UBBFriend: Email This Page to Someone!
  Reef Capital Coaching - Quality Education at a Reasonable Price
  Building Mechanical Trading Systems
  Volatilty Breakout Systems (Page 1)

Post New Topic  Post A Reply
profile | register | preferences | faq | search

This topic is 3 pages long:   1  2  3  next newest topic | next oldest topic
Author Topic:   Volatilty Breakout Systems
Nick Radge
Moderator
posted 08-09-2000 08:35 PM     Click Here to See the Profile for Nick Radge     
Firstly, let me discuss what is intended by this thread and perhaps some following threads. There has been some discussion on Volatilty Breakout methods written in other area's and I think it beneficial to tie it all together and show you how to build a mechanical trading system using such a method. Once we have done this, perhaps we can move on and together build other systems that can be further explored, enhanced and tested by members here.

Many professional traders use such a system and for good reason. Firstly let me discuss the 'mechanical approach'. A 'mechanical' approach is where the signals have been coded into a trading software package such as TradeStation, SuperCharts or even MetaStock. This system code can then be backtested and improved using 'correct' techniques. Incorrect techniques such as blatant optimization will cause error and loss down in the real world. A 'mechanical' approach, once tested and satisfies the user, will spit out the buy and sell signals without regard to any external factors. These signals should then be taken by the trader, placed in the market and followed without regard for any other external factors. This is hard for many people due to their own ego, outside influence or personal confidence at the time. In other words, a mechanical trading system doesn't care for anything except the buy and sell signals it generates, and nor should the trader.

I will continue to build upon the theme of a mechanical trading system from scratch in the form of a volatility breakout system.

Feel free to ask questions as we go and also be aware that it may take soem time to cover...as I am going skiing tomorrow for a week and will not be particularly interested in trading doing 40 kms through the treeline....

Stay tuned

IP: Logged

Nick Radge
Moderator
posted 08-09-2000 08:55 PM     Click Here to See the Profile for Nick Radge     
Lesson 1
*********************************************

The volatility breakout method is a very prominent trading tool and is very easy to use. It is sometimes known as a 'volatility expansion' method and very simply takes its name from an aggressive change in the volatilty of the market. What does this mean? Volatility can be measured in various ways, but we will look at it from the perspective of the daily range, that is the distance from the high to the low on any given day. We will call this the "range". Another method is the Average True Range (ATR) which is also the high - low plus any opening gaps. We will ignore the ATR measure and stick simply with the "range".

The theory behind such a method is that a period of low volatility, that is a period where the daily ranges are smaller than in the recent past, will be followed by periods of higher volatility where the daily ranges are larger than in the recent past. This increase is known as the "expansion" and hence the name 'volatilty expansion'.

The type of market we need is where prices open on its lows and rallies hard (expands its range)to close on its high. This type of situation is called a "trend day" or a "run day". At a trend turn, these can be called 'thrust days' and by themselves can indicate a good trend change indicator. The ideal situation is an open on the extreme low and a close on the extreme high but unfortunately I haven't found the perfect way to predict these types of days befoe they happen. Here is an example of a "trend" or "run" day:

Open on the low and close on the high. Remember when trading futures we also look for the opposite.

On that matter this type of method works in stocks but is difficult to track. My suggestion is to only follow optionable stocks and use the leverage available to trade it. Futures traders need volatile markets, such as stock inices and bonds to get best results.

So in summary, we are looking for an incraese in the daily range, an open near the low, a close near the high and some way to predict it.

Its easier than you think.


IP: Logged

Nick Radge
Moderator
posted 08-09-2000 09:16 PM     Click Here to See the Profile for Nick Radge     
Lesson 2
*********************************************

So we know what a trend day is. We have previously discussed that periods of low volatility, as measured by daily ranges, are followed by periods of higher volatility.

The formula is very simple:

Range = (High-Low)

Here is a simplified indicator of the daily range:

You can see its quite choppy, but this is what we need. This can be smoothed like so:


The formula to smooth the range would be something like:

Smooth Range = (Average(Range,Length)

where Length of smoothing is 4 days.

The smoothing factor may or may not be included and can be, in theory, any length. Obviously having a longer smoothing factor will further remove the volatility from the current price action.

Ok. We can now see the swings in volatility. We can also see the 'trend' days such as on the previous chart, but just how to get on these?

