Developing a System to Trade Live

[RodC]

Although it was mentioned earlier that this is a good exercise for those who would like to start their own system later on we should really concentrate on starting one system now. Once all is in place we can start a new thread to show the system being traded live. I think it defeats the purpose of us putting our heads together if we go off on our own at the first opportunity. I assume we will be just testing one system for the first few months of operation before starting with real money.
So basically GP I think we are looking to create one system which we can start with and later people can modify it if required.

Yep, that's how I see it as well.

There's a couple of aims to this exercise:

1. To develop and trade one system live,
2. To learn the process of system development for those that wish to take it further themselves later.

Rod.

 

[cabbie]

With the main goal to maximize Expectancy & Opportunity and avoid Risk of Ruin etc as presented by Tech/A, Van Tharp & others.. ?

How to get there is what I'd like to know.

 

[RichKid]

With the main goal to maximize Expectancy & Opportunity and avoid Risk of Ruin etc as presented by Tech/A, Van Tharp & others.. ?

How to get there is what I'd like to know.

I guess all that will be in the numbers once we get these preliminary issues sorted out. Do a search on those topics you mentioned using the ASF search tool and you'll find weeks of reading!

We'll be coming to those topics eventually so you'll have time to read up. I don't know much about it all myself but I'm learning.

 

[mit]

I'm unclear on how the ultimate live-trading part of this will work.

Is the intention that individuals will each trade their own version using their own capital? If so, will each individual be using different parameters and thus selecting different stocks so that everyone's not trying to make the same trades in the same stocks at the same time?

GP

On Reefcap it basically worked as a thread where it was synthetically traded by a volunteer. Some people did trade. But we must stress that it is an educational exercise and not a set of recommendations. Depending on the system. If it is a medium term system we may still have 100 possibilities to pick from and a number may come up simultaneously. I trade two systems that basically have the asx200 as their universe and I get out of step with the backtest so I think it would be okay.

If we pick a smaller number, they would probably be the more liquid stocks anyway and I think our combined might would not shift it. Although we might keep that in mind and make sure that the liquidity is greater than say $500k per day.

So ASX200 is the first step, still a huge number to cull.

Not with Amibroker and and Excel . I think the problem is getting a system that looks like working.

MIT

 

[Milk Man]

Just got Amibroker!

$149us ($205 aus)

Bring on the pain!

Formula's are not my cup of tea but i'm sure my love of money will soon change that.

I'm quite willing to try help with culling back the universe of stocks but I may
need some help taming this new Amibroker beast of mine.

If there's enough of us with AB maybe we could start a thread on it.

 

[mit]

Now where to next? Looking at the development of T/T, it was more of a monologue with some ideas thrown into the side. Tech was a couple of steps ahead and would reveal his thoughts as he went. Tech is not here anymore unfortunately.

Again I think we have a number of choices. Either we throw up a few systems to discuss and vote on one of those and that gets slowly developed
OR
those of us who have developed systems can "think aloud" and develop their systems in parallel culling out the ones that don't seem to make it.

Personally I think the second way might be better. Its hard to know what will work and I think that getting agreement on every step will take forever. I have developed three systems that I trade successfully and although I have a couple of ideas for systems 95 percent of my ideas don't work. My last system, a forex system, came from misreading Larry Williams. So I am happy to start developing live and for other experienced people to develop live.

Another good part is that I use Amibroker which seems pretty popular here.

MIT

 

[RichKid]

Just got Amibroker!
$149us ($205 aus)
Bring on the pain!

Formula's are not my cup of tea but i'm sure my love of money will soon change that.
I'm quite willing to try help with culling back the universe of stocks but I may
need some help taming this new Amibroker beast of mine.
If there's enough of us with AB maybe we could start a thread on it.

I've started an Amibroker thread as I have a feeling we'll need one soon. There has been some general discussion of Amibroker in various threads so you could start with a keyword search.

Here's the new thread: https://www.aussiestockforums.com/forums/showthread.php?p=15657#post15657

 

[Milk Man]

I've started an Amibroker thread as I have a feeling we'll need one soon. There has been some general discussion of Amibroker in various threads so you could start with a keyword search.

Here's the new thread: https://www.aussiestockforums.com/forums/showthread.php?p=15657#post15657

Thanks mate,

I do try to look for what I want to know as long as ive got the patience.
I'd rather use this site for forum chat however as theres a great sense of teamwork (99% of the time anyways). If no-one here knows ill go somewhere else and try to find some answers.

 

[Milk Man]

I've come up with a couple of ideas of culling back our universe,
these seem to have a good effect:

Close above/below a certain price-
I was to lazy to manually put in the asx200 so I just included everything that traded over $500000 a day average for the previous year. This may be worth tweaking. I have been testing below certain dollar amounts aswell with some success.

Integration of a bang for buck requirment-

care of one of Tech/a's posts:

"Divide a $10,000 account by the closing price on any given day. This number is then multiplied by the average range of the stock for the last 200 days. Dividing this number by 100 converts the result to dollars and cents which inturn indicates the possible dollar return on any given day."

