PureBytes Links
Trading Reference Links
|
Thanks a lot for the code. I have archived your message for later use. Great contribution.
Check out the good thread on expectancy back on June 21 and 22 (post no.'s 20006, 20021, 20028, 20029, 20031, 20032, 20033, 20034, 20035, 20038, 20056, 20061, and 20066). I'd like to point out especially post 20056, not because it's mine, but because it provides another way to calculate expectancy that may even be simpler to code than your code. There is an explanation there for why, if you are using position sizing in AB, you cannot use the average win/average loss ratio that AB gives you. Maybe this and the other posts may answer some of the questions you posed. I'll try to answer your questions as best I can:
1. If, by compounded trading methods, you mean pyramiding, you'd have to doit trade by trade and accumulate everything. That's why normalizing to risk may make it easier. Mark provided a simple Excel spreadsheet displaying Tharp's method of calculating E (see post 20066). You just add up all your R-multiples to give you your present expectancy.
2. Yes, results SHOULD be different for different lengths of trading history. After all, the longer you trade, the more data you will have along with more wins and losses upon which to calculate expectancy. The more data you have, the better off you are.
3. Yes, results will be affected by single large wins or losses. But, that's real life. I don't think you'll ever get a nice normal distribution of trading profits in any system. You will experience a large R-multiple profit or loss occasionally. I believe that's a truism.
4. Time in market and trade durations should have nothing to do with expectancy. You can calculate expectancy for short term trading systems as well as long term systems. There is no 'time' term in the equation.
5. Volatility is not considered by the expectancy calculation, but it is inyour trading system. If you are using a volatility-based system, volatility will have a huge effect on your trading results, which eventually affectsexpectancy indirectly.
6. System rating based on expectancy and your equity curve is a good idea. The higher expectancy, the better the system.
7. It shouldn't matter what the Pay-off ratio uses in its calculation. If your average win is $1000/trade and your average loss if $500/trade, your payoff ratio is still 2, so your average win is 2 while your average loss is 1.
8. If the stock's price changes from 100 to 5 (presumably, you mean via themarket, not via splits), you should be going short during that time. So, your equity curve should still be increasing as the stock price declines. Expectancy is determined by winning and losing trades, not by price.
Hope this helps clear up some points. Good thread.
Al Venosa
----- Original Message -----
From: Herman van den Bergen
To: Amibroker@xxxx Com
Sent: Saturday, October 19, 2002 2:41 PM
Subject: [amibroker] A closer look at "Expectation" (afl)
Hello,
Thanks to William for posting the math formula, attached an afl version. I am not sure if this formula falls under MM or system development however Expectation can be quite easy incorporated into a system as a simple indicator or screen. Might be worthwhile trying.
See chart below. Some first thoughts and questions:
1) the results will be different for fixed and compounded trading methods, for which is this formula?
2) the results will be different for different length of data history
3) the results will be dependent on size of profits, simple averages can be swayed by single large profits.
4) time in market should be a factor, so should L&S trade durations.
5) Volatility is not considered.
6) imho, system ratings should be based on a moving Equity window which contains a significant number of trades.
7) shouldn't the Pay-Off ratio use normalized values, percent trade profit?
8) The stock's price can change from $100 to $5, this formula doesn't account for that.
Perhaps there is a more complete formula out there that we can code? If you use this formula to trade by you might want to play with the attached code. You could test it on different Range settings and note how the results change, try it on some of your trading systems and different stocks.
thanks for your participation, I invite further comments.
Best regards,
Herman.
------=_NextPart_001_0085_01C27754.4009FDB0
Content-Type: text/html;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=Content-Type content="text/html; charset=iso-8859-1">
<META content="MSHTML 6.00.2719.2200" name=GENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=#ffffff>
<DIV>Herman,</DIV>
<DIV> </DIV>
<DIV>Thanks a lot for the code. I have archived your message for later use.
Great contribution.</DIV>
<DIV> </DIV>
<DIV>Check out the good thread on expectancy back on June 21 and 22 (post no.'s
20006, 20021, 20028, 20029, 20031, 20032, 20033, 20034, 20035, 20038,
20056, 20061, and 20066). I'd like to point out especially post 20056,not
because it's mine, but because it provides another way to calculate expectancy
that may even be simpler to code than your code. There is an explanation there
for why, if you are using position sizing in AB, you cannot use the average
win/average loss ratio that AB gives you. Maybe this and the other posts may
answer some of the questions you posed. I'll try to answer your questions as
best I can:</DIV>
<DIV> </DIV>
<DIV>1. If, by compounded trading methods, you mean pyramiding, you'd have to do
it trade by trade and accumulate everything. That's why normalizing to riskmay
make it easier. Mark provided a simple Excel spreadsheet displaying Tharp's
method of calculating E (see post 20066). You just add up all your R-multiples
to give you your present expectancy. </DIV>
<DIV> </DIV>
<DIV>2. Yes, results SHOULD be different for different lengths of trading
history. After all, the longer you trade, the more data you will have alongwith
more wins and losses upon which to calculate expectancy. The more data you have,
the better off you are.</DIV>
<DIV> </DIV>
<DIV>3. Yes, results will be affected by single large wins or losses. But,
that's real life. I don't think you'll ever get a nice normal distribution of
trading profits in any system. You will experience a large R-multiple profit or
loss occasionally. I believe that's a truism.</DIV>
<DIV> </DIV>
<DIV>4. Time in market and trade durations should have nothing to do with
expectancy. You can calculate expectancy for short term trading systems as well
as long term systems. There is no 'time' term in the equation. </DIV>
<DIV> </DIV>
<DIV>5. Volatility is not considered by the expectancy calculation, but it is in
your trading system. If you are using a volatility-based system, volatilitywill
have a huge effect on your trading results, which eventually affects expectancy
indirectly.</DIV>
<DIV> </DIV>
<DIV>6. System rating based on expectancy and your equity curve is a good idea.
