say clearer.

Break-Out System With Trade Management

Rule Type	Description
Long - Entry Rules	Close Below High Of Previous 100 Bars
Long - Entry Values	High Of Previous 100 Bars
Long - Init. S/L Values	Low Of Previous 50 Bars
Long - S/L Mgmt Rules	Every Bar
Long - S/L Mgmt Values	Low Of Previous 50 Bars

Reverse for shorts.

Explanation: Whenever the close is below the high of the previous 100 bars, an order will be placed at that high of the previous 100 bars. If the order is taken up, i.e. the price goes above the intended entry point, the initial stop loss will be set at the low of the previous 50 bars.

On the completion of each new bar, the stop loss is re-set at the low of the previous 50 bars. Often this means the stop loss will not be moved, but when it does, it will only move to a position of decreased risk. Eventually, hopefully, the stop loss will be moved to a spot above the original entry price, and that's how this system would make it's money.

Break-Out System With Take Profit

Rule Type	Description
Long - Entry Rules	Close Below High Of Previous 100 Bars
Long - Entry Values	High Of Previous 100 Bars
Long - Init. S/L Values	Low Of Previous 50 Bars
Long - Init. T/P Values	50 Pips Above Entry

Reverse for shorts.

Explanation: Whenever the close is below the high of the previous 100 bars, an order will be placed at that high of the previous 100 bars. If the order is taken up, i.e. the price goes above the intended entry point, the stop loss will be set at the low of the previous 50 bars. The stop loss will not be moved from that point. The take profit will be set at 50 pips above the entry.

This system puts barriers on either side of the entry, and doesn't move them. Either we get our 50 pips profit, or we take the loss.

%b System With Trade Management

Rule Type	Description
Long - Entry Rules	Close Above Simple Moving Average (Period=50)
Long - Entry Rules	%b (Period=20) Value Must Be Below 1
Long - Entry Values	Price Where %b (Period=20) Would Equal 1
Long - Init. S/L Values	Simple Moving Average (Period=50)
Long - S/L Mgmt Rules	Every Bar
Long - S/L Mgmt Values	Simple Moving Average (Period=50)

Reverse for shorts.

Explanation: The %b value specifies where the price is in relation to the bollinger bands. A value of 0 means the price is at exactly the lower bollinger line. A value of 1 means it is at exactly the upper bollinger line. 0.5 is in the middle, etc. I could have just used the "Upper Bollinger Line" value in this case, but %b allows greater flexibility when it comes to experimentation.

This system has two Entry Rules. Firstly, the price must close above a moving average with period 50. It must also be below the upper bollinger line with period 20. When both of those rules are true, an order will be set at the upper bollinger line.

Should the price rise above the upper bollinger line, our order will be taken up, and the initial stop loss will be set to the value of the moving average.

With the completion of each new bar, the stop loss is adjusted so it remains at the same price as the moving average. However, my software does not allow stops to be moved to positions of increased risk. Therefore, if the moving average went down, the stop loss would not follow it.

This system relies on the moving average to eventually rise above the entry value.

By the way, these are just example systems, and may or may not be actually useful in real trading. My point is to simply give examples of purely objective trading systems.

Initial Take Profit Values

This is optional. You don't always need an Initial Take Profit Value.

There are two ways to exit the trade - either your stop is taken out, or the take profit is hit.

Your Initial Stop Loss Value will always be, for long trades, below the Entry Value. Obviously.

Sometimes you rely on your Trade Management Rules (next article) to gradually move your stop loss from that loss position, to a break-even position, and eventually to a spot where even if your stop is taken out, you will make a profit.

Or, you can have a Take Profit set. Your trade will be closed at a profit if that level is hit. Or you could do both. Anyway, Take Profits are optional, but only if you have Trade Management Rules which will eventually move your stop to a profitable position.

Initial Take Profit Values are exactly the same as Initial Stop Loss Values, which are essentially the same as Entry Values. High, Low, Close of current bar, the value of a moving average, the lower bollinger line, low of the last X bars, a specific number of pips above (or below) the entry price.

Initial Stop Loss Values

Alrighty, Entry Rules say "yes or no", Entry Values set the price to enter, now we need an Initial Stop Loss Value. That is, at what price do we admit that the trade did not do as we thought? At what price do we cut our losses?

I've heard some people don't trade with stops. They reason that the stops are too restrictive, and are actually more likely to make people lose money. But I couldn't do that. First, too scary, and second, they'd be relying on some kind of intuition as to when to get out. I don't do intuition.

Stops also allow us to define in exact terms, how much are we risking on this trade. Sometimes there is slippage though, and so you might lose a bit more than you thought.

Anyway, the beauty of forex is that the stops are automatic. The worst trade I ever did was with Options. The price was already below my Initial Stop Loss Value. But I just couldn't call the broker to get out. "It'll come back" I kept saying, as the price kept falling. Automatic stops take away that problem. When the price hits your stop, you're taken out of the trade. Perfect.

So, where are we going to put that puppy? We need to give the price enough wiggle room. On the other hand, we don't want to risk too much. It's backtesting time.

Possible values are essentially the same as for Entry Values. High, Low, Close of current bar, the value of a moving average, the lower bollinger line, low of the last X bars, etc. You can also use a specific number of pips below (or above) the entry price.