Very simple as I said. We must base the 'range' or 'volatility' off something, like an anchor. An anchor is therefore the start of our measuring and would be simply one of the 4 certainiites that happen every day in every market; the open, high, low and close.

This where it starts getting interesting. We will start with one anchor and follow through, but then as part of our testing we will revisit and test basis other anchors.

We'll start with the opening price for two reasons; (1) it happens first and (2) its part of our 'trend' day setup.


[Note: This message has been edited by Admin5]

IP: Logged

Nick Radge
Moderator
posted 09-09-2000 09:25 AM     Click Here to See the Profile for Nick Radge     
Lesson 3
*********************************************

We will focus on the opening price as our anchor or measuring point of the days range. As mentioned earlier, we are attempting enter into market whose volatility is rapidly increasing and will potentially continue. If we can get aboard such a move and even better a situation where the close is the absolute high, we may be able to generate some low risk profits.

The opening price can be a little more complex than exactly that, the first price. But for our exercise here, the first print of the day will constitute the opening price.

Once the opening price has been established, we need to establish an increase in volatility. We know what the current volatility will be, basis the range indicator we have been able to calculate.

BY ADDING THIS RANGE ONTO THE OPENING PRICE, WE ESSENTIALLY HAVE A MEASURING LEVEL OF CURRENT VOLATILITY.

Lets use a very basic example. Say we have an index that trades between 1 and 100. It closed yesterday at 55 and has a daily range of 5 points (known as the 'expansion factor'). This means that on any given day the market will trade in a 5 point range, more often than not around its opening price...or will it? A trending market does not trade just around its opening price, it tends to move up in fits and starts. What we are looking for is an aggressive move up that starts on the days lows (ie the opening price) and continues to move in the same direction without looking back.

So our artificial market that settled at 55 yesterday opened today at 56. If we add the range measurement or expansion factor, 5 points, onto the opening price we potentialy have an extreme for the days move at 61 (56 + 5)if volatility remains static. If volatility is decreasing, that is the daily ranges are contracting, then 61 may not be reached. Remember, low volatility is followed by higher volatility and its this increase in volatility we are looking for.

What we should assume is that any increase in volatility will continue and that such a continuance of the volatility will allow the market to keep moving in the desired direction. Technically we are looking for a high probability point where, if raeched, the odds are still in our favour for the move to continue. This probability point is clearly the 'expansion factor' number, but how can we derive an accurate number?

Firstly, we have to ascertain how often we wish to trade. By having a 'factor' of less than the days range we may be forced to trade more often and potentially deal with more market "noise". In an ideal situation we want reasonable extremes which are outside the markets noise or normal range.

Saying this, we don't wish to have the factor too large or else we may rarely get to trade. A balance is what is needed and is something we can test quite easily.

IP: Logged

Nick Radge
Moderator
posted 09-09-2000 09:55 AM     Click Here to See the Profile for Nick Radge     
Here is an example of a range breakout in a visual sense:

You can see the resultant large up day. Although the open was not the extreme low, we still get the general picture of the setup. In the above example, which is the SPI futures contract, the details of the setup are:

Open: 3282

Range Factor: 35 points

Trigger: 3317 (3282 + 35)

Close: 3361

Profit on 1 contract would be $1100.

The close of the day was 3361, just 1 point from the days high.

This looks straightforward but there are several considerations to take into account. Our liitle example here does not take into account any retracement during the day. We all know that prices just don't go straight up without looking back. Who knows, we may have seen the lows AFTER we entered? You cannot tell from an end of day chart and these software packages rightly assume if the open is near the lows and the close is near the highs then the low was seen first and the highs seen last. This example was most likely that, but don't be fooled.

Eyeballing the charts can give us this picture in a visual sense, but this is not a mechanical system and it would certainly take soem time and patience to backtest the method.

What we need to do is mathematically define the 'expansion factor' and incorporate it into the trading system. We will now test several variations around the single days range. Therefore, the 'factor' will be a multiple of the days range. Remember, a multiple of less than 1 will mean more trades and more noise and a multiple of greater than 1 will mean less trades and less noise. Not necessarily better though.

Lets test 3 factor lengths; 0.75, 1.0 and 1.25 where the factor is multiplied by the previous days range and added to the opening price.

Some assumptions must be made to get an accurate test. We will come back later to straighten them out, but we need them now to find a suitable range factor.

