beero1000
7 posts
msg #49166
- Ignore beero1000 |
1/12/2007 12:55:48 AM
is anybody here working on their own indicators?
I've started coming up with something based on expected value, programming it out in stockfetcher and then testing it. its still not consistant, but I think i'm making progress.
I was wondering if anyone else was coming up with their own and if they wanted to share. maybe a peer review kind of thing.
|
TheRumpledOne
6,411 posts
msg #49175
- Ignore TheRumpledOne |
1/12/2007 9:14:34 AM
If you look through all of the posts then you'll find lots of indicators that people have created.
|
corsino
259 posts
msg #49218
- Ignore corsino |
1/13/2007 12:58:58 AM
I'm not aware of any new indicators.Most indicators, such as RSI, OBV, MACD,Williams %R, Stochastics, etc...were created quite some time ago. There are a lot of filters using either single indicators or combinations of them.Are you writing about a new indicator or a new filter?
|
beero1000
7 posts
msg #49239
- Ignore beero1000
modified |
1/14/2007 1:16:51 AM
I'm working on a new indicator.
I don't know of any really recent indicators. the most recent ones i know of are welles wilder's in the 70's and 80's.
why haven't there been any new ones? I'm not convinced that all the good ones have been invented yet.
I've been taking techniques from my mathematics classes and trying to see if I can come up with something that shows a pattern that will be tradeable.
for example, one of my early tries is a simple probability counter. I took different periods, 5,10,25,50 days and calculated the historical probability of an up day.
that was very simplistic and acted as a really basic oscillator. the longer the period the more reliable it was, if you hit 0 on the 5 day one, it meant it will probably have an up day in the near future, but not necessarily. If you hit 0 on the 50 day it is a stronger signal. but you could also do crosses, if the 50 day prob. crossed above 0.5 then you have a statistically larger chance of having an up day as opposed to a down day. this still wasn't really good, so I took a very long sampling period, 250 days, and looked for stocks who had a much higher than 50% chance of going up on any given day. that started to look better, but I noticed somehting, some of the stocks that had high % up, were still in downtrends. so I started weighting the up and down days by the range of close - open. that worked better. then I incorporated volume into the weighting.
thats pretty much where I am now. I have an indicator thats kinda cool. in some ways it leads the price and acts as an oscillator and in some ways it lags and acts kind of like a moving average. I'm trying to figure out the rules on entry and exit. and I'm trying to fine tune it so it works in all types of markets.
it would help if my stockfetcher backtesting worked, but what can I do?
anyone else doing something like that? : )
and TRO, ive looked through the posts and still haven't found anything. do you have any links.
|
corsino
259 posts
msg #49240
- Ignore corsino |
1/14/2007 1:44:39 AM
Well, good luck with your new probability indicator. All I know is that the longer a stock has been declining, the higher the probability that it'll go up soon !
There are quite a few good programmers here at SF, but I don't know if they are good mathematicians too.
|
rharmelink
81 posts
msg #49249
- Ignore rharmelink |
1/14/2007 6:01:26 PM
What makes you think there are no new indicators? There are new ones being discussed in "Technical Analysis of Stocks and Commodities" and similar publications all the time.
But, they are all just variations on a theme, since you only have a limited amount of data items to permutate. Now, if you expand to fundamental items, such as using something like the rank %'s of SI Pro or Portfolio123, you have a lot more things to deal with. However, that many data points creates a big problem for backtesting.
|
beero1000
7 posts
msg #49251
- Ignore beero1000 |
1/14/2007 6:26:44 PM
thanks for the info, didn't know about that
|
estetson
9 posts
msg #49347
- Ignore estetson |
1/17/2007 10:20:37 PM
beero1000,
What you're working on sounds very similar to a filter I recently created and posted, in a thread called "frequency of consecutive up or down days". This filter produces an oscillator for the probability that if the stock went up today, it will go up tomorrow, or if down today then down tomorrow.
I don't know how similar this is to what you're trying to do, but you might want to check out the filter I wrote and see if anything there might be useful.
|