|
PureBytes Links
Trading Reference Links
|
Yeah, Rich, I remember it well because Bob was responding to my post
regarding real time trading results verses long system backtests. I still
agree / disagree on the relevance of backtests to real time system
performance. Can anyone produce a backtested system that has comparable
real time trading results? I stll believe that there will be a divergence
between the two with time if no further tweaking is done to the system.
Comments?
Bob Brickey's post is reproduced below.
-Tony Haas
>Resent-Date: Wed, 25 Feb 1998 13:21:28 -0800 (PST)
>To: Omega Mailing List <omega-list@xxxxxxxxxxxxxxx>
>Subject: Statistical Significance
>Date: Wed, 25 Feb 98 14:19:44 -0500
>From: Scientific Approaches <sci@xxxxxxxxxx>
>X-Mailer: E-Mail Connection v2.5.03
>Resent-From: omega-list@xxxxxxxxxx
>X-Mailing-List: <omega-list@xxxxxxxxxx> archive/latest/14872
>X-Loop: omega-list@xxxxxxxxxx
>Resent-Sender: omega-list-request@xxxxxxxxxx
>
>Someone wrote:
>
>> What my rules / guidelines / suggestions try to do is
>> simply weed out the dishonest or even sloppy system vendors
>> by requiring documented real time performance results...<SNIP>
>
>Good real-time results tend to give us all a warm fuzzy feeling. However,
>contrary to the belief of most traders, real-time trading results often mean
>very little. They don't mean much, unless there have been enough trades to
>have statistical significance.
>
>The number of trades required for statistical significance is a complicated
>issue in a game like trading. It depends both on the percentage of
>profitable trades during a test period and the ratio of the average amount
>won to the average amount lost per trade.
>
>The result of 100 real-time trades generally means almost nothing, given
>typical win to loss and win-amount to loss-amount ratios. The result of 1,
>000 real-time trades under the same conditions typically has a little
>significance, but not nearly as much as most traders would think. A mere 1,
>000 trades can be very misleading. From a scientific perspective about all
>that could be said is that the result is interesting and probably warrants
>further study. The result of 10,000 trades typically is significant enough
>that it can be said there is a degree of scientific evidence, but even that
>number can be misleading.
>
>The problem with the "I won't trade a system without real-time trading
>proof" philosophy is it is practically impossible for statistically
>significant real-time results to exist. A trading method that has been
>found profitable in a back-test over 10,000 trades is statistically much
>more likely to work into the future than a system that happened to be
>profitable over a mere 100 trades in real-time, regardless of what most
>traders might think.
>
>Anyone who is going to limit their trading methods to systems proven to work
>in real-time must consider only systems that were developed a very long time
>ago and that have been traded consistently without modification since. They
>are not easy to find.
>
> -Bob Brickey
> Scientific Approaches
> sci@xxxxxxxxxx
|