What do you reckon about these results?

[Harro26]

Not sure if I understand correctly, but why wouldn't you just have your system buy in the opening auction? If you put a limit order in, you would just be filled at the opening price in the auction up to your limit price would you not?

Or am I completely missing something?

Hi Razza,

This is still a WIP. That may turn out to be the solution but what I have found is that if I set a relatively tight limit I may miss out on the trade. If I set a loose limit I may significantly increase/decrease the open price as I am often traing small - mid cap stocks without a huge amount of volume and often a large bid/ask spread.

 

[Wysiwyg]

That's interesting, my data has a volume on the 29th of 20 650 000 with a 100 day MA of 11 748 000. I'd like to think premium data is more accurate than Yahoo data but perhaps not. My results show an open and buy prcie of 12.5c on the 30th which is obviously what my data says also, not sure what you were saying WRT that.

Thanks.

Patience will hopefully bear fruit.

The test you sent me is as follows with SDL.

IRE Long 3/30/2007 6/19/2007 17.34% 6657.23 4872.95 38788.7 296267.93 54 123.28 -1.76% 0/0
7.96 9.34 17.16% 23.12%
SDL Long 3/30/2007 6/19/2007 292.00% 113110.34 310310 38788.7 409378.27 54 2094.64 0.00% 0/0
0.13 0.49 291.61% 300.00%

13 cents on the trades list. Incidentally my tester took the IRE trade on the 29th.

 

[Wysiwyg]

As you can see with the previous screen shot of IRE, a trade without set trade delays, how wrong it is.

This screenshot is with a 1 day delay. So the signal comes in on the 29th EOD data and the trade is made at open price on the next day, the 30th.

 

[Harro26]

When I open the trade list it shows the entry for SDL @ .125, I think you may have excel set to display to 2 decimal places, if you right click on the cell and select format cell, set it to number and set the decimal places to 3 it should show you .125.

Commsec charting data shows the same volume on the 29th for SDL as I have (Doesn't display the MA of volume though)

Taking a look at the sample of trades that you have displayed it looks as though all of them except deep yellow are in the ASX 100 (And deep yellow is a fairly common share), are you sure you are backtesting the entire ASX? This would also explain the lower exposure you have.

Anyway, i'm stumped as to why the difference in results. Hopefully a few others will post their results and help solve the mystery.

Regards,

Harro

 

[Wysiwyg]

When I open the trade list it shows the entry for SDL @ .125, I think you may have excel set to display to 2 decimal places, if you right click on the cell and select format cell, set it to number and set the decimal places to 3 it should show you .125.

I am viewing the data you sent in .htm . The screenshot below is the Trades List with SDL on.

Taking a look at the sample of trades that you have displayed it looks as though all of them except deep yellow are in the ASX 100 (And deep yellow is a fairly common share), are you sure you are backtesting the entire ASX? This would also explain the lower exposure you have.

Anyway, i'm stumped as to why the difference in results. Hopefully a few others will post their results and help solve the mystery.

Regards,

Harro

Yes I am using the list of symbols you posted. All 2002 of them. Might have a look at this problemo another day hey. Thanks for your friendly manner.

 

[captain black]

I'd like to think premium data is more accurate than Yahoo data but perhaps not.

Premium Data is more accurate.

Yahoo data is unadjusted. Backtesting with it will give inaccurate results. Premium Data and other good data suppliers adjust their data for corporate actions etc. This thread explains it:

https://www.aussiestockforums.com/forums/showthread.php?t=18057

 

[Superboot]

Heya Harro,

I have just started trading a trend system last month so I am very interested in your post.

Some of the actions I have been doing recently include:

1. Performing a 6 monthly breakdown of my backtested results (and equity) from 2000 to now to give an indication of the likely drawdowns and expected returns a trend system might deliver - this gives me confidence that a -15% or 60% return are possible returns over 6 months

2. Analysing my backtested results (i.e. charts) focusing on the attributes of the 'losers' and 'winners' to help tailor my system. Specifically, I am trying to reduce my drawdown without hurting my return.