The assumptions:

(1) positions will be closed at the close of each day
(2) we will use 1 contract
(3) we will assume an account size of $20k
(4) we will risk no more than 2% of our account on the trade ($400)
(5) we will therefore use a "hard dollar" stop of $400

A hard dollar stop is simply a dollar amount that when reached will stop us out. It is not a technical point as I discuss in my Online Course. In this case, the SPI is $25 per point and therefore the stop will equal 16 points (400/25=16). So, when we enter a trade, we place a stop 16 points from entry which equates to our $400 or 2% account limit. In the example above, the hard dollar stop would have been placed at 3301 (3317 - 16).

IP: Logged

McScrooge
Member
posted 09-09-2000 12:55 PM     Click Here to See the Profile for McScrooge     
Nick
Great thread. It's enlightening to watch a professional at work.
Is this system suitable for stocks? Wouldn't the position size need to be very large to get a similiar return?
Your avid student
Mike

IP: Logged

Nick Radge
Moderator
posted 09-09-2000 01:44 PM     Click Here to See the Profile for Nick Radge     
Lesson 4
********************************************
This is the TradeStation coding:

Inputs: Mult(1.0), PositionBasis(True), ShowText(False),Length(1);

Buy Next Bar at Open Next Bar + Average(Range,length) * Mult stop;


Sell Next Bar at Open Next Bar - Average(Range, length) * Mult Stop;

The close at end of day and hard dollar stop can be set without coding.

Here's the tested parameters we have chosen. The only thing not included is brokerage, which will have to be included at some stage.

Here is the results with the factor set at 1.25 * yesterdays range:

Here are the results with the factor set at 1.0 * yesterdays range:



And lastly, the results with the factor set at 0.75 * yesterdays range:


You can clearly see that the lowest factor has produced the best result, but also many many more trades. We haven't optimized or done any out-of-sample testing as yet, we've just picked several random numbers for the factor calculation.

Also to note is that by closing the trades at the end of the day, we don't allow the profits to run a great deal and this will only decrease the effectivenes of the expansion of volatility.

What we should now do is to find a 'robust' parameter for the expansion factor. This is where "smart" optimization is used as opposed to blatant optimization.

To do this, and again at this stage we aren't concerned about the exit, we must test on some out of sample data. Out Of Sample Data is where the system has been optimised on as yet and is sometimes referred to as forward testing.

Very simply, we select 10 years worth of data, in this case May 1990 to May 2000. We will select the middle 2 years, May 1994 to May 1996 and optimise the expansion factor during this period of time. This will give us the best parameter setting.

We will then apply that same optimized setting to the May 1990 to May 1994 data and again to the May 1996 to May 2000 data. If the results more or less come out the same, we can assume that the parameter is reasonably robust and can be counted on to perform the same way in the future.


[Note: This message has been edited by Admin5]

IP: Logged

Nick Radge
Moderator
posted 09-09-2000 03:13 PM     Click Here to See the Profile for Nick Radge     
Lesson 5
********************************************

I have now run a basic test to find the most optimal yet robust parameter without over-optimizing the data.

I initially chose the May 1994 to May 1996 data and optimized the expansion between 0.50 to 1.50 in increments of 0.1.

The best result was 0.6 and can be seen here. Don't forget this is only 2 years worth of data:

Here is the backtest, or the same 0.6 parameter on the first data set, that is the May 1990 to May 1994 period:


And one more, the forward test, that is the period from May 1996 to May 2000. The 0.6 parameter is used again:

So, this shows us that the same briefly optimized parameter has worked reasonably well on three different sets of data which indicates its reasonably robust. Here are the total results for the entire period:

And the equity curve doesn't look too bad either:


So, we have found a parameter that suits our need as the entry and has returned a respectable 19.57% pa on our $20k. We simply multiply yesterdays range by 0.6 and add/subtract from the opening price. This is our entry level and we exit at the end of the day.

Before moving on we must try and find problems with this method as opposed to concentrating on the result. If we can pull it apart and not find any major issues, well, we may just have something.

Firstly, the average trade is a profit. This is a good start. However, its only $59 and we haven't included brokerage!

Secondly, it trades a lot. 1919 trades in 10 years is alot of trading, in fact you place a trade every second day.

