Developing my trading plan

[Dan_]

Well after many pages of documenting my thoughts into a rough trading plan the time has come to test and develop my system. I'm about to purchase Amibroker & will be looking for some data to back test my theories with.

I have had a look at Bodhigold data as they have an ASX starter pack which for $99 provides you with 3 months of EOD data and 20 years of historical data. Due to cost being a limiting factor at the moment I was wondering if anyone has had any experience with their data and can provide any feedback?

Also whilst I know that there are many threads on trading plans and testing if anyone could offer any suggestions (not advice ASIC ) on what thoughts or theories they had whilst developing their plan would be fantastic. I'm not looking for details of anyone's particular trading plan, rather experience in walking through the development phase

As an example I will first test my theories/methodologies to see if I can meet my satisfactory rate of positive expectancy. Then I’ll look at the variables which will affect opportunity.

One area I am unsure of is "tweaking" of the system. I don't wish to wander off on a tangent and loose focus of the base system so any thoughts in regards to tweaking (e.g. 1 change at a time for a max of 5 variables then return back to the basic system and deviate on variable 2 for a max of 5 times etc..) would be helpful

Thanks in advance for any suggestions and support

 

[GreatPig]

Dan,

Just one comment about tweaking, or optimising, a system: you need to be careful that you don't end up on the pinnacle of a mountain.

By that I mean you need to look at how much changing the variables affects the result. Your system might show a return of 30% pa with the optimised figures, but that could drop to almost nothing - or even negative - with a minor change in one of those variables. You want your final system to be as insensitive as possible to changes in the variables.

For example, if you run an optimisation in AmiBroker (over a year, for simplicity) and the profit results start listing something like this around your optimised value:

2%
-5%
30% <-- optimised value
4%
20%
-10%

then the chances of that optimised result returning 30% in the future is not good.

Something more like this would be preferable:

24%
26%
27%
30% <-- optimised value
28%
27%
25%

All just my personal opinion of course, and should not be taken as any kind of advice.

Cheers,
GP

 

[Dan_]

GP,

Thanks for the thoughts I'll definitely be optimizing with this in mind. With your expertise in AmiBroker I'm sure I’ll be bothering you for a bit of assistance now and then

 

[tech/a]

Optimising should only be used if trading a singular entity.
Optimising a portfolio approach should be avoided.

Expecting a group of independant entities to conform to a statistical singularity consistantly is like expecting 50 flies to fly in a straight line every time they go from A-B.

Weekly systems are easier to design than Daily

Daily easier than shorter term IE Intraday.
Short term methods need a discretionary element to be able to react quick enough.
Reward to Risk should be as high as you can get it.
Shorter the term lower it will be.
Win Rate is higher the shorter the term.

There are 100s of elements to discuss best is to ask if you dont think something is right.
I also agree with GP strive for a method that doesnt deviate from the mean that much.
Also look at the best performing stock in your testing it maybe and often is that 1 or 2 trades make a method profitable.
Dont rely on leverage to make a method work.
IE 2% profit trading normally 20% trading CFD's

 

[Milk Man]

One thing I have done is with universe selecting in backtesting. I plan to use ASX 200 as my universe so when I backtested the system I got historical data and updated it in 6 monthly intervals in different watchlists. I then went through and tested the system using the ASX 200, current at the beginning of the test period and only used that up to 12 month timeframes.

Is there a way of automatically swapping watchlists or add/removing stocks at certain dates? That would make it a lot easier than doing numerous tests then averaging (as I am now).

If you use the ASX 200 of today for backtesting your sure to get somewhat inaccurate results because it should pick the big movers that werent on the list at time of purchase. If your universe reflected that volatility youd need to highly modify parameters, even key components.

PS- I can probably post the lists if anyone wants them. Might save you some time: your shout at the pub in the Bahamas though.

:bier:

 

[happytrader]

Heres some commonsense stuff

How many trades can you easily handle at a time?

Trade one at a time to start - one stock, worse case scenario - one drawdown at a time.

When is the best time of the day over all to trade this stock - am, pm? look on intraday charts and get specific.

When is the best time of the day over all to take profits on this stock and what percentage is that likely to be - am, pm? Look on intraday charts and get specific.

When will I know this trade is not going to plan and at what percentage will I take the loss. Look on intraday charts and get specific.

Check out the fundamentals of this stock. Is there a profit report or agm or dividend coming up. Check it out and get specific.

Cheer
Happytrader

These are only my thoughts and should not be taken as financial advice or recommendations. I could be a raving looney for all you know.

 

[kaveman]

Dan
If you have the ASX200 set up in various watchlists corresponding with the index for periods you can add conditions to your buy/short signals. Just make certain the stocks list you do the backtesting includes all the stocks from all the lists.
this assumes you have the lists in annual update on 31 march each year.
This should be enough idea to get started with. There are various ways to do this, here is one.

buy = OrdinaryBuyConditions and (
( datenum()>1030331 and datenum()<=1040331 and inwatchlist(1) ) or
( datenum()>1040331 and datenum()<=1050331 and inwatchlist(2) ) or
( datenum()>1050331 and datenum()<=1060331 and inwatchlist(3) ) );

 

[Dan_]

Thanks to everyone for your feedback and thoughts. I have printed this out to digest whilst playing with my plan.

