Sector Rotation

kkseal

Well-Known Member
#21
Oxy, just wondering if the calculation can be done on a day-to-day basis i.e. (Close today/Close yesterday) * 100 This would eliminate the need for choosing a base year Stocks can be included/replaced at any time (may be 3-6 mths after listing when the daily range is more in tune with other constituents)
 

kkseal

Well-Known Member
#22
The long term investor doesn't need to go by technicals He will focus more on the Fundamental outlook of a Stock & sector (like despite the current battering of infra & cap goods sector stocks they'll remain good long terms bets to him/her & hence a buying opportunity In fact Buffet advises to buy & hold stocks which would not be affected if the stk mkts were to close down tomorrow!)

It's in fact the short term trader who needs to look at short term sectoral strength/outperformance. The only category that can be excluded perhaps is the Nifty trader, but even such a trader could use a collective view of the various sectors as a short-term sentiment indicator.
At the same time, there's no sacrilege in a long term investor taking a graphical view of how a sector has performed compared to the mkt &/or other sectors over say the last 5 yrs.

And i also don't see why a swing or even an intraday trader shouldn't seek outperformance (based on one or more parameters like sectoral strength).
 

oxusmorouz

Well-Known Member
#24
Oxy, just wondering if the calculation can be done on a day-to-day basis i.e. (Close today/Close yesterday) * 100 This would eliminate the need for choosing a base year Stocks can be included/replaced at any time (may be 3-6 mths after listing when the daily range is more in tune with other constituents)
It can be done Kalyan. Substituting cum(ROC(C,1,%) instead of 100*(C/ValueWhen(1, value on nth day,C) - 1) will do the trick. Tried this and got the same result for ranking, only the values varied. Another problem with cumulative ranking is, if a stock is delisted for a particular time, then the index will be deflated to that extent during that time. The base date will avoid the problem because "IsDefined" function adds 1 to the divisor only if the stock is traded on that day.
 

kkseal

Well-Known Member
#25
It can be done Kalyan. Substituting cum(ROC(C,1,%) instead of 100*(C/ValueWhen(1, value on nth day,C) - 1) will do the trick. Tried this and got the same result for ranking, only the values varied. Another problem with cumulative ranking is, if a stock is delisted for a particular time, then the index will be deflated to that extent during that time. The base date will avoid the problem because "IsDefined" function adds 1 to the divisor only if the stock is traded on that day.
Owing to some of the pitfalls with indices i have hit upon an idea which can provide the same view/info that indices would without actually needing to create indices (sectors will still be needed off course) I'll let it gestate for some time before actually trying it out. In the meanwhile will continue to follow your work (& there's every likelihood my half-baked idea will turn out to be crap :D)

Regards,
Kalyan.
 

kkseal

Well-Known Member
#26
...

Here are some reason it may not work for short time trader.

Suppose, Cement and Infra are the weakest sectors but highly oversold. There shall be confusing signals in any system...in that situations to go long...but later that may be treated as whipsaws.

I have seen here in traderji most of the members went long on these two sectors considering them bottoming out, and has burnt their fingers.

My idea is to catch the fresh breakouts/stocks changing their trends and finally something about to happened, but not the things happened.
One can find out if a sector is ob/os the same way one'd figure out if the mkt is ob/os (using a mkt proxy like Nifty or CNX500 &/or mkt data comprising all its constituent stocks); In fact once one has the base material - the sectoral index & it's constituents - one can slice & dice it any way one wants, to extract whatever info one seeks with the advantage of being able to view a sector as a whole rather than extrapolating what is conveyed individually by 1 or 2 stocks (i.e. you get to work with a larger sample without needing to repeat the rather tedious sampling process every time. This process is abstracted & encapsulated in the form of the Sectoral Index created once).

Finally, everything need not be useful to everyone but i think it's better to keep an open mind rather than being needlessly (& arbitrarily) dismissive.
 
#27
gr8 oxy
learning from u as usual:)
one request to u...
can u spare some time to light up the correlation of the sector heavywgt with the index itself...
looking fw for ur valuable input
regards
 
#28
One can find out if a sector is ob/os the same way one'd figure out if the mkt is ob/os (using a mkt proxy like Nifty or CNX500 &/or mkt data comprising all its constituent stocks); In fact once one has the base material - the sectoral index & it's constituents - one can slice & dice it any way one wants, to extract whatever info one seeks with the advantage of being able to view a sector as a whole rather than extrapolating what is conveyed individually by 1 or 2 stocks (i.e. you get to work with a larger sample without needing to repeat the rather tedious sampling process every time. This process is abstracted & encapsulated in the form of the Sectoral Index created once).

Finally, everything need not be useful to everyone but i think it's better to keep an open mind rather than being needlessly (& arbitrarily) dismissive.
sir
again a request to u..
a small flowchart would be helpfull....
for members like me
regards
 

kkseal

Well-Known Member
#29
sir
again a request to u..
a small flowchart would be helpfull....
for members like me
regards
Flowchart of what? If you mean the index creation process Oxy outlines it here (post #13) :-

http://www.traderji.com/metastock/21228-how-create-index-sector-chart-2.html#post173802

You can also go through Oxy's posts (#13 & #14) on pg2 of this thread.

If an Index is equal-weighted (what Oxy has done) there's no 'heavywt' as such (vis-a-vis the Index) since all constituents carry equal weight. However you may choose to ascribe different weights to different constituents in which case the ones assigned the highest weights will be the heavyweights for your index. The movements in the higher weighted stocks will have a greater influence on the movement of the index value.

And no 'Sir' please Makes me feel old :)

Regards,
Kalyan.
 

Similar threads