Next, the system has had 7 consecutive losers. This is very normal and a professional traders will not have a problem with seven losses. But maybe you would? This seven is not a one off. here's the numbers:

4 consecutive losses occurred 13 times

5 consecutive losses occurred 13 times

6 consecutive losses occurred 1 time

7 consecutive losses occurred 3 times

Getting the picture. Psychologically it may be harder to trade than first impressions.

Lets move on though. The brokerage is a serious problem as is the average dollar profit per trade. The brokerage bill for this system assuming $20 per contract would be a whopping $38,800 for the 10 years! Jeez, this would take the profit to $75,000 and accounts for 34% of our hard work. Perhaps being a broker is the trick.

What needs to happen is to increase the average dollar profit of the trade. To do this, we need to (a) negotiate free brokerage, and (2) expand the ability to make a profit.

Good luck with (1). With (2) though, we can give it a go.


[Note: This message has been edited by Admin5]

IP: Logged

KamiCrazy
Member
posted 10-09-2000 08:05 PM     Click Here to See the Profile for KamiCrazy     
Hello, I was just wondering is this system suitable for stocks? and if it is how do we pick which stocks are suitable, surely there would be 10's if not hundred's of entry signals a day. Also how do you program this into the Metastock System Tester? or is the Explorer better?

Regards
KamiCrazy

IP: Logged

biggles
Member
posted 12-09-2000 09:38 AM     Click Here to See the Profile for biggles     
OK Nick,
Focussing on the profitability, here's the results of my mucking around with it, with input from Elizabay.
The exit seemed to be the most obvious place to start, so we tried using Larry Williams' Bailout (next profitable opening) as the exit.
Next, we eliminated all short trades, so it's now a system for long trading only. In doing this, I found that a volatility factor of 0.8 gives a better result than the 0.6.
Lastly, I added a $400 trailing stop to go with the MM stop.
Tested it using your parameters (back and forward).
See what you think...!
First screen is as yours - without any costs added.
Second screen is with my own cost details - $30 brokerage, $50 slippage, $2500 margin.
Cheers,
Steve

IP: Logged

biggles
Member
posted 12-09-2000 10:01 AM     Click Here to See the Profile for biggles     
P.S.
Forgot to mention - 'smoothing factor' over which average range is calculated is 14 days.

IP: Logged

th
Member
posted 15-09-2000 11:34 AM     Click Here to See the Profile for th     
Nick please dont take the lack of replies as lack of interest.It is really appreciated the time you have obviously put into the posting. I would venture to say the majority of us are Metastock users trading patterns and would appreciate how a mechanical back testing system can be applied in our situation. I am venturing into end of day trading US futures using your course pattern techniques and know it is vital to backtest. How does one do it effectively over say 30 commodities over 10 years for each ?

IP: Logged

Malloc
Member
posted 15-09-2000 07:31 PM     Click Here to See the Profile for Malloc     
TH,
I might be wrong here, but as for lack of replies, I think everyone is waiting for Nick to return from his week away skiing.
As for MS, the Metastock language seems to be rather limited in its backtesting capabilities in comparision to Tradestation

IP: Logged

Dave
Member
posted 16-09-2000 12:34 AM     Click Here to See the Profile for Dave     
Malloc
As a user of Ezychart with no backtesting facility, your comments begs the question, what software would best suit backtesting?
Tradestation is probably over the top for me so Metastock or Supercharts or something else.

Dave

IP: Logged

danny
Member
posted 16-09-2000 08:28 AM     Click Here to See the Profile for danny     
Dave,
From what I know, SuperCharts is superior to MetaStock for backtesting.
However, I very recently read that Omega Research (makers of SuperCharts and TradeStation) discontinued with SuperCharts.

IP: Logged

Jeff Eisman
Member
posted 16-09-2000 09:11 AM     Click Here to See the Profile for Jeff Eisman     
Dave
I think if you search through the archives, there has been much discussion of the relative merits of Supercharts vs Metastock

For a klutz like me who can't program, there is no choice - SC- at least I can write simple systems etc

On the other hand, when you see what TechA and others can do with MS.....

Suggest contact Paritech who certainly supply MS and I think SC

Last version SC fairly old, and you need to download y2k compliant updates from Omega site (easy enough, I managed it )

Cheers

Jeff

IP: Logged

Nick Radge
Moderator
posted 17-09-2000 04:00 PM     Click Here to See the Profile for Nick Radge     
Thanks all, and yes I have been skiing in the lovely Perisher. I will continue to add to this thread over the coming days, but alas, I have a full Olympics schedule now!!

