Posted by leekaufman on 12th of May 2009 at 10:29 am
Matt, given the code that you use, and therefore the backtesting
results that you highlight in the daily/weekend updates, did your
SRS 15 minute system take the 9:30am buy today?
I see whether or not to have taken the buy at the open today
as completely and only dependent upon the "code" one uses
(tradestation easylanguage in my case), not being discretionary. I
guess in my mind if the trade is discretionary, it's not really a
"system".
Some of the users have suggested that since this is an
"end-of-day" system, one would not take the trade at the open since
it is triggered by a EMA cross from the prior day.
I can of course code it either way, but I'd like to know whether
BPT views a cross on the last candle of the prior day as a concrete
signal to take a position on the open the next day.
This may seem a bit trivial, but with the usual volatility in
these ETFs in the last 30 minutes, and more often in the last few
minutes, these end of day crosses are acually not that rare.
Posted by mamaduck on 12th of May 2009 at 10:38 am
Matt, I echo Lee's sentiments.
The question is not today's trade (which we still don't know if
it should have been taken or not, nor do we know if it is going to
turn out to be profitable or not).
A system is a system. As you so often mention, in order to
achieve those results, one must take every trade. So the question
is what does the system say about EOD cross and buying/shorting on
open.
And does the same rule apply to other EOD systems?
My own backtesting leaves this in, and is still pretty
impressive. So the question is to think systematically and to trade
based on a system not discretion. Even if the system says do
something or another on a gap up/down, we still need to program
that in.
FWIW -- my programming for the systems is continuous -- so if a
cross occurs on the last candle of the day, the trade is entered at
the opening of the first candle of the next day. When I
posted the other day I was thinking I had programmed to start each
day fresh. Therefore this is what all my backtesting is based
on, and all my systems. It would be interesting to program it
the other way -- to start every day fresh and compare the
two. It may be that this occurs infrequently enough, and the
next day moves are varied enough, that it doesn't matter
much.
the systems are all back tested with every trade regardless of
gaps. In my experience, most trades work out, even those with
gaps. You would have to manually go through all the trades
and take them out to see the difference, or write a program that
takes out the gaps and see the difference in back testing
results. Also, the other day someone here posted the stats of
the SRS system for each hr of the day
But again, I leave the sytems unoptizmed, i.e. they are raw
simple systems. I leave it up to you guys to improve them and
make modifications. Like a house you buy, choose to keep it
as it is, or make improvements additions to the house
anyway, here's the post from the other day from davetrader:
Here is a study I did
on
SRSI thought I would share for
those interested, it breaks down trade history YTD based on the
hour of the day the trade is taken. I used Matt's setup on
the 15 minute chart (9/39 EMA cross and sell at EOD). This
study was completed using StrategyDesk which of course is subject
to error.
The numbers represent using $4,000 dollars for
each trade taken. Draw your own conclusions from the
information in this table, except I will comment “for whatever
reason the 1330 to 1430 time period seems to be a low win
probability time of the day”.
Note: Average trade,
average win and average loss are all dollar amounts.
I have to agree that, even by looking at the real time dynamic
link above for the SRS 15 min. system, it crossed just after
3:30 yesterday. It's been repeated 100 times and is in
the mechanical systems explanations page, you only take the trade
on the day that it crosses, you may be out for several days, which
would include today, since it crossed yesterday. And if your
system at Tradestation, TD Ameritrade is set up properly to only
take the trade when it crosses, it will never take it at the end of
the day at 4PM because you're already out, but will take it at 3:30
or 3:45, which it would've done yesterday since they only crossed
yesterday. Anything else is a discretionary trade and not
going by when the EMAs actually crossed, especially if it's set up
to run automatically while you're at work.
Here's all I could find from Matt, all interpretations regarding
today's SRS signal welcome (including Matt's!) as there seem
to be different opinions on the blog from long standing
members:
matt- January 13, 2009 04:02pm
Previously posted by mamaduck (Title: SRS 15 min
mechanical) :(
toggle)
Matt, what happens if the cross happens at the end of the
last candleof the day? Does that mean tomorrow's opening is the
buy?
The EMA's could not technically cross on the
last candleduring the day because it would take the next day's opening
candle and therefore it would be next day's trade. The EMA's
could cross on the second to
last candlebut not the last one
It does not work well when u change the rules...so whatever they
are stick with them, you can't change them halfway through to make
it look good....2 cents chucked in
SRS Mechanical System
Posted by leekaufman on 12th of May 2009 at 10:29 am
Matt, given the code that you use, and therefore the backtesting results that you highlight in the daily/weekend updates, did your SRS 15 minute system take the 9:30am buy today?
I see whether or not to have taken the buy at the open today as completely and only dependent upon the "code" one uses (tradestation easylanguage in my case), not being discretionary. I guess in my mind if the trade is discretionary, it's not really a "system".
Some of the users have suggested that since this is an "end-of-day" system, one would not take the trade at the open since it is triggered by a EMA cross from the prior day.
I can of course code it either way, but I'd like to know whether BPT views a cross on the last candle of the prior day as a concrete signal to take a position on the open the next day.
