03bsbme
20 posts
msg #102397
- Ignore 03bsbme |
8/22/2011 12:40:58 AM
Kevin, in the most recent filter you posted on this page are you evaluating these on a weekly, monthly, or quarterly basis?
|
Kevin_in_GA
4,599 posts
msg #102398
- Ignore Kevin_in_GA |
8/22/2011 7:48:02 AM
Weekly, with any reallocation decisions made on Friday. If the leader has changed during the week I'll usually make the reallocation by noon.
|
03bsbme
20 posts
msg #102410
- Ignore 03bsbme |
8/22/2011 10:24:17 PM
So you would typically run the filter sometime Friday morning after the open and make the decision then? I'm probably incorrect here, but I thought that the weekly data from SF was only valid after the Friday close.
Any thoughts on stops? Perhaps an ATR trailing stop. This may be useful to prevent drawdown, but may also hurt returns if the ETF rebounds. My thinking is that the ETF may be the highest ranked, but still dropping albeit not as quickly as the others. The problem (or possibly good thing) if your stop gets hit is that you may not have any holdings for a while if you do this. And when do you jump back in? Wait for another ETF to pass it or buy it again if you believe the selloff is over? Anyone have ideas on this?
Of course the system as is seems to work great. I'm just curious about ways to improve it if possible.
|
glgene
618 posts
msg #102430
- Ignore glgene |
8/23/2011 8:46:26 PM
Kevin,
You are simply amazing! You win the Gold Star award, wrapped in gold!
Keep up the great SF coding, and being so kind to share it.
Gene in OH
|
Kevin_in_GA
4,599 posts
msg #102437
- Ignore Kevin_in_GA |
8/24/2011 8:35:30 AM
So you would typically run the filter sometime Friday morning after the open and make the decision then? I'm probably incorrect here, but I thought that the weekly data from SF was only valid after the Friday close.
True, but what drives the decision is the ranking, not necessarily the close price. In most cases the ranking leadership changes mid-week, and by Friday it is pretty much set. I deviate occasionally (recently EFA passed BND to assume the top spot, but I stayed in bonds because of the obvious issues in Europe - good call since EFA dropped out of the top spot and BND was once again the leader by the end of the next week).
I also look at the volatility and Sharpe ratio. If the difference in 14 week performance between say, SPY and BND is less than a percent, I would likely stay in BND if I were there already even if the signal said to go into equities. Just being honest here - the system is mechanical, but I am not always going to agree with the call.
Any thoughts on stops? Perhaps an ATR trailing stop. This may be useful to prevent drawdown, but may also hurt returns if the ETF rebounds. My thinking is that the ETF may be the highest ranked, but still dropping albeit not as quickly as the others. The problem (or possibly good thing) if your stop gets hit is that you may not have any holdings for a while if you do this. And when do you jump back in? Wait for another ETF to pass it or buy it again if you believe the selloff is over? Anyone have ideas on this?
The system does not use stops. I have played with this, but it really did not improve performance. Also, stop thinking about these as ETFs - the ETFs are simply the proxies for the underlying asset class. It is the asset class into which you are investing as much as it is the specific ETF.
Kevin
|
mahkoh
1,065 posts
msg #102448
- Ignore mahkoh |
8/24/2011 5:58:32 PM
Kevin, if I still want to use the 8 etf's you were using earlier in the thread would this be correct?
|
Kevin_in_GA
4,599 posts
msg #102450
- Ignore Kevin_in_GA |
8/24/2011 7:28:54 PM
Close - the ranking code would now look like this:
SET{RANK1EE, RANK1A + RANK1B}
SET{RANK1FF, RANK1C + RANK1D}
SET{RANK1GG, RANK1E + RANK1F}
SET{RANK1HH, RANK1G + RANK1H}
set{rank1ii, rank1ee + rank1ff}
set{rank1jj, rank1gg + rank1hh}
set{rank1kk, rank1ii + rank1jj}
SET{RANK, 8 - RANK1kk}
That way you are summing all 8 ETF relative ranks.
|
jackmack
334 posts
msg #102469
- Ignore jackmack |
8/26/2011 8:01:53 PM
Kevin
It would appear as thought it is GLD on Monday with TLT in a close second.
Correct?
|
Kevin_in_GA
4,599 posts
msg #102471
- Ignore Kevin_in_GA |
8/26/2011 9:27:03 PM
Yes, both clearly "risk off" plays. However, jumping in without a re-allocation signal is not how the system's performance was determined. I think either of these are good trades right now, but both have been riding an up trend (GLD since 4/15, now up about 22%).
|
mahkoh
1,065 posts
msg #102473
- Ignore mahkoh |
8/27/2011 1:48:31 PM
Agree with Kevin, get into GLD now you'll risk picking the top. Better wait until the system generates a new number one and then step in.
|