Action 2 is particularly time consuming as I will come up with a common 'attribute' and then backtest variations of it. I am getting there, slowly but surely

Cheers

 

[Wysiwyg]

That's interesting, my data has a volume on the 29th of 20 650 000 with a 100 day MA of 11 748 000. I'd like to think premium data is more accurate than Yahoo data but perhaps not. My results show an open and buy prcie of 12.5c on the 30th which is obviously what my data says also, not sure what you were saying WRT that.
That is correct for the signal!

Hi again,

On further investigation of why my tester does not take SDL (or many of the trades in your test) I have found that my data on this occasion does not meet all the criteria.
On the 29th. March 2007 for SDL I have volume of 20 650 300 with a 100 MA of 11 685 300 which is correct for the signal as yours is. My XAO (All Ords) Index MA10 is above MA30.

There is one criterion that I hadn't looked at. The V1 = EMA(C*V, 50) > 25000;
I changed 25000 to 15000 and bingo, SDL was taken as a trade on the 2nd April which means my data indicated a MA cross on the 30th. Again this is a data issue. This brought the AR to 9.08%

I still can't understand how the results can be so far apart but I now know about the data issue. As can be seen with this shot of the SDL trade, my tester took exactly 197,000 less shares compared to your test result. But that is another twist to the saga.

 

[Wysiwyg]

What are your settings here?

At zero I can take as many traded shares for that day as my percentage of equity will allow. I use 10% which means I can only take a maximum 10% of traded shares for that day. Set at zero I get a huge difference in results but this is not practical because one cannot expect to take all available shares on trade open day. 10% of available shares is a nominal figure in which I believe in practice I would get my order filled from available shares on the day. This may or may not be the case.

This is the catch with back testing in which the total traded shares for the day is the figure we work with. In reality at open, the numbers may not match up. What are the number of available shares at open? We don't know!!!

 

[Harro26]

What are your settings here?

At zero I can take as many traded shares for that day as my percentage of equity will allow. I use 10% which means I can only take a maximum 10% of traded shares for that day. Set at zero I get a huge difference in results but this is not practical because one cannot expect to take all available shares on trade open day. 10% of available shares is a nominal figure in which I believe in practice I would get my order filled from available shares on the day. This may or may not be the case.

This is the catch with back testing in which the total traded shares for the day is the figure we work with. In reality at open, the numbers may not match up. What are the number of available shares at open? We don't know!!!

The difference in position for SDL that you mentioned a couple of posts above is due to the equity at the time, as my backtest was giving a greater annual return I had greater equity and my entry into SDL was 10% of this as I imagine yours was also.

I had my settings the same as yours but noticed that often my cash was not being drawn down 100%, this is due to max open positions and limit trade size as % of entry bar volume. The second is often limiting the position size and as only 10 open positions were allowed my account was not being fully drawn down. Replacing the top 3 lines of the formula:

PosQty = 10;
SetOption("MaxOpenPositions", PosQty );
PositionSize = -100/PosQty

To simply show:

PositionSize = -10;

And setting max open positions to 100 gave an extra 2% annual return (~21%)

That said given the light volume of some of these stocks your slippage would probably be large. If I was to trade this system with $100K I would change the V1 indicator to V1 = EMA(C*V, 50) > 100000; or the like.

Backtesting this reduces the results to an annual return of 12%

 

[Wysiwyg]

The difference in position for SDL that you mentioned a couple of posts above is due to the equity at the time, as my backtest was giving a greater annual return I had greater equity and my entry into SDL was 10% of this as I imagine yours was also.

Yes that is the reason.

limit trade size as % of entry bar volume. The second is often limiting the position size and as only 10 open positions were allowed my account was not being fully drawn down.

As I pointed out the % of entry bar volume, in other words total number of shares traded on the entry day, is the trick in back testing.
The back tester uses that days volume before that days volume is known at the day open!! At the open on the trade date, who knows how many shares are available.
It is certainly not the volume for the whole day. That is the trick. Is there a way around this I wonder?