Sorry for those waiting on the futures course; I really am trying.

N

IP: Logged

biggles
Member
posted 17-09-2000 04:40 PM     Click Here to See the Profile for biggles     
Life's a real bitch, eh Nick?!!

IP: Logged

Dave
Member
posted 17-09-2000 11:14 PM     Click Here to See the Profile for Dave     
Thanks Danny and Jeff
I guess there's no easy answer, the MS programming looks challenging (do I really need a huge learning curve) and SC looks like its being made obsolete. What I'm hoping for is some one with both systems that can compare pros and cons of each. Of course the demos are available but that doesn't give the true picture for backtesting just an overview of each system.
In the meantime I'll ring Paritech.

Dave

IP: Logged

Ross B
Member
posted 18-09-2000 01:18 PM     Click Here to See the Profile for Ross B     
Dave,

I am a backtesting maniac, and although I have Metastock, I prefer to use Ezycharts manually, going back several years and magnifying charts where buy/sell signals have been indicated, taking the data straight off the bars and recording it in Excel for subsequent analysis. Maybe this would not work for all systems, but it's great for backtesting pattern setups and indicators like MA crossovers, ST & RSI divergence etc. When I first started, I was mindful that Tradestation etc would automate the task and save me the laborious hours I was putting in, but at a certain point I became aware that doing it the slow way gave me a "feel" for the charts that automated testing would not. I would not do it any other way now. Try it for a while before you dismiss the option as too labour-intensive.

I actually find that it is a good tonic for those times - like now - when I can't seem to do anything right. Gets me back in touch with the notion of an edge being possible, and having probability work for you in the long-term, even with the run of losses that is statistically inevitable (puts this in perspective, which is a big positive at the moment). Most importantly, re-emphasises that executing stop losses is integral to any system's success, and that picking exact tops and bottoms systematically is impossible.

Cheers

IP: Logged

Nick Radge
Moderator
posted 18-09-2000 02:52 PM     Click Here to See the Profile for Nick Radge     
Beach Volleyball out of the way so I can reply to Biggles,

Why leave out short trades? Thats makes no sense and will create a bias in the system. At this stage we are testing futures and ALL trades, both long and short, should be taken into account.

We will move onto stocks after this and work our way through.

As for which program, well, TradeStation or SuperCharts do make life easier. Next to this my choice would be Excel. I have no experience with other programmes, but the coding should not be too much different. I believe MS cannot do intra day signals, so that programme has a major weakness in any kind of testing.

Baseball tomorrow.

Nick

[Note: This message has been edited by Admin5]

IP: Logged

biggles
Member
posted 18-09-2000 05:39 PM     Click Here to See the Profile for biggles     
Nick,
Short trades were eliminated from the system simply because they represented a very pronounced loss over all the periods tested.

With them included, overall profitability dropped from nearly 600% down to about 150% for the ten year period - short trades on their own represent about a 50% loss. I can only imagine that there is a greater psychological impact when prices break out significantly on the upside?

Why would you include a class of trades in your system that are clearly consistent losers, just to prevent a bias? For 'bias', can't we read 'edge'?

Cheers,
Steve

P.S. Hope the baseballers fare better than the bloody Boomers!

IP: Logged

AllanS
Member
posted 18-09-2000 06:20 PM     Click Here to See the Profile for AllanS     
What is this baseball nonsense anyway? It ought to be Cricket! Is there an Olympic Pub Darts competition?

IP: Logged

Nick Radge
Moderator
posted 19-09-2000 09:02 AM     Click Here to See the Profile for Nick Radge     
Biggles,
Not sure on where you are getting the results. Here is the TradeStation split on the long trades vs. short trades.

The question to be asked when leaving out the short trades is "What happens when a pronounced long term downtrend occurs?" You simply will be left sitting on the sidelines. This is because your system has a bias - you do not want any bias unless there is a firm reason for it.

Lets move on. We have an entry setup that seems robust, so no more fiddling should be done here. More fiddling = less robust = potential for future underperformance.

So, we need to expand the profitability of the system. The only way to do this is to increase the average dollar win by giving the market more room to allow profits to accumulate. Time is the obvious factor to do this.

The test used an exit on the closing price. This will obviously limit larger moves and naturally decrease our profitability. Many get concerned about holding positions overnight due to adverse moves against them. Using 24hr stops will decrease (if not eliminate them) any potential damaging move offshore and will in fact open the way for increased profitability because of favourable moves that will also occur.

