Said Bitar Daily Journal

[SaidBitar]

Sometimes the market makes you wonder,

SGI had earnings last night
estimated was -0.09$
Reported -0.05$

this is 44.44% better than estimation, but still the stock dropped 35%.

I have no idea what are the earnings stuff and frankly i do not care much to understand it but losing 5 cents is much better than losing 9 cents

[JulianCohen]

Buy the rumour, sell the fact

[Nick Radge]

Nasty…

[SaidBitar]

I wanted so much to get rid of it but the system sad NO
:angry:

[SaidBitar]

I was trying to arrange the folder in Amibroker where i keel all the folders that have the AFL files, with time I am forgetting which strategy was OK and which was not. So i decided to run all of them and then to sort them according to their types and profitability.

the first strategy i checked it is one i described before somewhere in the forum 5 day momentum.

this strategy is nice and it was hidden jewel then i remembered that i discarded it before due to huge variations in returns so i modified a bit in the code to make more stable.
Modifications are changing the number of positions from 10 to 20, and assigning 5% for each position and things became good i did not expect swing trading strategy to be with these returns

any how attached are some figures of the returns.

[Nick Radge]

Looks good Said. Be interested in how that performs on the ASX

[LeeDanello]

Are these results unleveraged.

[SaidBitar]

I don’t have data for the AUS market, so I did not test it.

here are the rules
ADX( 8 ) >30
DI+ >DI-
StochK < 40 Entry stop order 1 cent above the High of the signal day
Trailing stop 0.5% bellow the Low of the signal day

if there is no exit on the TS then u exit on the 5th day

[SaidBitar]

Actually no need to test it anymore

I found the reason the results were that good. I was using ranking on this type of systems.

I changed the ranking to random to be realistic and the results became bad

I felt it is not possible to be this good, anyhow good that things are sorted out

but if the system will be used for MRV then it will be OK

[ScottMcNab]

Any way to tweak it so enter on open next morning ? May lose some profit but gain from being able to use ranking system may make it worthwhile ?

[SaidBitar]

Scott McNab wrote:

Any way to tweak it so enter on open next morning ? May lose some profit but gain from being able to use ranking system may make it worthwhile ?

Great idea

works like a charm

so instead of entering on stop order it will be market order on the open, and the ranking is on Stochk value

the results look so good so i had to double check again before posting :blush:

everything looks good for me it will be great if someone can code it and check the results

[SaidBitar]

I tried it with margin and this stuff is becoming scary

CAR more than 128% and DD 24%

:blink:

Note that i set the limit trade size as % of entry bar volume to ZERO, the reason is just to get real feeling of the performance and returns per year

[LeeDanello]

I’m interested in giving it a crack although it won’t be until the weekend. When I can confirm your results I will test it on the ASX

[Nick Radge]

The first issue I can see here is that the ranking is not accurate and therefore proper testing requires the use of the trade skipping MCS or the ranking needs to be sorted.

Method:

I ran a backtest YTD 2016. I then selected a random day where new positions had been entered – in this case 29/4/16

The new positions on the back test were:

ANIK
UIHC
X
SABR
PPC

I then ran an exploration from close of biz on the 28th. In theory the top ranked stocks should be those listed above.

They weren’t.

X
CVC
SABR
AGIO
FRC
CCE
DERM
TROW
CATY
LLY
TMUS
CCF
NBIX
ACAD
CMCSA
PPC
RLYP
RVNC
UIHC
ANIK
VASC
CKP
AFOP
PFSW

You can see X and SABR are at the top which is great. However, PPC appears at #16, UIHC appears #19 and ANIK is at #21.

[Stephen James]

The error was with PositionScore. I’ve amended some text in the course on the function as it was not as clear as it should have been on its use.

When ranking using this function it is necessary to use the value from the previous bar:-

PositionScore = Ref(RSI(14),-1); – This is correct and will match up with ranking in the explorer where you would use RSI(14) for the setup bar.

PositionScore = RSI(14); – This is incorrect as the backtester will take the RSI value at the close on the same bar as a buy signal occurred, thus causing a post-dictive error.

[Author]

Posts