|
PureBytes Links
Trading Reference Links
|
I think Craig is correct here.
According to Kramers (TASAC 0494) you should only trade this when the mkt is
in a trending phase which he defines as a mkt with an ADX value between 26
and 40 or when > 41( hypertrend). He doesn't exactly specify his time-period
for the ADX but I guess it must be around 12 days. He just says: n-days. ADX
values between 16 and 25 are defined as trading range markets which should
be traded with the faster oscillators (MAM,moving average momentum and
Williams'%R) while prices "oscillate" between support and resistance.
Kramers states that this approach is good for S&P-500, D-Mark and T-bonds.
He uses intraday charts of 2 to 60 minutes depending on market conditions.
I had Malcolm Scott's formulas quickly run over daily data of the FTSE100,
S&P500 and AEX.
Though I agree with Harvey's conclusions on optimization I optimized over
wide ranges between 4 and 30 by 2 steps.
Harvey, I believe only in optimization when it shows that it works on
certain ranges, but not when say, in a optimization range between 10 and 20,
step 1 you only get results at 16 while 15 and 17 give bad results.
Primary conclusion: It works (S&P short signal on 160898 with opt1:4 and
opt2:8) but to me it seems like an illusion to put it into practice this
way. Too few trades over the last 5 years and too many and too big open
drawdowns.
I also noted(S&P500) that when in small ranges between say 4 and 10 for both
opt's, you are whipsawed in trading ranges like Feb-Aug 1996.
So we probably must add some trading/trending filter here. For example: Only
when ADX(12 or 14?)is > than 25? and rising?
Back to the drawing boards !!
Rgds
Frans
At 01:57 30-08-98 -0500, you wrote:
>James,
>
>There's no such animal as a 1.5SMA for -opt1. You'll need to "bound"
>both opt1 and opt2 test values to only positive integers in this formula
>construction. The "-opt*" will create an effective negative lookback.
>A Ref(*,+n) is how they play the game at your corner palm reading
>establishment. <G>
>
>Craig
>
>
>James Wolf wrote:
>>
>> Hello all.
>> I too am playing with the following formulas posted here lately.
>> The optimized values I have come up with so far are 1.5 for Opt1 and -8 for
>> Opt2.
>> Running this on the S&P500 I noticed that it should have given a 'GO-Short'
>> signal well before the recent decline. Somewhere around Aug 20th I would
>> think.
>> Looking at the formulas, the -8 for -Opt2 cancels out and turns into just 8.
>> Now given that, will there be an 8 day lag in the indicator?
>> Am I missing something obvious? Why the lack of a Short signal before the
>> decline?
>>
>> Enter Long:
>> (Mov(C,Opt1,S)-Ref(Mov(C,Opt1,S),-Opt2))>
>> Ref(Mov(C,Opt1,S)-Ref(Mov(C,Opt1,S),-Opt2),-1) AND
>> (Mov(C,Opt1,S)-Ref(Mov(C,Opt1,S),-Opt2))<0
>>
>> Enter Short:
>> (Mov(C,Opt1,S)-Ref(Mov(C,Opt1,S),-Opt2))<
>> Ref(Mov(C,Opt1,S)-Ref(Mov(C,Opt1,S),-Opt2),-1) AND
>> (Mov(C,Opt1,S)-Ref(Mov(C,Opt1,S),-Opt2))>0
>>
>> __________________________
>> James Wolf, KR9U
>> jbwolf@xxxxxxxx (hm)
>> jbwolf@xxxxxxxxxxxxxxx (wk)
>> Fax: 219-637-7074
>
>
|