I modified the formula to fit into Amibroker then put a > condition on the end.
(please see Amibroker thread to check- never programmed anything before!)
therefore if a stock moves $10 average a day then in a year on a $10000 position it will move $3650 a year = moving at 36.5%

The only downside to this is that it can move down faster than up (generally) so that has to be taken into acct. Therefore if we want stocks that can move at 35% a year we may have to enter 50% to actually get that. Theoretically that is.

My problem now is what to benchmark it on. I've just been using techtrader but we don't seem to want that. Any ideas?

What do you guys think?

 

[mit]

We can let Amibroker do it for us. We can add these things as filters and find good values. Actually for the bang for buck you can set Amibroker up to once it has a number of candidates it will pick the shares in order of volatility.

Interesting comment about the going down faster than going up. The simple way to measure volatility is to use ATR. Bang for Buck is usually ATR/C. In the forex system I trade I found that directional volatility gave some interesting results ie EMA(H-O, period) and EMA(O-L, period).

Basically, the aim is to capture stocks likely to go up or stocks already in an uptrend "Buy High Sell Higher". I've found that GMMA discussed elsewhere is good for this. NOT as a trigger. I used the formula below:

AND BarsSince( Cross(MA(C,50), MA(C,45) ) ) > period
AND BarsSince( Cross(MA(C,45), MA(C,40) ) ) > period
AND BarsSince( Cross(MA(C,40), MA(C,35) ) ) > period
AND BarsSince( Cross(MA(C,35), MA(C,30) ) ) > period
AND BarsSince( Cross(MA(C,35), MA(C,5) ) ) > period
AND MA(C,50) < MA(C,45)
AND MA(C,45) < MA(C,40)
AND MA(C,40) < MA(C,35)
AND MA(C,35) < MA(C,30)
AND MA(C,35) < MA(C,5)

Notice I am only interested in the long term lines. This shows me that for the last "period" days the MAs haven't crossed so the trend is well and truly established. GMMA on the other thread was being used as a trigger and it doesn't work as well. We want the lines to already be established.
Also the lines are in the order we want. If you really want to get picky you can test that all of the lines are sloping upward. There are other indicators of trend

So we filter on
1. liquidity
2. volatility
3. In a nice trend

Another thing to look at is stocks reaching a 52 week high. Once they pass this the skys the limit as there is little overhead resistance. However, the 52 week high itself is a resitance and stocks hitting this may subsequently dive. We would need some confirmation such as the stock has kept going up once it has hit the 52 week high or we can look at a 1-2-3 formation as this increases the probability of success and gives us an initial hard stop at the 2 point.

MIT

 

[mit]

I created a new thread for the longer term system: https://www.aussiestockforums.com/forums/showthread.php?t=1709

MIT

 

[tech/a]

I seem to be working on the short term method like MIT, I've moved the discussion to a thread dedicated to a Short Term System.

Perhaps general conversation and questions relating to systems developement and testing not specific to either thread could remain here.

 

[RichKid]

I seem to be working on the short term method like MIT, I've moved the discussion to a thread dedicated to a Short Term System.

Perhaps general conversation and questions relating to systems developement and testing not specific to either thread could remain here.

Great idea. And thanks guys for the suggestions re splitting the thread. I've done it now. Please browse the Trading Tips forum and checkout the threads on testing- let me know if I still need to move or edit any posts or threads. Should be okay now as I've provided cross-references. So we now have a general systems testing thread (this one) and a med/long term thread and a short term thread. I'll add nicknames to the threads once that's sorted, I think tech already has a name for his.

 

[tech/a]

Rich..
Are you available for general tidy ups!! Man I'm gonna nick name you "Hoover".

Good Job!

 

[tech/a]

Master Keys
Secondary Keys
Minor Keys

These are the essence of systems construction. Defining these keys in order and applying them to your system design will help you in BUILDING a good system.

MASTER KEYS.

These are conditions or variables which when applied can make the difference between a system failing or being profitable.
(I have found one of these in the Short term systems developement we are currently working on).

SECONDARY KEYS

These are conditions or variables which when applied can dramatically alter a systems performance----IE Key systems aspects like---% profit,% winners/losers,Risk Reward Ratio's,Consecutive winners/losers---are DRAMATICALLY altered.
(I have also found one of these in the short term method)

MINOR KEYS

These are conditions or variables which are often found when TWEEKING a method.Optimisation could fit into this area of systems design for singular entities like futures or indexes.

I mention these here as I will be using this terminology in the threads on systems design and I want people to know what it means.

 

[RichKid]

Rich..
Are you available for general tidy ups!! Man I'm gonna nick name you "Hoover".

Good Job!

Thanks, I prefer tidy discussions. And yes, I may have to do odd jobs cleaning-up for people if my trading doesn't improve, so I might as well get the practise while I can!

 

[mit]

Great idea. And thanks guys for the suggestions re splitting the thread. I've done it now. Please browse the Trading Tips forum and checkout the threads on testing- let me know if I still need to move or edit any posts or threads. Should be okay now as I've provided cross-references. So we now have a general systems testing thread (this one) and a med/long term thread and a short term thread. I'll add nicknames to the threads once that's sorted, I think tech already has a name for his.

Thanks, you even cleaned up my mistake in naming the medium/long term thread. Although for nicknames I have been writing programs for 27 years and still can never think of catchy names for systems. I have lefts a trail of systems with unpronouncable acronyms.

MIT