Lots more steps to go but I look forward to journey.

Thanks again

 

[doctorj]

The above posters are correct in the warnings. Optimising is a very powerful tool, but only when used correctly. If you optimise too much, or as Greatpig suggested ignore the reality behind what you're optimising, you run the risk of making a system that's fantastic at trading the historical period you're testing but useless over any other period.

Optimizing is useful when looking for trends. Lets say there's an imaginary world where you've noticed that a system using MA crosses might be useful. You could use optimisation to determine which MA's are best suited to this. Lets say you note, that there is a general uptrend in the returns on the system up to a 34 period MA but then it begins trending down again. That is good use of optimising (if using two variables, the charting function is very useful to help visualise trends).

Another useful technique is to optimize over one period (say 1/1/2005-current) and then backtest over several other periods that do not include the one you optimized. It's also worth looking at the individual trades when you're backtesting to see if the profit is due to a big outlier you couldn't normally expect (say 1 trade with a 1000% return could make an otherwise unprofitable system look attractive) - but that's getting off topic and is largely dealt with through montecarlo.

 

[It's Snake Pliskin]

Where would you place contingencies in a trading plan? the contingencies to handle emergencies.

 

[tech/a]

In Risk Management

 

[It's Snake Pliskin]

In Risk Management

The risk of your computer breaking down, of a blackout, of someone stealing you computer etc.?

 

[tech/a]

The risk of your computer breaking down, of a blackout, of someone stealing you computer etc.?

Hmmm.....
How you would handle adverse moves.
Setting stops--% of capital stop or % move stop.--A trailing Stop perhaps.
Position sizing,number of positions held.

Leveraged--un leveraged----

Its an involved topic

Mastering Risk by Mike Lally is a good publication worth the read.
ISBN 0 7016 3667 X

 

[emma]

Yep, I have had all of the above happen to me and none of them made me particularly happy - but the world didn't come to an end! But don't think they can't happen to you.

 

[Milk Man]

Seems like a great idea to incorporate into a trading plan! Seems like these guys really know what theyre doing.

www.rlmbusiness.com/PageSystemPhilosophy.asp

Might be a good idea to implement a bit more optimization and keep changing the variables slightly.

 

[happytrader]

Why don't we stop beating around the bush? A trading plan will only get you so far. The reality is YOU must prepare yourself psychologically and emotionally to cope with ongoing RISK and UNCERTAINTY. The real battle is within.

I've known quite afew trading course 'junkies' over the years. All were on a search for the holy grail. Some had even been mentored by the 'guru' to no avail. I'm not joking when I say their are definite behavioural similarities between trading and dieting.

No plan will ever work for you until you DECIDE to do whatever it takes to let it work.

PS I notice sport and peak performance psychologists are doing a roaring trade up here with 3 month waiting lists.

Cheers
Happytrader

 

[It's Snake Pliskin]

Why don't we stop beating around the bush? A trading plan will only get you so far. The reality is YOU must prepare yourself psychologically and emotionally to cope with ongoing RISK and UNCERTAINTY. The real battle is within.

I've known quite afew trading course 'junkies' over the years. All were on a search for the holy grail. Some had even been mentored by the 'guru' to no avail. I'm not joking when I say their are definite behavioural similarities between trading and dieting.

No plan will ever work for you until you DECIDE to do whatever it takes to let it work.

PS I notice sport and peak performance psychologists are doing a roaring trade up here with 3 month waiting lists.

Cheers
Happytrader

How true Happytrader!

A plan is extremely important for blocking out the emotions. Without emotions you can begin to master the psychological aspect.

Snake

 

[tech/a]

Happy/Snake.

Well I agree only to the point of dissolving/diminishing the emotion and the risk--do that and you wont have a problem.

If your emotional about any trade then chances are your,
(1) Over trading
(2) Trading a position which is over sized.
(3) You dont understand or havent quantified the risk/s.

As an example If you have a single trade of $20K on say a small cap like TOX and your nett worth is the $20K then your stress is likely to be higher than someone with the same stake on the same stock who has a net worth of $1 mill.

Secondly defining "to do whatever it takes to let it work." and understanding what parameters the "plan" should fall in while your doing and what parameters the "plan" shouldnt fall outside of will again go a long way in killing off that fear.

But I will agree that even when people know all the above--second guessing and "Instant" improving (Tampering) with proven methodology whilst being applied is plain suicide.

 

[It's Snake Pliskin]

What are your thoughts on trading the plan and discretionary trading as a hybrid methodology? A bit like intuitive trading. Do these sorts of traders exist?
:22_yikes:
Snake

 

[tech/a]

I think with short term trading (and the shorter the term the more it applies)
Discretion needs to be an intrugal part of any trading method.
But you still need to have expectancy,and position sizing issues answered with each and every trade.If trading portfolio's.

Where as with longer term methods these can be purely mechanical.

Single stock and or index/future methods short term could be mechanised without a discretionary component.

From what I have found