g*****g
发帖数: 34805 | | b*****h
发帖数: 3386 | 2 没有precious metal, agriculture commodity ETFs.
【在 g*****g 的大作中提到】
: heihei
: http://www.ritholtz.com/blog/2009/03/how-well-does-diversification-work/
| g*****g
发帖数: 34805 | 3 You can add DBA, DBC and USO to the list. The only one acting
different may be gold.
【在 b*****h 的大作中提到】
: 没有precious metal, agriculture commodity ETFs.
| b*****h
发帖数: 3386 | 4 you are right. Agriculture ETFs are similiar, but I'm wondering if it is
good time now to invest on them...
【在 g*****g 的大作中提到】
: You can add DBA, DBC and USO to the list. The only one acting
: different may be gold.
| K****D
发帖数: 30533 | 5 what are these charts trying to show?
If you compare the components of a total market index fund, on the same
chart with the total market index itself, by definition, their performance
will be the same, i.e.,
index YTD = weighted average of (all components YTD)
Maybe the author was trying to say no sector can out-perform S&P 500
consistantly? That's not related to diversification ah.
【在 g*****g 的大作中提到】
: heihei
: http://www.ritholtz.com/blog/2009/03/how-well-does-diversification-work/
| K****D
发帖数: 30533 | 6 Ok, from another perspective, these two charts EXACTLY proved that
diversification works.
Since S&P's curve has lower volatility than any of the individual index
(strictly speaking, S&P does not include some of the indices like
European Index. I just don't understand why the author list them
in the same chart), it proves that diversification lowers the portfolio
beta, a.k.a. risk.
Investing 100% in S&P would be safer than investing 100% in, say, small
cap index.
【在 K****D 的大作中提到】
: what are these charts trying to show?
: If you compare the components of a total market index fund, on the same
: chart with the total market index itself, by definition, their performance
: will be the same, i.e.,
: index YTD = weighted average of (all components YTD)
: Maybe the author was trying to say no sector can out-perform S&P 500
: consistantly? That's not related to diversification ah.
| g*****g
发帖数: 34805 | 7 I think you should compare to Treasury Bond performance, or at least
cash performance. If we are in a depression like 1929, you sure you just
want to beat the S&P index?
【在 K****D 的大作中提到】
: what are these charts trying to show?
: If you compare the components of a total market index fund, on the same
: chart with the total market index itself, by definition, their performance
: will be the same, i.e.,
: index YTD = weighted average of (all components YTD)
: Maybe the author was trying to say no sector can out-perform S&P 500
: consistantly? That's not related to diversification ah.
| s********n
发帖数: 1962 | 8 You didn't get it. Diversification is to reduce risk. It's not to eliminate
risk. You are proposing a systematic risk of stock market, which, every
econ 101 textbook will tell you, cannot be eliminated by diversification
over stocks.
And generally over many of your posts, when we say some strategy is better,
by no way it means it works better all the time. It simply means it's
better in term of risk-reward ratio on a prabability base.
【在 g*****g 的大作中提到】
: I think you should compare to Treasury Bond performance, or at least
: cash performance. If we are in a depression like 1929, you sure you just
: want to beat the S&P index?
| h*******y
发帖数: 864 | 9 I was trying to avoid replying to goodbug's post as I consider that just a
wasted effort.
But since BM replied, I have something to add as well. So what bond
beats the stock in the last 40 years? We invest in stock NOT because
we think it does not have risks. I really look down those kind of "TA"
conclusions as I think those people are just completely clueless of
the history. You don't need your TA goddess to tell you that stock
has risks (of losing the principle) and you don't need TA to tell y
【在 s********n 的大作中提到】
: You didn't get it. Diversification is to reduce risk. It's not to eliminate
: risk. You are proposing a systematic risk of stock market, which, every
: econ 101 textbook will tell you, cannot be eliminated by diversification
: over stocks.
: And generally over many of your posts, when we say some strategy is better,
: by no way it means it works better all the time. It simply means it's
: better in term of risk-reward ratio on a prabability base.
| K****D
发帖数: 30533 | 10 I set my target as S&P because I trade stocks only, no options,
no bonds, no CDs. Natually S&P is the best target.
If I set my target as "the current best CD rate", my investment
strategy would be completely different. In years like 2008 that
means I could almost do nothing but put 95+% of my money in
cash. That's super boring and I don't like it at all. Even in years
like 2006, I probably still have to keep 70+% in cash. And my showoff
would become "Wow! My APY is 3.6% in 2006! Beat the best CD
【在 g*****g 的大作中提到】
: I think you should compare to Treasury Bond performance, or at least
: cash performance. If we are in a depression like 1929, you sure you just
: want to beat the S&P index?
| | | m******t
发帖数: 2416 | | g*****g
发帖数: 34805 | 12 Numbers don't lie. I'll show a very simple TA strategy here as attached.
