Reducing Entry Signals

[LeeDanello]

Said Bitar wrote:

first run backtest in Amibroker without any commissions and set the maximum number of positions to 200 (for example) and position size to 1% the aim is to get the maximum number of possible trades. then save the CSV file.

I’m assuming you’re explaining this for a 2:1 leverage? So for an unleveraged account would this be 100 for 1%?

[TrentRothall]

I think the idea Maurice is just to set it high enough to capture ALL POSSIBLE trades regardless of position sizing. I believe you do the position sizing in the spreadsheet.

[ScottMcNab]

I went down the excel path Said. I am interested to try your file out. The excel I have used selects the worst 25 or 30 trades (depending on max positions) for each day

I found the results quite interesting…all tests 1/1/2011 until 31/12/2016

one of the SP500 system I have been trading with 25 positions at 8%
Amibroker backtest
exposure 25.57%
CAGR 37.26%
took 73% of trades (9557/13112)
excel results when select 25 worst trades each day CAGR = 29%

new SP500 system with tighter entry (added volatility line in buy setup) and 30 positions at 8%
Amibroker backtest
exposure 12.36%
CAGR 22.90%
took 83% of trades (5583/6762)
excel results when worst 30 trades selected each day CAGR = 21%

So certainly lesser impact with lower selection bias.

[ScottMcNab]

Not sure how to post excel file for others to check/test/confirm sorry

[ScottMcNab]

Hmm…. formula used for CAGR was ((L17000/L1)^(1/5))-1)*100 when it should have been
((L17000/L1)^(1/6))-1)*100…counting to 6 was my downfall

corrected results for excel worst possible outcome for the above 2 systems drops to 24.3% (from 37%) and 16.2% (from 22%) respectively

[SaidBitar]

I will try the excel as well just to verify the results of this program
but from the snapshots i posted the worst scenario was really bad for me ended up giving all the money that was made in the early 2000 until 2009 when i tested from 2011 till end of 2016 the worst was 40% underwater

this is why i need to verify with excel . if excel showed the same results then i need to get worried

[SaidBitar]

I did it again in excel and i got the same results, maybe because i used the same logic.

not good

[SaidBitar]

I ended up testing almost all the MOC strategies my conclusion is that if I have the worst luck to pick the biggest losers everyday it is better to stay far from the stock market

[ScottMcNab]

I have a rounding error somewhere. I have used the buy and sell prices for every trade from the csv backtest export and calculated a cumulative equity to calculate profit and loss for each trade….starts off matching exactly or within 1 cent for first few weeks but then before long profit/loss in Amibroker for a position may diverge from my excel spreadsheet by $1…etc.

Can I email you the excel file to look over the logic Said ?

[SaidBitar]

Sure
[email protected]

[TrentRothall]

Said Bitar wrote:

wow it is not allowing me to attach the file it is only 3M and zipped

still it is saying that the size is huge

anyhow here is the link to download it

https://www.dropbox.com/s/7byni28fxm202od/MCs%201.0.0.zip?dl=0

Hi Said,

When i click on the link it just goes to a blank page? any idea what’s happening…

[ScottMcNab]

Said Bitar wrote:

Sure
[email protected]

Hi Said,
I wont waste your time…wasn’t a rounding error…once found mistake in excel the better system on SPX I had (with worst possible outcomes selected) returned (over 2011-2016) only +5%….the other system was -4%

[JulianCohen]

So I’m a little confused here…why would this “bad luck” situation not occur in MCS testing?

Or would it occur if you chose to do 100,000 tests?

[SaidBitar]

Because to get this specific combination the probability is very small.

[ScottMcNab]

I guess the more MCS done the higher the chance of it occurring ?…the probability of the MCS picking the worst or close to the worst possible outcome each time would be low

Not surprisingly I guess, the ASX system which captures in the order of 94% of trades (22%CAGR in amibroker backtest 2011-2016) still had reasonable “worst case scenario” in excel with CAGR of 14.7%

[Author]

Posts