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

RE: [amibroker] Re: How to calculate RS Rank?



PureBytes Links

Trading Reference Links




To 
those of you interested in how IBD calculates its 
rankings...
<FONT face=Arial color=#0000ff 
size=2> 
I 
cannot get to the books and documents that I need in order to quote the exact 
methodology.     However, the news is not all 
bad:
<FONT face=Arial color=#0000ff 
size=2> 
<FONT face=Arial color=#0000ff 
size=2>1.  The primary book that contains that information is "The Hedge 
Fund Edge" by Mark Boucher.   I can think of dozens of reasons 
why this is one of my favourite books and I urge you to have a look at it when 
you have some time.    Mark goes into a fair amount of 
detail as to how IBD calculates (or has calculated) the RS 
rankings.
<FONT face=Arial color=#0000ff 
size=2> 
<FONT face=Arial color=#0000ff 
size=2>2.  I do have my AFL code that was based on what was quoted in the 
book, but I have modified it somewhat.   From that code, I can pretty 
much confirm what Peter has offered:
<FONT face=Arial color=#0000ff 
size=2> 
12 month 
performance + 
9 month performance 
+ 
6 month performance 
+ 
two times the 
3-month performance
<FONT 
face="Courier New"> 
<FONT face=Arial color=#0000ff 
size=2>3.  The exact formula used has changed over time and I can't say for 
sure that the above is being used now.
<FONT face=Arial color=#0000ff 
size=2> 
<FONT face=Arial color=#0000ff 
size=2>4.  As I was saying on another Yahoo board, I don't think our 
objective should ever be to exactly match the numbers calculated by someone else 
when doing our research.   It is more important to come up with 
numbers that work well.
<FONT face=Arial color=#0000ff 
size=2> 
<FONT face=Arial color=#0000ff 
size=2>5.  Given the basic formula (either from Mark's book or Peter's 
email), you have enough to play with using AmiBroker and PT to come up with 
something that should work for you.   For instance, I did a 15-year 
research run on 14,000 stocks where I tried ten different lookback periods, 
weighting each lookback period with a different value.   This research 
ran on ten PC's for over six days and gave me the weightings that I 
use.    
<FONT face=Arial color=#0000ff 
size=2> 
<FONT face=Arial color=#0000ff 
size=2>6.  Amongst the papers that I can't get to right now are several 
documents written by Columbine Capital Services in Colorado Springs.   
I strongly urge that you do some searching on Google  (columbine alpha 
forecasting ranking) for some articles that you should find to be of 
interest.   Amongst the wealth of information generated by this 
excellent organisation are several articles on ranking stocks.    
If you dig very deep into what is freely available as well as write to them for 
reprints of some articles that are not so easily available, you will discover 
new and exciting ways of ranking stocks.    I won't spoil the fun 
for you, but the most interesting thing I found in Columbine's work is the fact 
that they negatively rank the performance in the most recent time 
period.    Think about that one!!    If I hadn't 
read their articles and attended one or two of their presentations, I would 
NEVER have thought of applying a negative weighting to one or more of the 
lookback periods.
<FONT face=Arial color=#0000ff 
size=2> 
<FONT face=Arial color=#0000ff 
size=2>7.  I'm happy to work with anyone in this area, but I won't be able 
to share the exact methodology that I am using.   In any event, 
what I am using may not work well with the rest of your trading systems and/or 
methods.
<FONT face=Arial color=#0000ff 
size=2> 
I hope 
this helps those of you who are interested in ranking.   You won't be 
able to do the job (as thoroughly) without AmiBroker and PT.   If you 
don't know about PT, I think you are missing the boat if you want to do any kind 
of ranking.
<FONT 
face="Courier New"> 
<FONT 
face="Courier New"> 
<BLOCKQUOTE 
>
  <FONT 
  size=2><FONT face=Arial 
  color=#0000ff> 
  <FONT 
  size=2> 
  <FONT 
  size=2> 
  <FONT 
  size=2> -----Original 
  Message-----From: Phsst [mailto:phsst@xxxxxxxxx]Sent: 
  Saturday, September 06, 2003 1:23 AMTo: 
  amibroker@xxxxxxxxxxxxxxxSubject: [amibroker] Re: How to calculate 
  RS Rank?Peter,Prior to your post, I 
  was unaware of the Nov 20, 2000 QRS Accuracysuggestion of Gary Lyben 
  (Quotes Plus Owner).Here is all that I can offer to this 
  thread:I have a handful of profitable trading stradegies that are 
  based uponbacktest resulta against my Quotes Plus database. ( No need 
  toelaborate upon degree of profitability of those stradegies 
  ).What I found was that even a marginally profitable stradegy that 
  usedany form of trend analysis would get an almost exponential boost 
  byeither adding or substituting the QP2 QRS rank figure to the 
  filteringprocess.However QP2's QRS figure is calculated, it is 