Buy when 13 weekly when EMA crosses up 34 weekly EMA, and sell when crossing
down. Backtest on SPX (I have only data from 1950), the maximum single
trade loss is 7.05%, the max portfolio drawdown from peak is 15.84%, the
average yearly return is 9.98%, with 59.14% exposure (So that you can
collect some percent on CD when idle).
Now consider the average yearly return of SPX in the same period is 6.8%.
The max portfolio drawd
【在 h*******y 的大作中提到】
: I was trying to avoid replying to goodbug's post as I consider that just a
: wasted effort.
: But since BM replied, I have something to add as well. So what bond
: beats the stock in the last 40 years? We invest in stock NOT because
: we think it does not have risks. I really look down those kind of "TA"
: conclusions as I think those people are just completely clueless of
: the history. You don't need your TA goddess to tell you that stock
: has risks (of losing the principle) and you don't need TA to tell y
| d*****z
发帖数: 114 | 13 Time frame is misleading. Of course 5 year chart looks more diversified than
1 year chart. So that doesn't prove "Diversification works until it does".
From the charts it sounds to me like "Diversification works in 5 years , but
less effective in 1 year", because 1 year is too short to show the effect
of diversification, since market system risks of various (equity) markets
are more converged/correlated in short timeframe, but the converging/
correlation would weaken as timeframe increases.
【在 g*****g 的大作中提到】
: heihei
: http://www.ritholtz.com/blog/2009/03/how-well-does-diversification-work/
| d*****z
发帖数: 114 | 14 Right. Diversification is normally meant to reduce company or sector
specific risk. In theory market risk can not be eliminated through
traditional diversification.
eliminate
【在 s********n 的大作中提到】
: You didn't get it. Diversification is to reduce risk. It's not to eliminate
: risk. You are proposing a systematic risk of stock market, which, every
: econ 101 textbook will tell you, cannot be eliminated by diversification
: over stocks.
: And generally over many of your posts, when we say some strategy is better,
: by no way it means it works better all the time. It simply means it's
: better in term of risk-reward ratio on a prabability base.
| d*****z
发帖数: 114 | 15 Hmmm...
Until last year, I had got an even better strategy. Leverage my capital at 1
and buy any residential mortgage pool or mortgage-backed securities. I had
backtested to at least 1935. Since no single year of the 70+ year had a YOY
housing price decline (measured as consecutive YOY decline for more than 6
month), my investment's max draw down is <10% (even for leveraged capital).
Given the spread between libor and mortgage rate, the avearge return for
capital should be at least 10%. Does it | g*****g
发帖数: 34805 | 16 Certainly not, I show a strategy that works through multiple bear markets,
the strategy can escape early in 30's depression and today's crisis, while
yours is wiped out today. Also, since you leverage 1:5, you have to consider
your margin cost, assume you borrow at prime rate, you pay 20%+ of your
principle for interest, and only get back 10% in raw return? You are
under water!
If there's a permanent bull market, then congratulations, you find an
arbitrage. In the real world, however, you'd back
【在 d*****z 的大作中提到】
: Hmmm...
: Until last year, I had got an even better strategy. Leverage my capital at 1
: and buy any residential mortgage pool or mortgage-backed securities. I had
: backtested to at least 1935. Since no single year of the 70+ year had a YOY
: housing price decline (measured as consecutive YOY decline for more than 6
: month), my investment's max draw down is <10% (even for leveraged capital).
: Given the spread between libor and mortgage rate, the avearge return for
: capital should be at least 10%. Does it
| o*****c
发帖数: 241 | 17 no strategy works forever,just as no perpetual motion machine exists.
simple like that.
any reasonable strategy has its assumptions and needs to be closely
monitored and modified by trader himself.
consider
before
【在 g*****g 的大作中提到】
: Certainly not, I show a strategy that works through multiple bear markets,
: the strategy can escape early in 30's depression and today's crisis, while
: yours is wiped out today. Also, since you leverage 1:5, you have to consider
: your margin cost, assume you borrow at prime rate, you pay 20%+ of your
: principle for interest, and only get back 10% in raw return? You are
: under water!
: If there's a permanent bull market, then congratulations, you find an
: arbitrage. In the real world, however, you'd back
| g*****g
发帖数: 34805 | 18 I agree, TA is assuming history will repeat itself. And Livermore predicted
stock market will repeat itself since human nature of greedy and fear will
never change. At least he's right so far.
Any strategy can work if you do it right. Buffet doesn't do TA. But it's
funny some people will bash TA but couldn't provide a better strategy
than the simple one I mentioned.
【在 o*****c 的大作中提到】
: no strategy works forever,just as no perpetual motion machine exists.
: simple like that.
: any reasonable strategy has its assumptions and needs to be closely
: monitored and modified by trader himself.
:
: consider
: before
| K****D
发帖数: 30533 | 19 Biggest issue: how do you execute your strategy?
It's very obvious at present. But what about those days when the
two EMA curves cross each other often? Do you strictly trade whenever
the curves cross? Or do you set a dead-band threshold (as most electronic
engineerings do) and only trade if the trend is more obvious than the
threshold?