The higher expectancy, the better the system.</DIV>
<DIV> </DIV>
<DIV>7. It shouldn't matter what the Pay-off ratio uses in its calculation.
If your average win is $1000/trade and your average loss if $500/trade, your
payoff ratio is still 2, so your average win is 2 while your average loss
is 1. </DIV>
<DIV> </DIV>
<DIV>8. If the stock's price changes from 100 to 5 (presumably, you mean via the
market, not via splits), you should be going short during that time. So, your
equity curve should still be increasing as the stock price declines. Expectancy
is determined by winning and losing trades, not by price. </DIV>
<DIV> </DIV>
<DIV>Hope this helps clear up some points. Good thread.</DIV>
<DIV> </DIV>
<DIV>Al Venosa</DIV>
<BLOCKQUOTE dir=ltr
style="PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
<DIV style="FONT: 10pt arial">----- Original Message ----- </DIV>
<DIV
style="BACKGROUND: #e4e4e4; FONT: 10pt arial; font-color: black"><B>From:</B>
<A title=psytek@xxxx href="mailto:psytek@xxxx">Herman vanden
Bergen</A> </DIV>
<DIV style="FONT: 10pt arial"><B>To:</B> <A title=amibroker@xxxxxxxxxx
href="mailto:Amibroker@x... Com">Amibroker@xxxx Com</A></DIV>
<DIV style="FONT: 10pt arial"><B>Sent:</B> Saturday, October 19, 2002 2:41
PM</DIV>
<DIV style="FONT: 10pt arial"><B>Subject:</B> [amibroker] A closer lookat
"Expectation" (afl)</DIV>
<DIV><BR></DIV>
<DIV><FONT face=Arial size=2><SPAN
class=660435416-19102002>Hello,</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN
class=660435416-19102002></SPAN></FONT> </DIV>
<DIV><FONT face=Arial size=2><SPAN class=660435416-19102002>Thanks to William
for posting the math formula, attached an afl version. I am not sureif
this formula falls under MM or system development however Expectation
can be quite easy incorporated into a system as a simple indicator or
screen. Might be worthwhile trying.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN
class=660435416-19102002></SPAN></FONT> </DIV>
<DIV><FONT face=Arial size=2><SPAN class=660435416-19102002>See chart below.
Some first thoughts and questions:</SPAN></FONT><FONT face=Arial size=2><SPAN
class=660435416-19102002></SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN
class=660435416-19102002></SPAN></FONT> </DIV>
<DIV><FONT face=Arial size=2><SPAN class=660435416-19102002>1) the
results will be different for fixed and compounded trading methods, for which
is this formula?</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN
class=660435416-19102002>2) </SPAN></FONT><FONT face=Arial size=2><SPAN
class=660435416-19102002>the results will be different for different
length of data history</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN class=660435416-19102002>3) the results
will be dependent on size of profits, simple averages can be swayed by
single large profits.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN class=660435416-19102002>4) timein market
should be a factor, so should L&S trade durations.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN class=660435416-19102002>5) Volatility is
not considered.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN class=660435416-19102002>6) imho, system
ratings should be based on a moving Equity window which contains a significant
number of trades.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN class=660435416-19102002>7) shouldn't the
Pay-Off ratio use normalized values, percent trade profit?</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN class=660435416-19102002>8) The stock's
price can change from $100 to $5, this formula doesn't account for
that.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN
class=660435416-19102002></SPAN></FONT> </DIV>
<DIV><FONT face=Arial size=2><SPAN class=660435416-19102002>Perhapsthere is a
more complete formula out there that we can code? </SPAN></FONT><FONT
face=Arial size=2><SPAN class=660435416-19102002>If you use this formula to
trade by you might want to play with the attached code. You could test iton
different Range settings and note how the results change, try it on some of
your trading systems and different stocks.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN
class=660435416-19102002></SPAN></FONT> </DIV>
<DIV><FONT face=Arial size=2><SPAN class=660435416-19102002>thanks for your
participation, I invite further comments.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN class=660435416-19102002>Best
regards,</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN
class=660435416-19102002>Herman.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN
class=660435416-19102002></SPAN></FONT> </DIV>
<DIV><IMG
src="cid:008301c27775$c70fb6d0$6401a8c0@xxxx"></DIV></BLOCKQUOTE></BODY></HTML>
------=_NextPart_001_0085_01C27754.4009FDB0--
Attachment:
Description: ""
|