acfncp3
59 posts
msg #64176
- Ignore acfncp3 |
6/22/2008 4:36:08 PM
I have been running some of the forums filters against the backtest feature. I always seem to get a %ROI return of -110 percent or so. Am I missing something . My best backtest on some very simple filters, ( cross above rsi 50, DMI cross yields a ROI of about 18 percent or so. Am I missing something on the definition of ROI. Does someone have a filter that they have backtested that has in their opinion an outstanding %ROI. By the way long time SF user .... first time posting.
|
acfncp3
59 posts
msg #64180
- Ignore acfncp3 |
6/22/2008 6:02:36 PM
Show stocks where MACD fast line(12,26) is above 0
Show stocks where MACD slow line(12,26) is above 0
and Average Volume(90) is above 500000
and close is between 30 and 250
stocks are optionable
rsi above 50
stock touched MA(30)
Show stocks where Williams %R (14) has been increasing over the last 2 days
+DI(14) above -DI(14)
The above filter is just so simple compared to some of the filters I've seen in the forum. I'm almost embarrassed to show it. It does return a %ROI of 14.88 percent though. I like using the MA(30) to confirm a bullish entry and then adjust to a option spread if it goes below the MA(30) to keep loses to the minimum.
When I run some of the other filters on backtesting I'm getting some weird numbers like -110 -120 %ROI. Is this a shortcoming of the SF backtesting tool. Can some share filters that have double or even triple %ROI returns.
|
cunparis
71 posts
msg #64220
- Ignore cunparis |
6/23/2008 2:11:22 PM
Here's one I'm working on, I am trying to get the reward/risk higher but I can't figure out a way to do it:
and volume of 1 day ago is above 500000
and Close is above 4
and Open of 1 day ago is above Day Position(0.95,1)
and Close of 1 day ago is below Day Position(0.05,1)
and volume of 1 day ago gained more than 100 percent over the last 1 day
and close below open
and volume dropped more than 50 percent over the last 1 day
and high less than high of 1 day ago
and low more than low of 1 day ago
it's a 1 day daytrade on the short side:
There were 178 total stocks entered. Of those, 177 or 99.44% were complete and 1 or 0.56% were open.
Of the 177 completed trades, 85 trades or 48.02%resulted in a net gain.
Your average net change for completed trades was: 0.16%.
The average draw down of your approach was: -1.99%.
The average max profit of your approach was: 2.04%
The Reward/Risk ratio for this approach is: 1.15
Annualized Return on Investment (ROI): 40.05%, the ROI of ^SPX was: 3.41%.
I've got a lot of filters but they're all stuck around 50% and 1.2 to 1.5 ratio. Furthermore, most are $2 stocks with relatively thin volume that can't be daytraded in any serious amount. All are daytrades, which are hard to backtest with stockfetcher because it uses EOD data.
Also when I try to make it multi-day, with a stop loss, the results go below 50%. I get stopped out. I can't figure out decent exit conditions.
|
TrendSurfer
109 posts
msg #64223
- Ignore TrendSurfer
modified |
6/23/2008 2:57:14 PM
...
|
cunparis
71 posts
msg #64231
- Ignore cunparis |
6/23/2008 3:43:26 PM
I agree that one can't program discretion, which is why I'm trying to get set up with a mechanical system. I tried discretion for 6 months and have a losing record. With the backtesting I've been able to rule out a lot of things and have picked up on some new stuff as well. As to whether it'll work with real money, only time will tell. But I feel better knowing that the backtest is positive.
|
TrendSurfer
109 posts
msg #64242
- Ignore TrendSurfer
modified |
6/23/2008 4:49:32 PM
...
|
TrendSurfer
109 posts
msg #64245
- Ignore TrendSurfer
modified |
6/23/2008 6:08:48 PM
::: To further clearify cunparis #2 :::
The power of understanding technical analysis is not the ability to predict a stock’s every move.
The best technical traders I know, trade charts that make sense. They don’t try to make sense out of every chart.
Back-testing applies your test whether the charts make sense or not and spits out a result, it may not be the right result for a system that in reality works.
If the chart does not have any obvious areas of support and/or resistance then don’t try to find something that doesn’t exist!
Back-testing applies your test whether something exists or not and can miss the mark completely.
One of the biggest mistakes new technical traders make is that they assume that every chart has some pattern that can be traded.
The best approach to reading charts is to focus on the charts that have OBVIOUS PATTERNS rather than trying to interpret patterns that are not clear or have a back-test find them.
The best way to trade using charts is to become your OWN EXPERT in analyzing a select few chart patterns ferreted from refined scans and then trade stocks that have your patterns.
In my view, SF power is in allowing a novice and a pro to write powerful code in a simplistic way and refine scans to find stocks with tradable patterns. Pretty simple right!
I've worn this out. Just wanted to be clear where I'm coming from.
I truley hope this helps...
|
chetron
2,817 posts
msg #64248
- Ignore chetron |
6/23/2008 6:56:00 PM
CLICKABLE.....
THANX FOR SHARING.
|
chetron
2,817 posts
msg #64249
- Ignore chetron |
6/23/2008 6:57:46 PM
CLICKABLE...
|
acfncp3
59 posts
msg #64278
- Ignore acfncp3 |
6/24/2008 10:58:05 AM
again I'm new ... how do you make the post clickable... that was cool...
Over time I have taken a lot of ideas and built SF screens out of them ... BB squeeze, DMI crossovers, rising three candles.... so I've accumulated quite a few filters.... I just discovered the back test so I have been using it to weed out all these filters I have .... to at least show a positive %roi... so I'm not trying to create a black box....
So is the BT any good to evaluate some of the recent filters that some of the experts here have posted that look so good and which I could never write ...
|