Exits come in all shapes and sizes. Where to start? I geuss we need to look at our trading goals and your personality type. I cannot answer these - as I continually point out in the course. Only you can answer these questions ans you will move naturally toward a trading plan or system that suits you.

I wish to keep this system relatively short term, that is under 5 days. I would prefer not to have a same day exit because the profitability, as we have seen, will always be limited.

So, lets investigate some exits and I welcome some more for our testing purposes. Here's a quick list, and you can add and I will test (after the baseball - Aus v Jap).

(1) Exit next open

(2) Exit 1st profitable open

(3) Exit at profit target

(4) Exit next close

(5) Exit second open

(6) Exit on n'th day

....and so on. Again, we do not wish to optimise. We need to select our middle two year test period, find the best result and then see how it works in the backtest and the forward test.


IP: Logged

elizabay
Member
posted 19-09-2000 01:37 PM     Click Here to See the Profile for elizabay     
Nick ....... Are we running transaction costs with these tests? If yes, should we standardise to simplify comparisons? Say, $30 round trip broker, $50 round trip slippage? PJ

IP: Logged

Nick Radge
Moderator
posted 25-09-2000 08:53 AM     Click Here to See the Profile for Nick Radge     
I have done some basic look backs with several exit techniques. These do not include any costs as yet as our goal at the moment is to increase the average winning trade in dollar terms. The following are also non-optimized and were just some examples of shorter term exits that came to mind:

(1)
50% Trailing Stop
$400 Floor
$400 Hard Dollar stop

Stop gets trailed by 50% of gains once $400 floor profit achieved.


(2)
No Exit
$400 hard dollar stop

No exit used. Essentially a stop and reverse system when new opposite signals generated.

(3)
Exit Next Open
$400 hard dollar stop

Exits first price next day. This is the classic Larry Williams bailout technique.

(4)
1st profitable close
$400 hard dollar stop

exits on 1st profitable close only.

(5)
Exit on 5th profitable close
$400 hard dollar stop.

Same as aboce, except takes 5th profitable close.


(6)
Profit target of $400
$400 hard dollar stop

Uses profit target equal to initial risk to exit.


Summary:

Well, our goal was to increase the average winning trade's profitability but keeping the system reasonably short term. The best result in that regard was the % trailing stop with an average trade of $103.89. The strike rate is still acceptable at 54.18% yet it wasn't the highest volume trading system which will help keep the brokerage bill down. The downside to this is the amount of work, especially intra-day to monitor the trailing stop. Most brokers will not perform that duty for you.

The Larry Williams bailout or 1st open exit worked well with an strike rate of 59.52%, but notice the tradeoff: the small avge dollar win at $86.85 and also makes a whopping 1927 trades. I think your broker would do better than you out of this one. This method had the highest risk adjusted reward ratio and if brokerage wasn't an issue, this would be the most comfortable to trade.


[Note: This message has been edited by Admin5]

IP: Logged

Jeff Eisman
Member
posted 25-09-2000 09:27 AM     Click Here to See the Profile for Jeff Eisman     
Nick

My broker is the only one who has made any money out my trading last few months anyway

Jeff

IP: Logged

elizabay
Member
posted 26-09-2000 12:44 PM     Click Here to See the Profile for elizabay     
Nick - Even $103 is not too attractive when costs are added in. Or am I placing too much emphasis on costs too soon? Cheers, PJ

IP: Logged

Nick Radge
Moderator
posted 26-09-2000 07:38 PM     Click Here to See the Profile for Nick Radge     
PJ,
This is what is expectd of a short term system; higher risk, higher probability, smaller reward. We risk $400 to make $103 and do so 55% of the time, we can make money.

Costs are the difference between theory and reality. Costs are both commissions and slippage and you are completely correct, they tend to kill high volume trading systems.

Paul Tudor Jones and Monroe Trout, two massively successful hedge fund managers who use a very diverse range of short term trading systems both pay next to no commisssion and have specialist brokers to get better fills or "positive" slippage. The retail sector does not have this advantage.

As an example, I pay $4 R/T for the SPI, so I could net $99 per trade on that system excluding slippage. Multiply that by 1400 trades, add some risk by multiplying the contracts traded to 10, and all of a sudden you can make reasonable money. Find 3 to 5 of these such systems,put them together and your talking about a mechanical way to make $500,000 per annum with little daily input.

