I'm trying to test the new SRS swing system using SD -- so far I
have not been able to duplicate Matt's results. Is anyone else
doing backtesting using SD that would care to try? Maybe you can
see if I have errors in my formulas. Some of the parameters are
below so that we can see if we get the same results. I show a
profit with the regular 9/39 system of $26,929. per the set-up
below with 59% winners and an average trade gain of $163. However
the swing system only gained $22,133. had 34% winners and an
average trade gain of $123. Seems not worth it unless I've
made a mistake somewhere :
Both systems set to $10,000. positions, 1/1/2008 to 12/31/2008,
exit at end of testing, 15 min with override interval checked
Comparison system -- 9/39 15min with EOD exits (this one seems
to work correctly).
Long Entry:
(Bar[Hour,15]=15 AND Bar[Minute,15]=45)=0 AND
ExpMovingAverage[EMA,Close,9,0,15]>ExpMovingAverage[EMA,Close,39,0,15]
AND
ExpMovingAverage[EMA,Close,9,1,15]
Long Exit:
Bar[Hour,15]=15 AND Bar[Minute,15]=45 OR
(ExpMovingAverage[EMA,Close,9,0,15] ExpMovingAverage[EMA,Close,9,1,15]>ExpMovingAverage[EMA,Close,39,1,15]
)
Short Entry:
(Bar[Hour,15]=15 AND Bar[Minute,15]=45)=0 AND
ExpMovingAverage[EMA,Close,9,0,15] ExpMovingAverage[EMA,Close,9,1,15]>ExpMovingAverage[EMA,Close,39,1,15]
Short Exit:
Bar[Hour,15]=15 AND Bar[Minute,15]=45 OR
(ExpMovingAverage[EMA,Close,9,0,15]>ExpMovingAverage[EMA,Close,39,0,15]
AND
ExpMovingAverage[EMA,Close,9,1,15]
************
New system: remove EOD exit and Add 10.5% stop loss and 21%
profit target (also removed rule to prevent entry on last bar of
day)
Long Entry:
ExpMovingAverage[EMA,Close,9,0,15]>ExpMovingAverage[EMA,Close,39,0,15]
AND
ExpMovingAverage[EMA,Close,9,1,15]
Long Exit:
(ExpMovingAverage[EMA,Close,9,0,15] ExpMovingAverage[EMA,Close,9,1,15]>ExpMovingAverage[EMA,Close,39,1,15]
) OR
(Bar[Close,15](Bar[Close,15]>EntryPrice*1.21)
(ExpMovingAverage[EMA,Close,9,0,15]>ExpMovingAverage[EMA,Close,39,0,15]
AND
ExpMovingAverage[EMA,Close,9,1,15](Bar[Close,15](Bar[Close,15]>EntryPrice*1.105
Thanks to Junkboxmayl(SP?) I found an error in the formulas I
posted before. I have the 1 in the offset position rather than the
previous bar's position. Here's my message to him:
Wow am I glad you wrote and is my face red!! My formula is
wrong! I have
the comparison average offset by 1 rather than set to give
the previous
bar's value which is what the SD formula does. Offsetting by
1 on a chart
shifts the MA one bar to the left. I'm not sure what it does
in a back test.
Anyway if it was some freaky thing that worked better anyway
we might use it
but it doesn't. See attached backtest comparing the 10/49
system for QLD
with the offset version vs the corrected version. To be clear
the correct
long rule is :
(Bar[Hour,15]=15 AND Bar[Minute,15]=45)=0 AND
(ExpMovingAverage[EMA,Close,10,0,15] >
ExpMovingAverage[EMA,Close,49,0,15]
AND
ExpMovingAverage[EMA,Close,10,0,15,1] <=
ExpMovingAverage[EMA,Close,49,0,15,1])
Similar correction for the other signals.
I'll have to go through all my set-ups and see if they need to
be corrected.
Note the interesting data that at least over the backtest the
way to go was
to short QLD/short QID. I have to check with other time
periods.
Thanks for helping me catch that.
[Thought I should post in case anyone else copied those
formulae Sorry for the trouble!]
Posted by averageguy on 11th of Feb 2009 at 09:57 am
(ExpMovingAverage[EMA, close,9,0,15,1]>ExpMovingAverage[EMA,
close,39,0,15,1]AND........rest repeats except 2 in place of 1 .
worked well for me .pm & I`ll give u phone # .
Posted by junkmaylbox on 11th of Feb 2009 at 01:51 pm
Do you get stable alerts with the formulas, as you posted them?
I don't get alerts even though the formulas have been entered. I am
not using any strategy, by the way. I am using version 3.0. I tried
upgrading to the latest 3.1 this morning, and began to see the
program crash within minutes of a restart. I had to uninstall the
update, delete the registry keys and run the installer for 3.0.
Let me know your set-up, if it's different. Thanks for your
help!
But I *think* that unless you check the box that says "Test
every "X" min inside the interval bar the signal will not trigger
till the close of the bar. Let me know if that's not right but
that's the way it looks in backtesting. Admittedly I auto trade
with Ninja and IB even though I like to backtest with SD as well as
Ninja so I guess it's possible that a rule that works for
backtesting might not work in real time trading.
Newsletter
Subscribe to our email list for regular free market updates
as well as a chance to get coupons!
