Re[2]: CASH index is poor model for Oddball & Variants

[Jim Johnson <jejohn@xxxxxxxxxxx>]

Hello DH,

I'm in the throws of running the OddBall variaitons posted to the O
List (Brown, Folster, C Johnson, ,Dubell etc.) for the period 1994
thru 2001 in two year windows.  The thing that puzzles and somewhat
concerns me is that the system became more profitable as time goes
forward. With 2000-2001 being vastly more profitable than the other
periods.  I'll post some details soon.  But for now, what can cause
this?  Increased range and volatility?  Different composition of the
NYSE issues (McMillan's study shows more interest rate sensitive
closed end funds), quirks resulting from the cap weighted nature of
the SP500???

Although OB is profitable thru the entire period, the early years
average trade is small (<$200).  Unless one can determine what causes
this change in average trade size over time, one can't really guard
against it lapsing back into tiny mode going forward.


Best regards,
 Jim Johnson                           mailto:jejohn@xxxxxxxxxxx

-- 
Sunday, May 19, 2002, 2:10:05 PM, you wrote:

>> TNP on data from 3-14-02 till 5-17-02 using SPM2 data with Original OddBall 
>> at 7,3,1 = $11,925 TNP with $18,700 MIDD 
>> 
>> TNP on data from 3-14-02 till 5-17-02 using Cash index data with Original 
>> OddBall at 7,3,1 = $14,960 with $17,427.50 MIDD 

D> This demonstrates that the original oddball is a robust system that can
D> be used on different data streams and still make money. It is likely to
D> perform well in the future because it is relatively insensitive to minor
D> market/data changes.

>> I have an interesting, initially promising OddBall variant that shows: 
>> 
>> TNP on data from 3-14-02 till 5-17-02 using Cash index data = $27,787.00 
>> 
>> TNP on data from 3-14-02 till 5-17-02 using SPM2 data with = -450.00 
>> (negative) 

D> This is a classic example of overoptimization. It performs better than
D> the original on one data stream but it blows up completely on another.
D> This system is unlikely to perform well going forward. It has been fine
D> tuned to one set of data that will likely never be repeated.

D> All the evidence is there in plain sight. Your conclusion that "CASH
D> index is poor model for Oddball & Variants" is the wrong one. The
D> correct conclusion is that overoptimized systems can't be profitably
D> traded with real money.