|
PureBytes Links
Trading Reference Links
|
>Of course, if I ever get around to testing what would happen if trading
from
>signals based on Perpetual Contracts, I'll be glad to have that facility.
>
>Owen Davies
I did just that.
mb says you can get better results taking signals from the perpetual
contract. This intriqued me so I tested it. I tested my straightforward long
term channel breakout system (diversifed basket of commodities) using
closing prices only taking signals from both a back adjusted series as well
as a perpetual series. In all cases, trades were executed on the front month
as you would in real life. Slippage, commisions, and rollover costs were all
taken into account. I use a sophisticated Excel spreadsheet to do my
portfolio testing. I can program anything I want in VB. My back testing can
be replicated exactly in real life.
I discovered that taking signals from the perpetual series was significantly
worse than just a back-adjusted front month series. I found the same sort of
thing when testing systems that only use the 2 most active contracts in a
year rather than all active contracts (to cut down on rollover costs).
Amazingly, in every case, the best results was just to always trade the
front month and pay the rollover costs. Even in markets like CL where you
are rolling every month. The best I could determine was that the front
months had more movement, ie bigger swings than back months at the same
time.
FWIW,
Scott Hoffman
Issaquah, WA
-----Original Message-----
From: Owen Davies <owen@xxxxxxxxxxxxx>
To: Mark Brown <markbrown@xxxxxxxxxxxxx>; OmegaList <omega-list@xxxxxxxxxx>;
clydelee <clydelee@xxxxxxx>
Date: Thursday, April 29, 1999 1:35 PM
Subject: Re: Wish List
>
>Mark Brown commented:
>
>>someday i will make a believer of all of you and show you that indeed you
>>can trade a futures contract (and with better results) using signals
>>generated from a perpetual data series (tm CSI). (etc.)
>
>It's not that I don't believe it. I know of one successful trader who was
>doing
>just that, back when Perpetuals were new. But it's not the way I trade,
and
>I want my back-testing to model real life as closely as possible
>
>Beyond that, although I have not tested it myself, I understand that
>back-testing
>on Perpetual Data tends to give results about 25 percent more favorable
than
>similar tests on unadjusted contracts. Bad news if so, assuming you were
>going to use Perpetuals, but not critical for me. When using Perpetual
>Data,
>there's no way to know how closely the results of any given test mirror
what
>would have been experienced in practice. That pretty much rules it out for
>me.
>
>Of course, if I ever get around to testing what would happen if trading
from
>signals based on Perpetual Contracts, I'll be glad to have that facility.
>
>Owen Davies
>
>
|