[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[EquisMetaStock Group] Re: Twardy...Gain Risk Index



PureBytes Links

Trading Reference Links

Kees,

I believe this is right.

STEPS:
1. Determine the Linear Regression Trendline for the last 15 days?
the best fit data
2. Subtract the Close from the best fit data?becomes the residual
3. Determine the standard deviation of the residual
4. Determine the percentage gain over the last 15 days using the 
best fit data
5. Divide the percentage gain by the standard deviation of the 
reisduals?becomes the Gain Risk Index(GRI)

{Gain Risk Index}
{Based on a TASC Oct.2004} 
{article by John Twardy}
{Written by Preston Umrysh}  
BFD:=LinearReg(C,15,SIMPLE,15);
SDResidual:= Stdev(BFD - C,15);
GRI:=ROC(BFD,15,%)/SDResidual;
GRI; {end}


enjoy,

Preston



--- In equismetastock@xxxxxxxxxxxxxxx, pumrysh <no_reply@xxx> wrote:
>
> From TASC:
> 
> TRADING SECTOR FUNDS USING STATISTICS 
> 
> Editor, 
> The October 2004 article by John Twardy, "Trading Sector Funds 
Using 
> Statistics," piqued my interest, and the sidebar on determining a 
> trendline by least squares was very helpful. I do have a request 
for 
> a couple of clarifications. 
> 
> The gain/risk index is to be computed for each sector fund for the 
> last 30 days. The next step is to "rank the funds according to 
this 
> index." Please explain how the ranking relates to 30 days of 
> gain/risk index computations. Is a 30-day average to be used? 
Also, 
> there is reference to the value of the "sum of the gain/risk index 
> values for all funds for any given day makes up the sector trend 
> index. This gives you the overall trend of the market, which is 
> helpful for this short-term analysis." How is this value 
> specifically used when selecting the highest-ranking fund? 
> 
> Pete Bock 
> via email 
> 
> John Twardy replies: 
> Only the last 15 days of data is used. The article should have 
> stated to "do this for each of the past 15 days," not 30 days, 
> although this really doesn't affect the results, as only the most 
> current 15 days of data is used. 
> 
> The sector trend index is not used for selecting the highest-
ranked 
> fund -- it just gives you a "feel" for the market trend. 
> 
> MORE ON TRADING SECTOR FUNDS 
> 
> Editor, 
> After struggling for the past few days unsuccessfully to replicate 
> the gain/risk index shown on page 84 of John Twardy's October 2004 
> article, "Trading Sector Funds Using Statistics," for FIDSX (6.2) 
> and FBMPX (18.7), I wondered how would be the best way to ask 
> questions that might help me to successfully replicate Twardy's 
> work. I thought the most direct way would be to send you my 
> spreadsheet [not shown--Editor], which I produced based on my 
> interpretation of the instructions you gave on page 83. 
> 
> The attached spreadsheet has two tabs, one labeled FBMPX and the 
> other labeled FIDSX. I have reduced the dataset to only that data 
> which should be necessary to produce a gain/risk index result for 
> August 26, 2002. 
> 
> Any help you can provide would be greatly appreciated. 
> 
> Axel Gumeson 
> via email 
> 
> John Twardy replies: 
> The discrepancy appears to be because you have used adjusted Nav 
> data instead of closing data -- that, and it appears that some of 
> your historical data has been corrected, which resulted in slight 
> differences in our calculations. Taking all this into account, 
your 
> spreadsheet now duplicates the results I get with the adjusted 
data. 
> 
> 
> 
> -------------------------------------------------------------------
--
> -----------
> 
> METASTOCK CODE FOR TRADING SECTOR FUNDS 
> 
> Editor, 
> I have been a long-time subscriber to Stocks & Commodities. I 
> especially enjoy the articles that discuss trading methods. 
> 
> In the October 2004 article "Trading Sector Funds Using 
Statistics," 
> I had a couple of questions for the author, John Twardy. First, I 
> wanted to better understand the logic behind the gain/risk ratio. 
> 
> The way I understand it, the highest-performing 15-day sector with 
> the lowest standard deviation to a 15-day linear regression line 
is 
> the one to select. A high return with low risk -- that sounds very 
> good. But why 15 days? And could it be done on weekly data? What 
> kind of testing was done to end up with 15 days? 
> 
> Second, I attempted to create a MetaStock formula, but I was not 
> able to fully understand the basis for the formula in the article. 
> Here is what I tried: 
> 
> ROC(c,15.%)/(Sqrt(Var(Linearreg(c,15),15))) 
> 
> Unfortunately, I was not able to get the same results as in the 
> article. I know that you sometimes get the software vendors to 
> submit code. Any help would be appreciated. 
> 
> D.J. Vatalero 
> via email 
> 
> John Twardy replies: 
> I looked at various time periods -- 30 days and 15 days -- to find 
a 
> system that would trade often enough to keep up with market trends 
> but not so frequently as to create seesaw trading. In my testing, 
> the 15-day average outperformed the 30-day average. Taking the 30-
> day minimum holding period (for the Fidelity fund) and looking at 
> the performance over half the holding period helps to show the 
trend 
> that is hidden in the longer-term average. 
> Regarding MetaStock code, sorry, that's not something I can help 
> with. 
> 
> 
> 
> 
> 
> 
> 
> --- In equismetastock@xxxxxxxxxxxxxxx, "ktakkenberg" 
> <c.a.takkenberg@> wrote:
> >
> > HI, I am looking for a Metastock formula probably called The 
> Twardy 
> > Indicator?
> > In  October 2004 there was an article "Trading Sector Funds 
Using 
> > Statistics," by John Twardy, in Stocks & Commodities page 82.
> > Normally Equis supplies the formula with the article.But not 
this 
> time.
> > On the internet I found an email to S&C where  some-one wrote to 
> have 
> > tried to make the correct formula but was in doubt if he/she did 
> so?
> > The formula  was: 
> > 
> > ROC(c,15,%)/(Sqrt(Var(Linearreg(c,15),15))) 
> > I would like to know if this is the right-one or if anyone knows 
> the 
> > correct "Twardy Indicator"?
> > 
> > Thanks for helping me out!
> > 
> > Kees Takkenberg
> > The Netherlands.
> >
>




 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/equismetastock/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/equismetastock/join
    (Yahoo! ID required)

<*> To change settings via email:
    mailto:equismetastock-digest@xxxxxxxxxxxxxxx 
    mailto:equismetastock-fullfeatured@xxxxxxxxxxxxxxx

<*> To unsubscribe from this group, send an email to:
    equismetastock-unsubscribe@xxxxxxxxxxxxxxx

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/