dmewbourne
15 posts
msg #97599
- Ignore dmewbourne |
11/24/2010 11:58:08 AM
Using Double Stochastics, I'm really close to a filter with good returns. Problem is, there are a lot of hits that don't pan out. Any ideas on how to filter out more underperformers/losers?
|
jimvin
170 posts
msg #97600
- Ignore jimvin |
11/24/2010 1:00:27 PM
One of the tricks I've learned is to sort by % Change; those stocks which have gone up the most generally continue to go up the most. The default value for returns is Volume which works less well.
|
dmewbourne
15 posts
msg #97612
- Ignore dmewbourne |
11/25/2010 7:49:33 AM
But that seems to also produce some of the biggest losses, at least with this screen.
|
jimvin
170 posts
msg #97613
- Ignore jimvin |
11/25/2010 2:48:55 PM
Alright..that was my quick-and-easy answer and it didn't panout...so here's a longer one, which may or may not be as useful. I'm sure there are ways to achieve some of this through the use of filters. but consider any or all of these any way you chose to implement them: (1) Look at the candlestick charts; a long white candle on the last day of trading is often a sign of accumulation and may indicate a sustsained buying pattern; if you're good at candlesticks, look for the more detailed indicators such as Bullish Engulfing, Bullish Reversal, Bearish Reversal, etc.; (2) Consider the P&F Chart (I use the ones at Stockchart.com); (3) Consider other technical indictors such as the RSI and MACD. DISCLAIMER: I relied heavily in the P&F charts and RSI and MACD in my trading prior to August of this year (the classic RSI above 70 is overheating and over 80 is definitely overbought) and found that these no longer work well with the majority of filters I now use. In this market, for example (and every market has its own personality) stocks with an impossibly high RSI by classical standards (90 and above) have gone much higher in value. However the use of additional technical indicators may work well with this model... In any case, good lcuk with your trading; I hope your Thanksgiving is a happy one.
|
dmewbourne
15 posts
msg #97622
- Ignore dmewbourne |
11/26/2010 12:55:16 PM
Thanks jimvin. Happy Thanksgiving to you as well.
|
jimvin
170 posts
msg #97632
- Ignore jimvin |
11/26/2010 11:50:27 PM
I've run well over a hundred of the filters here on StockFetcher over the last year...and looking at those that use Stochastics my testing shows that one filter in particular that uses a Bollinger Band overlay may show over time a positive results and a strong correlation with traditional candlestick analysis. I don't know who to credit for this, but here it is:
draw cema(Double Stochastic(10,3),3) 1 day ago < 25
double stochastic(10,3) crossed above cema(double stochastic(10,3),3)
close > open
close < 1
avgvol(22) > 9785
price touched upper bollinger band(20,2)
For this week (a very volatile week) to be sure the picks using a candlestick analysis overlay were:
GNP (76.47% / 17.65%) and VLCO (29.19% / -8.11%)
The first number is the "sweet-spot" sale; that is, the return you would've gotten if you dold the stock at this week's high price; the second number is the Friday close; the return you would've gotten if you fell asleep after 9:30 Monday and didn't wake up until late Friday afternoon (missing out on turkey and somne good returns).
I haven't done a lot with this filter (I'm not big on Stochastics) but you might want to play with a Bollingert overlay on your filter
I'd also suggest testing RSI and MACD crtieria. And note that m best results for any filter have been to buy on Monday at market open with a 5% trailing stop-loss (other recommend 8% to 10%).
I'm sure there are other contributors here well versed in filter-construction that can run circles around me (and will when they get back from Thanksgiving) but this is my best shot.
Good luck with all of your trades.
Now for a turkey sandwhich...
|
four
5,087 posts
msg #97633
- Ignore four
modified |
11/27/2010 12:54:33 AM
|
jimvin
170 posts
msg #97643
- Ignore jimvin |
11/27/2010 7:57:03 PM
Wow! Elegant filter, but ugly results. The results of your posting of Slow Stochastic and Bollinger Bands filter on 11/6/2010 look a lot better...of course that doesn't address the double stochastics but every market has its own personality and Slow Stochastics may fit this better. (Just thinking out loud...)
|
four
5,087 posts
msg #123608
- Ignore four |
4/19/2015 10:48:38 PM
pop
|