This may seem a bit trivial, but with the usual volatility in these ETFs in the last 30 minutes, and more often in the last few minutes, these end of day crosses are acually not that rare.
Best, Lee
Mechanicals
Posted by mamaduck on 12th of May 2009 at 10:38 am
Matt, I echo Lee's sentiments.
The question is not today's trade (which we still don't know if it should have been taken or not, nor do we know if it is going to turn out to be profitable or not).
A system is a system. As you so often mention, in order to achieve those results, one must take every trade. So the question is what does the system say about EOD cross and buying/shorting on open.
And does the same rule apply to other EOD systems?
My own backtesting leaves this in, and is still pretty impressive. So the question is to think systematically and to trade based on a system not discretion. Even if the system says do something or another on a gap up/down, we still need to program that in.
tia
what site do you use
Posted by marita on 15th of May 2009 at 12:17 pm
what site do you use to backtest?
FWIW -- my programming for
Posted by Michael on 15th of May 2009 at 12:26 pm
FWIW -- my programming for the systems is continuous -- so if a cross occurs on the last candle of the day, the trade is entered at the opening of the first candle of the next day. When I posted the other day I was thinking I had programmed to start each day fresh. Therefore this is what all my backtesting is based on, and all my systems. It would be interesting to program it the other way -- to start every day fresh and compare the two. It may be that this occurs infrequently enough, and the next day moves are varied enough, that it doesn't matter much.
the systems are all back
Posted by matt on 15th of May 2009 at 12:26 pm
the systems are all back tested with every trade regardless of gaps. In my experience, most trades work out, even those with gaps. You would have to manually go through all the trades and take them out to see the difference, or write a program that takes out the gaps and see the difference in back testing results. Also, the other day someone here posted the stats of the SRS system for each hr of the day
But again, I leave the sytems unoptizmed, i.e. they are raw simple systems. I leave it up to you guys to improve them and make modifications. Like a house you buy, choose to keep it as it is, or make improvements additions to the house
anyway, here's the post from the other day from davetrader:
Here is a study I did on SRSI thought I would share for those interested, it breaks down trade history YTD based on the hour of the day the trade is taken. I used Matt's setup on the 15 minute chart (9/39 EMA cross and sell at EOD). This study was completed using StrategyDesk which of course is subject to error. The numbers represent using $4,000 dollars for each trade taken. Draw your own conclusions from the information in this table, except I will comment “for whatever reason the 1330 to 1430 time period seems to be a low win probability time of the day”.
Note: Average trade, average win and average loss are all dollar amounts.
Time
W/L
% Win
PF
Average Trade
Average Win
Average Loss
Overall % G/L
0930 to 1030
6/2
75
3.39
99
188
-166
2.5
1030 to 1130
2/1
66
3.7
136
286
-151
3.3
1130 to 1230
4/1
80
9.45
169
237
-100
4.1
1230 to 1330
3/1
75
4.26
121
211
-149
3
1330 to 1430
2/4
33
0.61
-33
160
-130
-1
1430 to 1530
3/1
75
5.7
49
79
-41
1.1
SRS Cross
Posted by curtis on 12th of May 2009 at 10:54 am
I have to agree that, even by looking at the real time dynamic link above for the SRS 15 min. system, it crossed just after 3:30 yesterday. It's been repeated 100 times and is in the mechanical systems explanations page, you only take the trade on the day that it crosses, you may be out for several days, which would include today, since it crossed yesterday. And if your system at Tradestation, TD Ameritrade is set up properly to only take the trade when it crosses, it will never take it at the end of the day at 4PM because you're already out, but will take it at 3:30 or 3:45, which it would've done yesterday since they only crossed yesterday. Anything else is a discretionary trade and not going by when the EMAs actually crossed, especially if it's set up to run automatically while you're at work.
Missing the SRS Forest ...
Posted by cdjd on 12th of May 2009 at 10:52 am
My understanding: Matt EOD system ignores yeasterday's crosses.
Each day is a new day.
So there was noSRS trigger of 15 min this am.
HOWEVER: using the yesterday's cross is another (non Matt) related (sibling) system.
It may be more or less profitable than the brother but is not the older brother.
Don't get lost in the details. If you are on automatic it doesn't matter.
If you are in semi automatic you can deviate from the system if you choose.
The world will not end.
You just wont be able to emulate exactly Matt's stats. That is All.
Entry at open on EOD systems
Posted by bkout3 on 12th of May 2009 at 10:51 am
Here's all I could find from Matt, all interpretations regarding today's SRS signal welcome (including Matt's!) as there seem to be different opinions on the blog from long standing members:
Matt, what happens if the cross happens at the end of the last candle of the day? Does that mean tomorrow's opening is the buy?
A system is a system
Posted by ravun on 12th of May 2009 at 10:50 am
It does not work well when u change the rules...so whatever they are stick with them, you can't change them halfway through to make it look good....2 cents chucked in
A system
Posted by mamaduck on 12th of May 2009 at 10:52 am
Agreed. The issue right now is there is confusion about what the system is... or maybe I am the only one confused here.