Thats how those guys do it.

IP: Logged

Nick Radge
Moderator
posted 26-09-2000 07:47 PM     Click Here to See the Profile for Nick Radge     
Here's 3 such systems that I trade in the SPI. These numbers are the net average monthly profits for the last 10 years. Scary thing is that the exact same systems work just as well in things like the Nikkei, T-Bonds and other large contracts. All signals across all markets take about 10 minutes per day to calculate and administrate...


This should be your goal as a trader: multiple systems that are proven to profit over time and that do not require excess daily input.

[Note: This message has been edited by Admin5]

IP: Logged

elizabay
Member
posted 27-09-2000 07:40 AM     Click Here to See the Profile for elizabay     
Nick - thanks. That's extremely interesting stuff. So, in practical terms for us scungey retail types, who don't get anything remotely close to $4 r/t on the SPI or anything else (and sometimes slippage can be a cause of chagrin), does this mean we need to filter the trading signals to find the opportunities where the potential dollar profits/trade are attractive? The corollary, of course, is a significant reduction in trading activity. PJ

IP: Logged

AllanS
Member
posted 27-09-2000 09:40 AM     Click Here to See the Profile for AllanS     
Am I right in thinking that if there are 100 American traders doing this sort of thing on the Spi for every 1 Australian trader doing it on the Dow, Australia will eventually go broke? After all, the money must come from somewhere. Is this what's happening to our dollar?

Keep writing that Futures Course, Nick!


IP: Logged

JonS
Member
posted 28-09-2000 08:35 PM     Click Here to See the Profile for JonS     
Sorry, as this slightly off the topic, though I believe it is still relevant to the discussion at hand.

Last time I brought options on the SPI, I paid $52 R/T.

What is the brokerage that us average retail types should expect? This figure is quite important to the tradeability of the sort of system we are considering here.

Further, what about slippage? Is it reasonable to assume a constant x points or a fixed percentage of the ATR? Thanks.

IP: Logged

biggles
Member
posted 01-10-2000 11:29 AM     Click Here to See the Profile for biggles     
Hi Nick,
Sorry for the slow reply - been down in the Emerald City myself doing some 'Oi, oi, oi' !

Further to PJ's point re costs - as these are the 'difference between theory and reality', isn't it reasonable to consider that we lowly retail types (I can only dream of the day when I can negotiate a $4 r/t deal!) might need to filter our systems a little more deeply in order to return a more profitable result?

This is why I arrived at the variant of this system as described above - when I took costs into account, short trades as a group were so clearly unprofitable it seemed sensible to leave them out - yes, this means I am introducing a bias I guess and that I will be sidelined during a long period of downturn, but I figure that, as long as I have a small group of other systems operating at the same time, I should be able to continue trading.

Below is my complete system, including the short side - I think you'll agree that there's a marked difference in return when its included.

Just to recap - it uses a volatility factor of 0.8, a range average of 15 days, $400 m/m and trailer, costs of $30 broker, $50 slippage and entry is on a retracement (close

Cheers,
Steve

IP: Logged

Nick Radge
Moderator
posted 01-10-2000 05:16 PM     Click Here to See the Profile for Nick Radge     
The average retail client will pay between $15 - $25 per side. Anything higher is a complete rip off. The brokers cost is generally $2 or so. A second tier brokers cost may be $4 or so.

The traditional futures broker expects to get 10 or so calls from a client in any given day and tends to think that they hold th clients hand throughout the day and hence why they believe the higher charge is relevent. I disagree and as those of you who have done my course, a broker is only required to execute and clear the trade.

JonS,
Slippage can be reepresented by both positive and negative. I tend to use $50 per contract as a guide as opposed to an ATR measure. I don't think it needds to be overly complex to get a fair idea.

What to do with our system? Well, it should be safely assumed that any short term system will have small average dollar winning trades and that these will be eaten away bu commissions and slippage. With no way around the inherent costs simply means we have to look elsewhere to increase the average dollar win and decrease the trading frequency.

This can be done using the same entry technique. What we will have to do is turn the timeframe out somewhat and even change the instruments being traded to get a slightly longer trend. Remember, most stock index contracts are backfill markets and will not present smooth trends.

Perhaps we can start a new thread and invetsigate this course of action? It should also give some good results with stocks as well.

