|
PureBytes Links
Trading Reference Links
|
Thanks for the tip Stephanos !
Theo Lockefeer
----- Original Message -----
From: "SR" <raftsp@xxxxxxxxx>
To: <metastock@xxxxxxxxxxxxxxxxxx>
Sent: Thursday, November 22, 2001 6:45 PM
Subject: More on MS Explorer
>
> Here is another possible use of the MS Explorer for calculating the
> profitability of its own past explorations or for back testing a system
> during a specified period. I have been away for some time now, so I don't
> know if any other members have posted something like this in the
meanwhile.
> I am aware of the fact that there have been a lot of interesting and more
> sophisticated exploring methods using the MS Explorer for back testing,
but
> this one might be also useful, since it's quite simple, fast and easily
> adaptable to various needs.
>
> Suppose we had performed an exploration on date xx/xx/xx, based on a set
of
> conditions, like <myCondition1> and <myCondition2> and ..<myCondition10>,
> all part of the Explorer's Filter.
> Say that 3 securities had passed our strict filter and had been candidates
> for long entry at the next bar's Open.
> 10 periods after our long entry we want to check the performance of the
> above securities, to evaluate our conditions either as part of a system or
> as part of the exploration.
>
> The following method checks the %profit gained 1, 2, 3, 4, 5 and 10
periods
> after our position has been opened.
>
> 1. Copy -paste the code into the corresponding fields of the Explorer.
>
> Column A
> Col Name: H1 %
> Code: (Ref(HIGH,-9)-Ref(OPEN,-10))*100/Ref(OPEN,-10) {%Gain 1 period after
> long
> entry}
>
> Column B
> Col Name: H2 %
> Code: (Ref(HIGH,-8)-Ref(OPEN,-10))*100/Ref(OPEN,-10) {%Gain 2 periods
after
> long
> entry}
>
> Column C
> Col Name: H3 %
> Code: (Ref(HIGH,-7)-Ref(OPEN,-10))*100/Ref(OPEN,-10) {%Gain 3 periods
after
> long
> entry}
>
> Column D
> Col Name: H4 %
> Code: (Ref(HIGH,-6)-Ref(OPEN,-10))*100/Ref(OPEN,-10) {%Gain 4 periods
after
> long
> entry}
>
> Column E
> Col Name: H5 %
> Code: (Ref(HIGH,-5)-Ref(OPEN,-10))*100/Ref(OPEN,-10) {%Gain 5 periods
after
> long
> entry}
>
> Column F
> Col Name: H10 %
> Code: (H-Ref(OPEN,-10))*100/Ref(OPEN,-10) {%Gain 10 periods after long
> entry ( the
> Exploration Date)}
>
> Filter
> Entry:= <myCondition1> and <myCondition2> and ..<myCondition10> ;
> Ref(entry,-11)=1 {since we have entered long 10 periods ago, the signal
has
> been generated 11 periods ago}
>
> 2. Click on the Options button on the Exploration Editor, check the radio
> button "Specific date" and enter the date for which you want the
exploration
> performed. Click OK, OK, Explore.
>
> 3. After the exploration is done click Reports:
>
> The results in this case concern the maximum profits possible on each one
of
> the above dates (periods).
> For example the column H1 % shows the maximum %profit possible 1 period
> after our long entry, whereas H10 % shows the maximum %profit possible 10
> periods after our long entry, which is the date for which we performed the
> exploration.
>
> This is a simple method, not taking into account commissions or slippage.
> Nevertheless, it might be useful if one wants to have a rough idea of the
> profitability of his trading rules, performing back-tests rather
> "horizontally" (on multiple securities, on various dates, for a few
specific
> date records) than "vertically" (on one security, on many adjoining date
> records).
> An additional advantage of this method is that one can specify custom
entry
> or exit prices, instead of fixed ones like Close, High, Low, Open.
> Furthermore, one is able to get an idea of how "crowded" his explorations
> usually are, since he knows how many securities pass his filter's
> conditions, on various dates.
>
> Needless to say, that we can easily use Close in the place of High if we
> rather want to base the profitability on more "realistic" conditions than
> calculating max profits.
> We may also want to check the profits on other than the above-specified
> dates, for example on 5, 10, 15, 20 and 25 periods after our long entry.
> What's more, we can calculate the %drawdown on specific dates after long
> entry, by substituting High for Low prices. We can back-test for short
> entries instead of long ones, etc.
>
> I hope this might be useful for some of you.
> Happy analysis
> Stephanos
>
>
>
>
>
>
>
|