View previous topic :: View next topic |
Author |
200 EMA problem |
chrome Yellow Belt
Joined: 28 Dec 2009 Posts: 645
|
|
Back to top |
|
|
|
|
SID2060 White Belt
Joined: 04 Nov 2009 Posts: 319
|
Post: #2 Posted: Sat Jan 01, 2011 5:15 pm Post subject: |
|
|
banknifty ema 200
Chartfeed 10996
Chart premium 10905
Screener (EOD) 10893.38
I have seen such difference in almost every scrip and index. Plz resolve the matter asap.
Regards,
Sid |
|
Back to top |
|
|
SwingTrader Site Admin
Joined: 11 Aug 2006 Posts: 2903 Location: Hyderabad, India
|
Post: #3 Posted: Sun Jan 02, 2011 10:31 am Post subject: |
|
|
To understand why there is a difference in the values of EMA you will first have to understand how EMA gets computed. EMAs use a weighting factor to give more importance to recent values (and lesser importance to old values). EMAs use ALL data one provides it (unlike SMAs that use only fixed number of values for its computation).
Now coming to the point....JCharts & Charts (Premium) (or chart feed etc) use different number of data points (number of days/weeks/months of data) for EMA computation. This number of data points depends on how much data you are loading the chart. In JCharts this means your "Historical Intraday Timeframe" setting. If you select 3 yrs here then 3 yrs of data is used for EMA computation. Whereas in Charts (Premium) or chart feed link we internally use around 8 yrs of data for EMA computation - this is fixed. So JCharts & Charts (Premium) / Chart feed EMA values will not match.
Check SMA values, they should match everywhere. EMAs are a totally different animal _________________ Srikanth Kurdukar
@SwingTrader |
|
Back to top |
|
|
SID2060 White Belt
Joined: 04 Nov 2009 Posts: 319
|
Post: #4 Posted: Sun Jan 02, 2011 11:09 am Post subject: |
|
|
Dear st,
If I take DLF one year data and then compute ema 14 and than i again compute ema 14 but from 6 month historical data of DLF. Does the ema14 will be different for full year and half yearly timeframe? |
|
Back to top |
|
|
SwingTrader Site Admin
Joined: 11 Aug 2006 Posts: 2903 Location: Hyderabad, India
|
Post: #5 Posted: Sun Jan 02, 2011 11:12 am Post subject: |
|
|
SID2060 wrote: | Dear st,
If I take DLF one year data and then compute ema 14 and than i again compute ema 14 but from 6 month historical data of DLF. Does the ema14 will be different for full year and half yearly timeframe? |
Yes, EMA value completely depends on how much data is used in its computation. So EMA(14) computed using 6 months & 1 year data will be different. This difference is due to the way EMA is computed...it is due to its formula. _________________ Srikanth Kurdukar
@SwingTrader |
|
Back to top |
|
|
SID2060 White Belt
Joined: 04 Nov 2009 Posts: 319
|
Post: #6 Posted: Sun Jan 02, 2011 11:42 am Post subject: |
|
|
Dear ST,
I will compute ema14 today and will see the difference. Also i think u should not compute ema from 8 year historical data in chart (premium) because if any adjustment like split or bonus is missed than it will have an adverse effect on the computed ema. I had reported such incidence in kfa where stock split happened in the past but was not adjusted and wrong emas were calculated. I just wana say that i think 2 or 3 years data is more than enough to compute ema 200 and all othe emas.
I have made screener (eod) for american stock mkt comprising of components of dowjones, s&p 500, nasdaq 100, s&p 100 and other most liquid100 etf's like spy, qqqq, dia,gld, slv,sds etc and i use 1 year timeframe data to compute ema 200 and all other emas. In one year the american mkt have 253 traded days this time.
Regards,
Sid |
|
Back to top |
|
|
SwingTrader Site Admin
Joined: 11 Aug 2006 Posts: 2903 Location: Hyderabad, India
|
Post: #7 Posted: Sun Jan 02, 2011 9:29 pm Post subject: |
|
|
SID2060 wrote: | Dear ST,
I will compute ema14 today and will see the difference. Also i think u should not compute ema from 8 year historical data in chart (premium) because if any adjustment like split or bonus is missed than it will have an adverse effect on the computed ema. I had reported such incidence in kfa where stock split happened in the past but was not adjusted and wrong emas were calculated. I just wana say that i think 2 or 3 years data is more than enough to compute ema 200 and all othe emas.
I have made screener (eod) for american stock mkt comprising of components of dowjones, s&p 500, nasdaq 100, s&p 100 and other most liquid100 etf's like spy, qqqq, dia,gld, slv,sds etc and i use 1 year timeframe data to compute ema 200 and all other emas. In one year the american mkt have 253 traded days this time.
Regards,
Sid |
We have arrived at 8 years timeframe based on detailed research. Shorter EMAs will work fine with 1 year timeframe but longer ones do not. The problem about splits/bonuses is the one we need to address, reducing EMA computation timeframe is not the solution.
About your comment about you building screener for US market - I do not get the point, not sure what you meant. _________________ Srikanth Kurdukar
@SwingTrader |
|
Back to top |
|
|
aromal White Belt
Joined: 22 Mar 2010 Posts: 302
|
Post: #8 Posted: Mon Jan 03, 2011 7:35 am Post subject: Re: 200 EMA problem |
|
|
|
|
Back to top |
|
|
|