| April 22nd, 2011 07:19 AM |
|
|
| membar |
Kind of scary to be short the Euro during this long weekend.
Allen: I've been putzing around with mx + b and can't get b to want to be above 0 for FAS (or TNA for that matter). In fact, lately, the strategy wants to make m very small, which makes the strategy tend towards a simple overnight hold with sell at open (or slightly thereafter). |
| April 22nd, 2011 07:20 AM |
|
|
| membar |
Quote:
membar wrote:
Kind of scary to be short the Euro during this long weekend.
Allen: I've been putzing around with mx + b and can't get b to want to be above 0 for FAS (or TNA for that matter). In fact, lately, the strategy wants to make m very small, which makes the strategy tend towards a simple overnight hold with sell at open (or slightly thereafter).
|
P.S. The backtest, like the strategy, hasn't performed well the last couple of months but also hasn't hit the max. drawdown either. |
| April 22nd, 2011 08:07 AM |
|
|
| membar |
P.S. Gold and silver have been in an uptrend for the past decade, albeit with sharp pullbacks. Maybe there's a strategy with those? I'm not a big fan of the TT/BB methods, because I think they might be curve fits, albeit over long periods of time. (How well did your UPRO TT/BB strategy perform after your first published it?)
Also, Van Tharp says that there are six market types: (normal. volatile, quiet) * (bull, bear) and that it's easy to design a strategy that works well in one of them, but difficult to design a strategy that works well for all of them. It would be nice to have an arsenal of strategies by market type and more importantly, understand why each strategy works in each market type. |
| April 22nd, 2011 03:20 PM |
|
|
| allen314159 |
member, I need to soak up all your messages before I reply. |
| April 24th, 2011 11:40 AM |
|
|
| membar |
Quote:
allen314159 wrote:
member, I need to soak up all your messages before I reply.
|
Looking forward to that reply :) |
| April 25th, 2011 04:33 PM |
|
|
| allen314159 |
Quote:
membar wrote:
Looking forward to that reply :)
|
Hey membar,
I tried doing some analysis, but I think I had a bug in my code.
But YES, I have noticed that the optimal stop parameter has drifted if you just look over the last year. In fact the profile over the last few months is much different than looking over the full period since inception.
So I'm trying to figure out what is the optimal look back period.
It might be that we have to look at longer term ETFs (maybe UYG, or XLF) to figure this out. |
| April 27th, 2011 03:45 AM |
|
|
| allen314159 |
Quote:
membar wrote:
Kind of scary to be short the Euro during this long weekend.
Allen: I've been putzing around with mx + b and can't get b to want to be above 0 for FAS (or TNA for that matter). In fact, lately, the strategy wants to make m very small, which makes the strategy tend towards a simple overnight hold with sell at open (or slightly thereafter).
|
membar,
I screwed up my signs.
For the following form:
SS = Open + (m x ATR5 + b)*PC
SS = Sell Stop
PC = previous close
m and b are both < 0
What numbers for m and b did you get? |
| April 27th, 2011 03:51 AM |
|
|
| allen314159 |
Quote:
allen314159 wrote:
membar,
I screwed up my signs.
For the following form:
SS = Open + (m x ATR5 + b)*PC
SS = Sell Stop
PC = previous close
m and b are both < 0
What numbers for m and b did you get?
|
Also, I'm getting very tight stops when you just use the last year's data... both m, b are close to zero (i.e., sell very close to open).
I've been debating, but I think I'm going to go with the numbers that include all FAS data.
I think what's happened recently is due to the very significant departure of financials from the general market.
Anyway. Not sure if it's a mistake or not, but I still think using more data is more conservative and generalized than using a shorter timeframe... |
| April 27th, 2011 06:59 AM |
|
|
| membar |
Quote:
allen314159 wrote:
Anyway. Not sure if it's a mistake or not, but I still think using more data is more conservative and generalized than using a shorter timeframe...
|
Hey Allen: I don't necessarily agree with this. I think using more data just gives you a curve fit which isn't necessarily applicable to the current timeframe. Check out my Jimmy Smith Strategy analysis: I believe I give a formula for computing system quality which uses expectancy as well as standard deviation. If you can optimize for system quality rather than expectancy, I might be sold. |