The futures course is half way through. Hopefully due by end of October.

Comments welcome.

N

[Note: This message has been edited by Admin5]

IP: Logged

elizabay
Member
posted 02-10-2000 08:50 AM     Click Here to See the Profile for elizabay     
Nick

It would be gratifying to take the discussion through at least to a point where we have a workable proposition for all of us, including those with brokerage costs on the high side. (And, yes, I need to do some work there too).

Could we perhaps select one or two commodities/currencies with a reputation for trending and build up the system there from where we have left off on the SPI? Might allow us greater flexibility with exits, including the opportunity to position over longer periods. A suggestion.

By the way, many thanks for the guidance to date. I have learned a lot and enjoyed the discussion. Cheers, PJ

IP: Logged

biggles
Member
posted 02-10-2000 02:34 PM     Click Here to See the Profile for biggles     
Ditto PJ's last line.

Steve

IP: Logged

Ligier
Member
posted 04-10-2000 09:42 AM     Click Here to See the Profile for Ligier     
Whoa!

Hope this thread isn't coming to an abrupt end. Been ghosting this for the past couple of weeks and been very intersting and informative. Been funny to see how we all have to change a "system", this time even before it's developed. I thought we were meant to stick with 0.6 and no slippage or brokerage for current annalysis and try and fine tune the exit side. (then add the costs) Not that I'm complaining as now I've learn't what the Pro's can pay for their brokerage.

That brings me to my question (now I'll sidetrack here). Those 3 combined trading system results for past 10 years you posted Nick, what size account (%return) does that convert to and is it heavily reliant on your low brokerage, positive slippage rate (lots of short term trades)? Or is it retail trader friendly. (eg: whats its average points per contract). Hope these questions aren't too personal for you, just enjoying the learning process here.

Looking forward to developing this thread further on another futures contract as I'm a chronic bear and need the oppurtunity to go short. ASX stocks generally too hard to short comfortably. (ASX Warrants & Options too thinly traded)

Ligier

IP: Logged

zipzap
Member
posted 15-10-2000 03:12 PM     Click Here to See the Profile for zipzap     
Nick

I have been reading ththis with great interest. I also trade the SPI(although I have only just started) and one of my systems is a volatility breakout system. As mentioned above the average trade is quite low and after brokerage the system in the form from above may not be suitable for the general private trader.

The idea then is to increase the average winning trade so that brokerage has minimal impact. I pay $20 each way and $2500 margin.
In supercharts the settings ar therefore brokerage $40 slippagae $0 and margin $2500. This has been included in the system.

Filtering out the small trades. Firstly a $400 stop may be to small. I use a $1500 stop. The exit remains at first profitable open. Enter long at next bar open+((h-l)*.7)
Enter short at next bar open -((h-l)*1.6).

A further filter is used using Larry Williams emthod of Day of Week and Month.
For January find out the days of the week since 1983 that this system performed well.

Example
The supercharts code is IF: Month(date)=1 and Dayofweek(date)=5 or DayofWeek(date)=1

Then Buy on stop
Open next bar + ((h-l)*.7)

You'll find that your system will improve dramitically. Simply test each month and find out the best days of the week to trade.

I hope the image I have pasted works ok.

Here are the short trades

Cheers
Zipzap

p.s. who said this post was getting stale

IP: Logged

Nick Radge
Moderator
posted 15-10-2000 06:00 PM     Click Here to See the Profile for Nick Radge     
Thanks zipzap and nice to have another new person on the forum. Two things strike me with your comments. Firstly, the differing expansion levels you use for the short trade and the long trade. I'm not sure why you'd use two except that optimising may have been used. Not sure, but I welcome your comments.

Secondly, the day of the week strategy. I haven't bothered to do too much testing on that as I always thought it was entering into the realms of curve fitting, but maybe it warrants some work from my part.

Thanks again and we'll have a look at the results once I've re-coded.

N

IP: Logged

This topic is 3 pages long:   1  2  3 

All times are EST (Aus)

next newest topic | next oldest topic

Administrative Options: Close Topic | Archive/Move | Delete Topic
Post New Topic  Post A Reply
Hop to:

Contact Us | Reef Capital Coach Home Page

Copyright Reef Capital Coaching Pty Limited

Powered by: Ultimate Bulletin Board, Version 5.38d
© Infopop Corporation (formerly Madrona Park, Inc.), 1998 - 1999.