Entry Values

Okay, so you've got Entry Rules, and they say "yes, you may enter a trade", or "no". Yes or no. If there's a "not sure" in there, it's not an objective trading system. I'm only interested in objective trading systems (due to my inability to trade subjective ones).

Now we need an exact price at which we will be entering the trade. These are pretty easy. Normally.

For example, "the high of the most recently completed bar". Your Entry Rules say "enter the trade", and your Entry Value says "...when the price hits the high of the most recently closed bar".

Some people like to add a bit, so they know the price has a bit of momentum. For example, "the high of the most recently completed bar, plus 1 tick". Or plus 2 ticks, or 5 ticks, or whatever.

Notice though, that we don't have a direction for the trade yet. Sometimes your Entry Rules will specify the direction, like "enter a long trade if the bar closes above the 20-bar moving average". And then the Entry Values specify exactly where to open that long trade.

Or, sometimes the Entry Rules simply allow you enter a trade, and the Entry Values specify the direction. For example, the Entry Rule is "enter a trade when the RSI with period 14 goes above 70". And if the price hits the high of the most recently completed bar, you enter long at that price. If the price hits the low of the bar, you enter short at that price. In this case your Entry Rules have told you that the price is about to move, but it doesn't know in which direction. Your Entry Values answer the direction question.

Here are the values that Thinking Stuff currently has programmed. There'll be more:

Close Of Current Bar
High Of Current Bar
Low Of Current Bar
Close Minus X Average True Range(s)
Low Minus X Average True Range(s)
Close Plus X Average True Range(s)
High Plus X Average True Range(s)
Upper Bollinger Line
Lower Bollinger Line
Price Where %b Would Equal X
Bar Swing High
Bar Swing Low
High Of Previous X Bars
Low Of Previous X Bars
X Percent Of Bar Range Above Entry
X Percent Of Bar Range Below Entry
Simple Moving Average
SMA Swing High
SMA Swing Low

And there's also a Pip Offset thing in there, if you want to add or subtract a certain number of pips from the value calculated above.

Sometimes though, the price is already above (or below) the price that your Entry Values say to enter. What to do then? Do you pass on any trade where you can't get your intended price? Do you get in at any cost? Do you get in, but only if the price hasn't gone "too far"? In that case you need to specify what "too far" means. 20 pips? 50? 100? Backtest.

Number of Trading Systems

In my previous posts, I seem to be fixated on exactly how many different trading systems a set of rules specify.

I've talked about having the word "OR" in a system effectively creates more than one system. If you've got a system which a whole bunch of rules, but one that specifies an entry when the close is above the 20-bar moving average, OR when the 14-bar RSI is above 70, then actually you have two systems - one which enters when the close is above the 20-bar moving average, and one which enters when the 14-bar RSI is above 70.

Then, in that system with the points (where any combination of rules adding to more than 36 points signals an entry), I said that effectively those rules made X number of different systems - one for each combination of adding the rules together to get more than 36 points. In that particular case, where there were 9 rules, with respective points of 8, 8, 8, 8, 6, 8, 6, 6, and 4, I did start to work out exactly how many different combinations could add to more than 36, but I quickly realised there were going to be many. Maybe about 20 or so. 20 different trading systems.

Why is this important?

Because one day the markets going to change its "personality", and the system you've used for years isn't going to work anymore.

When that happens, you need to find out what's wrong and fix it.

Is it the close above the 20-bar moving average that doesn't work anymore, or is it the 14-bar RSI above 70? Or both? Keeping track of them as one system will not tell you.

Trying to find out the problem in that points-based system would just be a joke.

Dare I say it, but knowing when to use a trading system and when to stop using it, is probably more useful than entry rules.

We all want equity curves that start in the bottom left, and end in the top right. But what if it starts bottom left, goes to middle top, and is currently working its way to bottom right? Is this a temporary drawdown, or catastrophic failure?

Much like we need exact, written-down, objective rules for when to enter, where to set the stop loss, when and where to move the stop loss, when to exit, how much money to use, etc, we also need such rules for when to stop using a system.

Without objective rules set out for when to stop using a system that's in a drawdown, we have essentially moved our "oh, I'll stay in a bit longer - it'll come back" thoughts from the price chart to the equity curve chart.

The reason these rules would be more important than entry rules, is because these rules would stop us using a poor-performing system - it wouldn't matter what rules made up that system.

So, what might be an example of an objective "This is not just a drawdown, but rather a catastrophic failure of your trading system" rule? I'll talk more about this at the end of the series. At this rate, maybe some time in 2007 :-) But you should check back everyday just to be sure.

I hope people are finding what I say interesting, and I'd certainly appreciate comments - anything at all. Anything.

Re-Think

I've said a few times that trading rules need to be individualised, to suit the person doing the trading. You can't just buy someone else's rules and expect they'll work for you. I said.

But I've changed my mind. That's for if some of the rules are subjective. Normally Support and Resistance, or Trend lines, are thrown into the system, and different people will place lines at different places.

If the rules are completely objective, it really wouldn't matter what the rules are - being completely objective, me, you, your neighbour's kid should be able to trade those rules with no problems at all.

So, if you purchased a trading system consisting of only objective rules, theoretically anyone should be able to trade that system.