klynn55
747 posts
msg #97950
- Ignore klynn55 |
12/10/2010 10:21:10 PM
mahkoh: the filters are great but the short filter seems at least half the time to pick stocks that are near, very near, highs , like a 52 wk high, often that mere fact seems to propel a stock upward by momentum traders, so perhaps look at the charts and ignore those at highs.
|
mahkoh
1,065 posts
msg #97966
- Ignore mahkoh |
12/11/2010 6:27:26 PM
The filter just looks for the pattern with the highest relative volume, it is up to you to put it in a context. It is somewhat funny that while many find a bullish pattern at a new low appealing, they become awkward when the inverse happens, a bearish pattern at a new high. Your rationale is very correct though, you have to deal with buyers and they are more dedicated than sellers. However, if you offset the days you'll see that the next day's low is very often below the close of the bear candle. This means that there is money to be made, albeit on a shorter timeframe. If I find a candidate I open a position near the close when I can be pretty sure that the pattern will remain intact. The next day I cover half when the price reaches the daily pivot point or S1, place a tight stop on the remainder and see how it goes. Of course this very much depends on the broader market action.
In general I believe that if the market has an up day the profit from your bullish positions will outweigh the loss from your bearish takes and vice versa.
|
jimvin
170 posts
msg #97968
- Ignore jimvin |
12/11/2010 8:49:11 PM
In my experience (over the last 10 years) markets have different "personalities." I was making 10% a week on average in the run-up to the dot-Net collapse using a weekly Top-10 RSI model; 3 years ago I was making 15% a week using a momentum model based on a combination of elements in an MSN Money filter. 8 months ago I was making about 10% a week on a variation of the momentum filter based on tachnicals at BarChart.com and StockCharts.com. Each of those models eventually failed as the market shifted (i.e. the big boys (and girls) who move the market changed their buying trends).
I am now in a holding pattern testing various filters here to see which best fits my preferred trading style: I wnat to buy 3 to 5 stocks on Monday, sell them during the week at a profit of from 2% to 10%, with a stop-loss of 5% if the stock price drops.
Over the last 6 months I've tested over 100 filters and found that some tennants of classic stock theory do not apply in this market; for example, an RSI of over 70 is usually a danger signal, with an RSI over 80 a defintie no-buy...but many stocks generated by the filters with RSIs over 70 and 80 continue to go up from week to week.
I have no filters to offer as "sure winners" yet, but my general suggestions for an investing come-back for you would be:
1. Determine what your investment goal is (i.e. what rate of return do you want/need and over what period of time, week/month/longer?)
2. Select a group of - say, a dozen filters - based on a variety of theories (Darvas, Bollinger Bands, MACD, RSI...).
3. Try sorting the results they return by Volume and again by % Change, and track the results form each approach to see which is the best predictor with the input to the model(s) you are tracking.
4. Try an analysis of the top 10 using some standard criteria (P&F charts - available at StockCharts.com, candlesticks - available here, etc.)
5. DO NOT get too deeply into technicals; for example, I did a study of the technicals at Barchart.com for several months and found little short-term correlation between their positives and stock prioce increase.
6. Know that penny stocks (i.e. stocks selling for less than $5) have different rules than non-penney stocks; they are - in my experience - unaffected by RSI, for example.
7. Any ONE approach, such as candlestick charts, is not enough; I tracked Bullish Engulfing, for example, for several weeks and found no strong correlation between that "pure" approach and the short-term movement of a stock's price; a blended filter that employs some combination of methods generally works better (in my research so far).
8. Know that back-testing options at StockFetcher may not reflect your particular stock-trading style; while other features of this site are excellent, I prefer to test on a daily basis (weekly cumulative) using end-of-day data each day (open/high/low/close) to develop a weekly profile of how the filter is working.
9. ALWAYS set a stop-loss at between 5% and 8%...10% max' (which you chose depends on your tolerance for risk and the beta of the stock price fluctuations based on your study of the model(s) returns. (In this market, as noted I generally prefer 5%).
In any case, this has gone on far too long...if any of this is of any value, thanks for reading; if not, at least my opinion was free.
|