  probably the mostimportant metric that is in my backtesting toolbox (on 
  both the Longside AND the Short side).Unfortunately, the QRS 
  figure only goes back to 6/1/1995, which isapparently when it was first 
  included in the database. This means thatabout 4 or 5 years of prior data 
  simply had a QRS figure of -1... ie.four or five years of historical data 
  became useless once I includedthe QRS metric on my backtests.All 
  of the above is offered FWIW.Phsst--- In 
  amibroker@xxxxxxxxxxxxxxx, "bluesinvestor" 
  <bluesinvestor@xxxx>wrote:> In November 2000, Gary Lyben of 
  Quotes-Plus offered the following> calculation for RS ranking that he 
  said (at the time) was similar to IBD> RS ranking.  > 
  > The new calculation however was rejected by the users due to lack 
  of> consistency (prior years would have been the old 
  calculation).> > Regards,> Peter> > 
  From:  Gary Lyben <gary@xxxx> > Date:  Mon Nov 20, 
  2000  9:09 pm> Subject:  RE: [quotes-plus] QRS 
  accuracy> > All -> > Attached is a csv file that 
  contains a totally different calculation of> the qrs.> > 
  Look it over and see if it looks better than the current calculation.> 
  > This one has ACTP at 18, which is pretty close to IBD.> 
  > This is essentially the calculation> > 12 month 
  performance + 9 month performance + 6 month performance + 2 * 3> month 
  performance> > versus the current calculation of > 
  > performance 4 quarters ago + performance 3 quarters ago + performance 
  2> quarters ago + 2 * current quarter performance.> > It 
  looks like the attached calculation is better at capturing the> 
  downtrends, look at NXWX, EVTC, and MITK. > > The 99 ranks look 
  like they should be 99's> > This is a significant change in the 
  formula, so I'll send out the> spreadsheets for a while to get your 
  feedback.> > The ranks in this spreadsheet are for 
  11/20/00> > Regards> > Gary> > 
  > -----Original Message-----> From: Fred [mailto:fctonetti@xxxx] 
  > Sent: Thursday, September 04, 2003 7:32 PM> To: 
  amibroker@xxxxxxxxxxxxxxx> Subject: [amibroker] Re: How to calculate RS 
  Rank?> > By using Portfolio Trader, a generic scoring, ranking, 
  trading (GUI & > AFL)> > --- In 
  amibroker@xxxxxxxxxxxxxxx, "jtelang" <jtelang@xxxx> wrote:> > 
  Hello,> > > > Has anyone performed calculations of RS Rank 
  using AB, such as the > > one that IBD provides? Although IBD's 
  formula is secret, lets say > > that if I define the gain to be 
  simply "ROC(C, 200)", how do I go > > about ranking the security 
  against all other securities in the DB?> > > > I'm new to 
  AB, so my thought process here could be a bit off. > > Wondering if 
  anyone can think of a better way to do this than > > following 
  -> > > > 1. Grab a list of all stocks in the DB, either 
  thru automation > > interface or by reading a file. > > 2. 
  Since there's no array in AFL to iterate these over, write them > to 
  > > a file (if read using automation interface).> > 3. For 
  each ticker in the file, calculate current gain (as above or > > 
  thru a better formula), and write it to same or another file.> > 4. 
  Sort tickers in the new file by the gain value.> > 5. Use a 
  percentile style calculation to generate the rank, either > by > 
  > creating percent buckets using value or by using number of > 
  securities.> > > > Is there a better way to do this than 
  above?> > > > Thanks in advance.> > > 
  > Jitu> > > > Yahoo! Groups Sponsor> 
  ADVERTISEMENT> > > > > Send BUG REPORTS to 
  bugs@xxxx> Send SUGGESTIONS to suggest@xxxx> 
  -----------------------------------------> Post AmiQuote-related 
  messages ONLY to: amiquote@xxxxxxxxxxxxxxx > (Web page: <A 
  href="">http://groups.yahoo.com/group/amiquote/messages/)> 
  --------------------------------------------> Check group FAQ 
  at:> <A 
  href="">http://groups.yahoo.com/group/amibroker/files/groupfaq.html 
  > > Your use of Yahoo! Groups is subject to the Yahoo! Terms of 
  Service.Send 
  BUG REPORTS to bugs@xxxxxxxxxxxxxSend SUGGESTIONS to 
  suggest@xxxxxxxxxxxxx-----------------------------------------Post 
  AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx (Web page: <A 
  href="">http://groups.yahoo.com/group/amiquote/messages/)--------------------------------------------Check 
  group FAQ at: <A 
  href="">http://groups.yahoo.com/group/amibroker/files/groupfaq.html 
  Your use of Yahoo! Groups is subject to the <A 
  href="">Yahoo! Terms of Service. 







Yahoo! Groups Sponsor












Send BUG REPORTS to bugs@xxxxxxxxxxxxx
Send SUGGESTIONS to suggest@xxxxxxxxxxxxx
-----------------------------------------
Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx 
(Web page: http://groups.yahoo.com/group/amiquote/messages/)
--------------------------------------------
Check group FAQ at: http://groups.yahoo.com/group/amibroker/files/groupfaq.html



Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.