[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Recreating the back adjusted vistorical number



PureBytes Links

Trading Reference Links

A futures roll is no different than a forex fwd roll (I used to do them
constantly) .. the fwd price is = spot adjusted for carry.  The back
adjusted is the sum of those carry adjustments.  Very simple.  So why is the
sum of the carry adjustments so much greater than I get when allowing for
t-bills-dividends .. this shouldn't be complicated .. I am making a stupid
mistake somewhere

-----Original Message-----
From: DH [mailto:catapult@xxxxxxxxxxxxxxxxxx] 
Sent: Tuesday, October 27, 2009 4:10 PM
To: Omega List
Subject: Re: Recreating the back adjusted vistorical number

It looks like you are using the wrong kind of continuous contract for 
the calcs you are doing. There are many ways of adjusting -- add so many 
points, multiply by so many percent, etc. I haven't really thought it 
through but percentage calcs would probably be better for buy and hold 
compounding. Any of them are just an approximation and, to really do it 
right, you'd take each contract's unadjusted price and crunch the 
numbers from roll date to roll date.

CSI offers these:

Transformed Computed Data
Unfair Advantage comes with software that will transform raw futures 
contract data for a given commodity into a single continuous series. 
This forces any futures series comprised of many delivery months into a 
single series that can be analyzed over an extended period. This type of 
data is suitable for individual study or for simultaneous analysis of 
multiple continuous series.

The available options include:
# A time-weighted Perpetual ContractR data series, which reduces a given 
futures market into a single continuous series using a constant 
future-period-forward perspective measured in days and months.

# A back-adjusted series in which individual contracts are concatenated 
over time and adjusted by a delta difference (to provide smooth 
continuity to the data) as each contract rolls from the current month to 
a more distant month. Delivery months visited may be dependant upon 
either a given calendar date relative to the start or end of the month 
or the magnitude of volume, open interest, either volume or open 
interest or both volume and open interest, etc. Precise roll timing may 
be controlled according to the availability of delayed volume and/or 
open interest reports relative to the current trading day.

# A proportionally adjusted series, which is quite similar to the above 
back-adjusted series, except that the delta adjustment is done in 
percentage terms. This minor difference helps to avoid situations where 
a back-adjusted series might move into negative territory. A 
proportionately adjusted series cannot move into negative territory, but 
it can approach zero.

# A Gann series, in which historical data is transformed into a series 
comprised of successive historical segments of the same delivery month 
over successive years. Using this algorithm, the output continuous 
series becomes a compilation of all July deliveries, for example.

# A Nearest Future series, which is an artificial contract representing 
a concatenation of successive contracts over time, reflecting the price, 
volume, and open interest of the Nth nearest future. In this option, 
there is no attempt to account for step-size jumps or drops in price as 
contracts change from one to another.

-- 
   Dennis

Paratrade Systems LLC wrote:
> OK.. the compounded return for the index without dividends is about 7.5% -
> that is WAY above the futures return (starts at 675 and ends at 1050
(about
> 1.75%) + the 2.3% compounded carry return for a grand measly total) of
4.05%
> 
> Why are these numbers so far apart (and I haven't even thrown in
dividends)?