|
PureBytes Links
Trading Reference Links
|
By
removing the stops and examinng the MFE/MAE, (IMO) I can see how good my "buy"
signal is over some period of time. You mention 15 days and that
sounds reasonable. I like to look at an MFE chart of my "buys" that
were held for exactly 15 days. I can get a feel for how many
trades went to various profit levels and I can see how far against me the trades
went. The later is handy for determing whether the system lends itself to
using stop losses. I may see, for instance, that the largest MAE is
50%. So, I can place my stop loss at 50% knowing that I at least
have a stop loss in place, even though it may almost never get
hit.
<FONT face=Arial color=#0000ff
size=2>
I may
then try reducing the holding period to 10 days. If the MFE looks
similar, I know that there is no reason to hold for more than ten
days. If the MAE looks a lot better, then I know that my buy signals
are valid for ten days, but not 15. Nothing wrong with that
information, but I like to see it.
<FONT face=Arial color=#0000ff
size=2>
I saw
a lot of people in this group "knocking" MFE/MAE when TJ released it in recent
betas. I'm a big fan of using it and I was doing it myself before AB
included the functionality.
<BLOCKQUOTE
>
<FONT face="Times New Roman"
size=2>-----Original Message-----From: Dave Merrill
[mailto:dmerrill@xxxxxxx]Sent: Saturday, November 22, 2003 1:45
PMTo: amibroker@xxxxxxxxxxxxxxxSubject: RE: [amibroker]
an entry signal evaluation model
<SPAN
class=274462918-22112003>thought of that, but I think the AFL approach gets
you some info that the built-in MFE/MAE stats don't.
<SPAN
class=274462918-22112003>
<FONT
size=+0>basically, you get stats for all
bars, side by side with stats for buys only. readouts for each bar cover the
spec'd number of bars into the future, or until the loss stop is hit,
whichever comes first.
<FONT
size=+0><SPAN
class=274462918-22112003>
<FONT
size=+0>if you buy the time and loss stop idea,
I think this is exactly what you want to know, and I couldn't figure out how
to get it out of MFE/MAE (not that those stats aren't great to have
too).
<FONT
size=+0><SPAN
class=274462918-22112003>
<FONT
size=+0>since you've been doing something like
this already, have you found it to be a useful way to evaluate entry signals?
if so, would you mind sharing which ones are looking good from this
perspective?
<FONT
size=+0><SPAN
class=274462918-22112003>
<FONT
size=+0><SPAN
class=274462918-22112003>dave
<SPAN
class=274462918-22112003>
<BLOCKQUOTE
>
An
excellent concept, but I may be able to give you an idea to make it
easier.
<FONT face=Arial color=#0000ff
size=2>
I
do what you propose, except I eliminate the "stop". Just leave
an exit after so many days.
<FONT face=Arial color=#0000ff
size=2>
<FONT face=Arial color=#0000ff
size=2>Then, using the portfolio backtester or other software, have a look
at the MAE and MFE reports.
<BLOCKQUOTE
>
I've been working on a new (to me) way
of evaluating possible entry methods,as distinct from exit methods.
I've got code in the works, but I thought I'drun the concept by folks
here for some feedback while I finalize it. here'sthe basic
idea:on the long side, profits come from selling at a higher
price than youbought. so, a good entry signal is one where the stock
frequently risesreasonably soon after a buy signal occurs. with the
exception of a permanentdown-trend or individual stock fatalities,
*everything* goes up eventually,so the question really is, how soon
after buys does price rise, and how far?this exploration makes two
assumptions, after Steve Karnish: 1) a relativelytight stoploss is
needed to limit risk and mental pain, and 2) we're lookingfor short
term signals, to generate a statistically meaningful number oftrades,
and to limit the opportunity cost of continuing to hold a
positionthat's treading water. specifically, we'll sell anything that
hits a 13%stoploss or is still held after 15 days. these aren't the
only possibleassumptions, but they're what's used here.given
those assumptions, the quality of an entry signal can be evaluated
asthe peak percentage rise in price, after commissions, before one of
thosetwo basic exit conditions occurs. to actually trade the signal
profitably,you might well need to exit before the loss or time stops
took you out, andthe design and testing of an exit signal specific to
a given entry model isbeyond the scope of this effort. however, unless
price reaches a healthyprofit fairly often before those basic stop
exits happen, the entry signalitself isn't useful; you could do better
with random entries.peak price rise before those time or loss
stops are hit can be evaluated forany bar, not just when there's a buy
signal. the efficiency of a buy signalcan then be expressed as the
percentage ratio of the average peak rise onbuy bars to that of all
bars. greater than 100% means it entered on barswith higher than
average gain before stopping out, less than 100%, lowerthan
average.to design a system using this idea, you'd need to select a
set of issues anda date range to test, evaluate a variety of entry
methods on this basis,then test a variety of exit signals with the top
entry methods and choosethe best combination.make sense?
is this a known concept that others have explored?
otherthoughts?daveSend
BUG REPORTS to bugs@xxxxxxxxxxxxxSend SUGGESTIONS to
suggest@xxxxxxxxxxxxx-----------------------------------------Post
AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx (Web page: <A
href="">http://groups.yahoo.com/group/amiquote/messages/)--------------------------------------------Check
group FAQ at: <A
href="">http://groups.yahoo.com/group/amibroker/files/groupfaq.html
Your use of Yahoo! Groups is subject to the <A
href="">Yahoo! Terms of Service.
Yahoo! Groups Sponsor
ADVERTISEMENT
Send BUG REPORTS to bugs@xxxxxxxxxxxxx
Send SUGGESTIONS to suggest@xxxxxxxxxxxxx
-----------------------------------------
Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx
(Web page: http://groups.yahoo.com/group/amiquote/messages/)
--------------------------------------------
Check group FAQ at: http://groups.yahoo.com/group/amibroker/files/groupfaq.html
Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
|