|
|
| Topie: Wednesday, May 25, 2011 |
|
|
|
|
| membar |
wanlorn, if you're listening, one thing I learned from you is to sync my time with a time server every morning. |
|
|
| allen314159 |
&#%6# found an error in my code!
Today is not Jan 6 2009!! |
|
|
| BullDog |
Do you any of you watch CNBC?
Mark Haines, passed away. he was one of their better anchors? Did enjoy his interviews and commentary. Will miss him. |
|
|
| membar |
Quote:
allen314159 wrote:
&#%6# found an error in my code!
Today is not Jan 6 2009!!
|
Sorry. If I had a nickel for every time that sort of stuff has happened to me in the past couple of years, I wouldn't be doing this anymore. (That plus Internet connections going down....)
Not stopped out of anything so far. Everything is in the green, including AGQ. This is where ISA_Coder would take profits and I just watch and weep as everything tanks towards the end of the day.... |
|
|
| BullDog |
McClellan oscillator is predicting that the market's going to rocket higher any day now. I do not follow the McClellan oscillator, but hope this is true.
Check this out:
There Is A Huge Move Coming In The Markets
http://etfdailynews.com/2011/05/25/there-is-a-huge-move-coming-in-the-markets/ |
|
|
| allen314159 |
Quote:
membar wrote:
Sorry. If I had a nickel for every time that sort of stuff has happened to me in the past couple of years, I wouldn't be doing this anymore. (That plus Internet connections going down....)
|
I discovered the error when I found that my overnight FAS/FAZ prediction happened to be the same everyday (and betting on FAS)... I missed out on some gains since my fixed predictor would've been FAZ yesterday. Oh well. At least it error'd on the "safer" side.
|
|
|
| BullDog |
woh, look at that jump! |
|
|
| membar |
Quote:
BullDog wrote:
woh, look at that jump!
|
I'm not complaining :) Good for you for sticking to your guns. |
|
|
| membar |
Quote:
BullDog wrote:
woh, look at that jump!
|
The ending was a bit gross, but I'll take it. |
|
|
| allen314159 |
Quote:
membar wrote:
The ending was a bit gross, but I'll take it.
|
Yikes!
My code sold FAS at 26.41, and is in FAZ at 45.04 for the overnight gap!!
This is scary - first time I've held FAZ overnight in a long time! Hope this isn't a programming error. |
|
|
| membar |
Quote:
allen314159 wrote:
Yikes!
My code sold FAS at 26.41, and is in FAZ at 45.04 for the overnight gap!!
This is scary - first time I've held FAZ overnight in a long time! Hope this isn't a programming error.
|
So you do a least-squares fit? Over what time period? |
|
|
| allen314159 |
Well, it's a little more complicated that least squares..
basically, I look at 11-days prior, considering OHLC data. Then I use each OHLC data point to come up with a estimations of tomorrows gap (everything is ATR5 volatility-adjusted). Then I have a weight that corresponds to each point, and I do a least squares fit based on the weighting. For example, the open-to-close of the current day is a much better predictor than the high of 7 days ago. So when doing the least squares fit, I weight each predictor according to how well it correlated. Near term weights tend to be higher than far term, but there are 5-day, 10-day cycles the market exhibits which creates peaks at these points.
Then I look at historical data for what matched the current data and see what the average next day's gap was for the historically sqrt(N/2) nearest points (~17, based on least squares ordering of the weights above).
I don't know if any of this makes sense... but I use the same idea for determining whether to use FKOS or FAZ-kos based on the morning's gap and historical performance of the nearest points for FKOS/FAZ-kos and see what had the higher average.
If I can be right 60% of the time, that's a huge advantage. Remember the curve I showed that described the returns gained on just getting either FAS or FAZ right from close to close back in the polling days? If you got anywhere near 60% you were basically hitting grand slams in terms of performance. |
|
|
| allen314159 |
Doh! Looks like FAZ was probably a bad choice. Oh well. |
|
|