Said Bitar Daily Journal

[SaidBitar]

I do not know about ranking stocks how it will perform but it sounds like some how creating your own index and trading it. Since indexes are revised every 6 months (maybe) based on the performance and company capital.
Instead i removed the price limitation from 1 to 10 and i tested on all the stocks that are above 1$ and the results are much better it gave lower DD, now i am running MCs on this but it is interesting how DD decreased and i am sure i tested this before and i found less than 10$ are the best anyhow testing on going.

[LEONARDZIR]

I ran MCS on WTT from 1/1/2006 to present, 1000 runs with 25% seed (because of selection bias) on Russell 1000 price limiters 1-10 with rules as written in Grail book.
Average Min Max
CAR 13.35 7.96 20.25
DD 25.6 36 18.82
#trades 551 522 583

These numbers weren’t as good as my rotation momentum system so I have been hesitant to trade this system and also concerned about selection bias. Perhaps ranking the stocks when there are too many signals for the amount of capital is the answer.

[LEONARDZIR]

Average Car 13.35,, DD 25.6 , #trades 551
Min Car 7.96 , DD36 , #trades 522
Max Car 20.35 , DD 18.82, # trades 583

[SaidBitar]

yes i have similar results on Russell 1000
and this makes sense you know Russell 1000 are the top 1000 companies based on capital and theoretically what are the chances of one company that is 10B $ to double or triple in value. this is why it makes more sense for me to use these stocks for mean reversion while smaller capital companies for trend following.

[LEONARDZIR]

Said sorry that was run on Russell 3000 not 1000

[JulianCohen]

Len Zir wrote:

Said sorry that was run on Russell 3000 not 1000

I’m also struggling to get anywhere near your results Said on Russell 3000 with$1-$10 and EMA(V,50) > 500,000

I’m getting around 12-13% CAGR and 25-30% MDD which is fine, but I get much better results on the ASX.

[SaidBitar]

now i am worried i will check again

[SaidBitar]

Please try the following:
1- instead of cross(price, top channel) use Price > top channel (it has advantages and disadvantages but apparently it is helping)
2- instead of Turnover AND Liquidity use Turnover OR Liquidity
everything else is per the book

[LEONARDZIR]

Craig pointed me to the correct module to add ranking to the WTT when there are too many signals for capital. I used roc to rank. Ran the numbers for a single run and clear improvement with CAR at 16.4% from around 13% Looks more promising. Will run mcs on WTT overnite.

[LeeDanello]

Len Zir wrote:

Craig pointed me to the correct module to add ranking to the WTT when there are too many signals for capital. I used roc to rank. Ran the numbers for a single run and clear improvement with CAR at 16.4% from around 13% Looks more promising. Will run mcs on WTT overnite.

What module is that in?

[LEONARDZIR]

Module 2, section 17

[JulianCohen]

Said Bitar wrote:

Please try the following:
1- instead of cross(price, top channel) use Price > top channel (it has advantages and disadvantages but apparently it is helping)
2- instead of Turnover AND Liquidity use Turnover OR Liquidity
everything else is per the book

Sorry for late reply Said, I had a corrupt data database yesterday and it took me all day to sort it out.

When you are referring to top channel do you mean the breakout point? I am buying on H +.01

[LeeDanello]

Len Zir wrote:

Module 2, section 17

Len, I can’t find it. Can you provide a link please

[LEONARDZIR]

Said, my code was testted with high>channel of highs as coded in the course not cross(price,channel).
Did mcs test 2006-2016 as noted above with ranking added and no significant change in CAR or DD., 13.4%,27%.
WTT does not seem as robust in US markets as rotation which has a higher CAR and more consistent returns.

[SaidBitar]

I agree about the results for the duration of 2006 – now.
I normally test from 1995 or 2000 the longer the duration the more trades the clearer the image.
maybe rotational system will have better results but also it will have more trades for the same duration.

[Author]

Posts