Title: Strategy Desk users I'm trying
Posted by bkout3 on 2nd of Feb 2009 at 12:34 pm
I'm trying to test the new SRS swing system using SD -- so far I have not been able to duplicate Matt's results. Is anyone else doing backtesting using SD that would care to try? Maybe you can see if I have errors in my formulas. Some of the parameters are below so that we can see if we get the same results. I show a profit with the regular 9/39 system of $26,929. per the set-up below with 59% winners and an average trade gain of $163. However the swing system only gained $22,133. had 34% winners and an average trade gain of $123. Seems not worth it unless I've made a mistake somewhere :
Both systems set to $10,000. positions, 1/1/2008 to 12/31/2008, exit at end of testing, 15 min with override interval checked
Comparison system -- 9/39 15min with EOD exits (this one seems to work correctly).
Long Entry:
(Bar[Hour,15]=15 AND Bar[Minute,15]=45)=0 AND
ExpMovingAverage[EMA,Close,9,0,15]>ExpMovingAverage[EMA,Close,39,0,15] AND
ExpMovingAverage[EMA,Close,9,1,15]
Long Exit:
Bar[Hour,15]=15 AND Bar[Minute,15]=45 OR (ExpMovingAverage[EMA,Close,9,0,15]
Short Entry:
(Bar[Hour,15]=15 AND Bar[Minute,15]=45)=0 AND
ExpMovingAverage[EMA,Close,9,0,15]
Short Exit:
Bar[Hour,15]=15 AND Bar[Minute,15]=45 OR (ExpMovingAverage[EMA,Close,9,0,15]>ExpMovingAverage[EMA,Close,39,0,15] AND
ExpMovingAverage[EMA,Close,9,1,15]
************
New system: remove EOD exit and Add 10.5% stop loss and 21% profit target (also removed rule to prevent entry on last bar of day)
Long Entry:
ExpMovingAverage[EMA,Close,9,0,15]>ExpMovingAverage[EMA,Close,39,0,15] AND
ExpMovingAverage[EMA,Close,9,1,15]
Long Exit:
(ExpMovingAverage[EMA,Close,9,0,15]
(Bar[Close,15]
Short Entry:
ExpMovingAverage[EMA,Close,9,0,15]
Short Exit:
(ExpMovingAverage[EMA,Close,9,0,15]>ExpMovingAverage[EMA,Close,39,0,15] AND
ExpMovingAverage[EMA,Close,9,1,15]
Anyone care to take a crack at it?
Title: Strategy Desk users --
Posted by bkout3 on 11th of Feb 2009 at 09:34 am
Thanks to Junkboxmayl(SP?) I found an error in the formulas I posted before. I have the 1 in the offset position rather than the previous bar's position. Here's my message to him:
Wow am I glad you wrote and is my face red!! My formula is wrong! I have
the comparison average offset by 1 rather than set to give the previous
bar's value which is what the SD formula does. Offsetting by 1 on a chart
shifts the MA one bar to the left. I'm not sure what it does in a back test.
Anyway if it was some freaky thing that worked better anyway we might use it
but it doesn't. See attached backtest comparing the 10/49 system for QLD
with the offset version vs the corrected version. To be clear the correct
long rule is :
(Bar[Hour,15]=15 AND Bar[Minute,15]=45)=0 AND
(ExpMovingAverage[EMA,Close,10,0,15] > ExpMovingAverage[EMA,Close,49,0,15]
AND
ExpMovingAverage[EMA,Close,10,0,15,1] <=
ExpMovingAverage[EMA,Close,49,0,15,1])
Similar correction for the other signals.
I'll have to go through all my set-ups and see if they need to be corrected.
Note the interesting data that at least over the backtest the way to go was
to short QLD/short QID. I have to check with other time periods.
Thanks for helping me catch that.
[Thought I should post in case anyone else copied those formulae Sorry for the trouble!]
(ExpMovingAverage[EMA, close,9,0,15,1]>ExpMovingAverage[EMA, close,39,0,15,1]AND........rest repeats except 2
Posted by averageguy on 11th of Feb 2009 at 09:57 am
(ExpMovingAverage[EMA, close,9,0,15,1]>ExpMovingAverage[EMA, close,39,0,15,1]AND........rest repeats except 2 in place of 1 . worked well for me .pm & I`ll give u phone # .
StrategyDesk formulas
Posted by junkmaylbox on 11th of Feb 2009 at 01:51 pm
Do you get stable alerts with the formulas, as you posted them? I don't get alerts even though the formulas have been entered. I am not using any strategy, by the way. I am using version 3.0. I tried upgrading to the latest 3.1 this morning, and began to see the program crash within minutes of a restart. I had to uninstall the update, delete the registry keys and run the installer for 3.0.
Let me know your set-up, if it's different. Thanks for your help!
Yes I agree. You need
Posted by twins on 11th of Feb 2009 at 10:07 am
Yes I agree. You need the bar delays set to 1 and 2, not 0 and 1, to avoid inadvertant crossovers during the intervals.
Bob in NJ
Title: Strategy desk rules But I
Posted by bkout3 on 11th of Feb 2009 at 02:59 pm
But I *think* that unless you check the box that says "Test every "X" min inside the interval bar the signal will not trigger till the close of the bar. Let me know if that's not right but that's the way it looks in backtesting. Admittedly I auto trade with Ninja and IB even though I like to backtest with SD as well as Ninja so I guess it's possible that a rule that works for backtesting might not work in real time trading.