Arb,

I've created an algo that tests for the following qualities:

. . . Relative Strength Indicator (RSI) higher than that from 5 days
ago. . . . Closing price is higher than the average in the past 30
days. . . . Volume at least 120% of that from the last 2 months. (The
120% is arbitrary. I used it because it seems to work) . . . Average
volume for the last 2 months at least 100,000 shares (note that
TC/2000 reports shares in hundreds). . . . Price today higher than 30
day average . . . Price today higher than in the last 3 weeks

Not quite sure what to do about the different stages of the stock, but this seems like a good starting point! You can put in your order logic in

if False in context.w_checks.values():  
        pass  
    else:  
        # Order here  
        pass  

-Seong

Hi Seong

This is awesome. Thanks for the support. For now the order can run the same as in the description provided. I am still just learning the platform and so I am likely to be a bit slow at first. I am a C# guy so taking a little while here.

Arb

Arb,
You got my interest. I'm reading Stan's book now. He does a great job of giving simple explanations.
In the first few chapters, he advises waiting for the 30 week moving average to be flat or positive to get a good stage 2 entry, so I'm trying to figure out how I can tell. To do so, I need to calculate the average at least a few days back and compare it. If I figure it out, I'll post it.

Rich

Hi Richard

It is a great read. The markets have been so fundamentally driven in recent weeks that everytime a talking head said anything about Syria, debt ceiling etc volatility surges. Hopefully things will now settle. Let me know what you are thinking.

Andy

Arb and Seong,
Could you please indicate where in Weinsteins book did you find the algorithm? Thanks in advance,
/Omar

Hey everyone,

I have been following Weinstein's method for months now and have realized this algo is slightly differentiated in a few ways from the methods he states in his book.

First and foremost I noticed that the RSI being used in this algo (RSI Wilder) is completely different than that mentioned in the book (RSI Mansfield). This indicator takes the RS Dorsey formula and breaks it down to compare it to the strength of the S&P 500. I built the indicator in my Thinkorswim platform, but I do not have experience with python yet so all I can do is provide you all will some material which can be used to code this into the algo.

http://www.chartmill.com/documentation.php?a=398&title=Relative%20Strength%20(Mansfield)

Second, what he does is screen the stocks to find which industry is outperforming the SPX and then finds stocks within those sectors that meet the criteria that has been set (Volume increase, RSMansfield Uptrend, Close above 30 and 10 week MA's). This usually can help indicate which stocks have entered Stage 2 or are already in Stage 2 of his technique. This can hopefully help you guys while I learn Python and begin to build this on my own eventually.

I have also written code for the TOS platform using colored bars on my charts utilizing the MA's that shows me exactly when the Weinstein method is in play, when I should be cautious, and when I should be out.

I definitely think this can be automated if the metrics are entered correctly. My only concern is sometimes its hard to confirm the different stages until you see them with your own eye.

Hey Jaan,
This is precisely what I suspected, but I wasn't sure. Thank you for confirming it and for the explanation. To write a function in Python calculating the Mansfield and Dorsey relative strengths is very simple. The problem is implementing it in the whole strategy. Weinstein measures the relative strength for each stock in relation to its sector. This means that we should use a different index for each sector and only calculate the relative strength for the securities in the sector using the respective index. Therefore, we have to partition the market universe into several sub-universes representing each sector, choosing an index that represents each sector and commence with calculating. I, unfortunately, don´t have an idea of how to do that, since we have no way to know which sector from the symbol of the security. Furthermore, as you mentioned, Weinstein´s definition of phases is somewhat subjective. Due to all of the above, I don´t think that it is wise to develop a fully automated algorithm to trade Weinstein´s methods. Maybe the optimal way to use the relative strength is as an indicator in stock charting tools.

Hi All

I have finally come around to wanting to action this algo and am looking for someone to help me code this up - initally as a screen so we can backtest and then to work with me on some specific market ideas. Anyone have the time to help me do this...?

Arb

@Omar, couldn't you use the 11 sector ETF's as an index for the sector? Here is a thread that has the sector to ETF mapping.
And now that we have fundamentals you use the fundamentals.asset_classification.morningstar_sector_code to map a stock to its sector!

What other problem needs to be solved here?

Paul Perry, It seems to be a good idea. However, I still need a list of all the companies divided into sectors. I don't know if this is possible in Quantopian. I will look at the thread you sent and see if it is any help in this point. Thanks for the tip.

Hi does anyone have a script in tos or algo of any kind to share?