If the former, you pay a lot of commission (and time). If the latter,
you lose a couple of percents during the transition.
2nd issue: as I pointed out before, yo
【在 g*****g 的大作中提到】
: Numbers don't lie. I'll show a very simple TA strategy here as attached.
: Buy when 13 weekly when EMA crosses up 34 weekly EMA, and sell when crossing
: down. Backtest on SPX (I have only data from 1950), the maximum single
: trade loss is 7.05%, the max portfolio drawdown from peak is 15.84%, the
: average yearly return is 9.98%, with 59.14% exposure (So that you can
: collect some percent on CD when idle).
: Now consider the average yearly return of SPX in the same period is 6.8%.
: The max portfolio drawd
| m******t
发帖数: 2416 | 20
I thought you said you only had data since 1950?
【在 g*****g 的大作中提到】
: Certainly not, I show a strategy that works through multiple bear markets,
: the strategy can escape early in 30's depression and today's crisis, while
: yours is wiped out today. Also, since you leverage 1:5, you have to consider
: your margin cost, assume you borrow at prime rate, you pay 20%+ of your
: principle for interest, and only get back 10% in raw return? You are
: under water!
: If there's a permanent bull market, then congratulations, you find an
: arbitrage. In the real world, however, you'd back
| | | g*****g
发帖数: 34805 | 21 I backtest dji last night, I think I got about 9% return.
【在 m******t 的大作中提到】
:
: I thought you said you only had data since 1950?
| g*****g
发帖数: 34805 | 22
In the real world, I'd use a more sophisticated strategy. But yes, you
can trade strictly whenever it crosses. We are talking weekly close for
the indicator here, so it's really easy to anticipate a cross and follow.
Buy/sell at Friday's close at market price. You should easily get +/- 10cent
on SPY close.
This is not true, in former, SPX yields 28 trades in a 60 years span. As you
can see here.
http://www.mitbbs.com/mitbbs_article.php?board=Investment&id=31197357&ap=1787
And as I used weekly c
【在 K****D 的大作中提到】
: Biggest issue: how do you execute your strategy?
: It's very obvious at present. But what about those days when the
: two EMA curves cross each other often? Do you strictly trade whenever
: the curves cross? Or do you set a dead-band threshold (as most electronic
: engineerings do) and only trade if the trend is more obvious than the
: threshold?
: If the former, you pay a lot of commission (and time). If the latter,
: you lose a couple of percents during the transition.
: 2nd issue: as I pointed out before, yo
| K****D
发帖数: 30533 | 23 If your back test is based on weekly close and still yielded
9.xx%, then yes, you won't accumulate a lot of commissions
and I aglee that your performance beat the index.
(I couldn't click on the links in the attachment you posted.
I can only see the statistics page.)
Please report back results on WMT if you have time. From naked
eyes I observe that buying-and-holding is better, compared with
the SMA205 method. I didn't check with your new SMA method.
I originally thought those two SMA curves wou
【在 g*****g 的大作中提到】
:
: In the real world, I'd use a more sophisticated strategy. But yes, you
: can trade strictly whenever it crosses. We are talking weekly close for
: the indicator here, so it's really easy to anticipate a cross and follow.
: Buy/sell at Friday's close at market price. You should easily get +/- 10cent
: on SPY close.
: This is not true, in former, SPX yields 28 trades in a 60 years span. As you
: can see here.
: http://www.mitbbs.com/mitbbs_article.php?board=Investment&id=31197357&ap=1787
: And as I used weekly c
| K****D
发帖数: 30533 | 24 Well, some people just bash all methods, like me, hehe.
I am a believer that whether a method beats the market is
completely based on luck -- take my cat chart as an example,
haha.
In another word, when I predicted the market right, or my
performance beats the market, it doesn't boost my confidence
that I am gaining skills. I consider it as luck and am
reluctant to adjust my future trading strategy towards
the currently winning method.
【在 g*****g 的大作中提到】
:
: In the real world, I'd use a more sophisticated strategy. But yes, you
: can trade strictly whenever it crosses. We are talking weekly close for
: the indicator here, so it's really easy to anticipate a cross and follow.
: Buy/sell at Friday's close at market price. You should easily get +/- 10cent
: on SPY close.
: This is not true, in former, SPX yields 28 trades in a 60 years span. As you
: can see here.
: http://www.mitbbs.com/mitbbs_article.php?board=Investment&id=31197357&ap=1787
: And as I used weekly c
| g*****g
发帖数: 34805 | 25 It's 13 and 34 weekly EMA crosses I used, and it only generated
28 trades in about 59 years.
【在 K****D 的大作中提到】
: If your back test is based on weekly close and still yielded
: 9.xx%, then yes, you won't accumulate a lot of commissions
: and I aglee that your performance beat the index.
: (I couldn't click on the links in the attachment you posted.
: I can only see the statistics page.)
: Please report back results on WMT if you have time. From naked
: eyes I observe that buying-and-holding is better, compared with
: the SMA205 method. I didn't check with your new SMA method.
: I originally thought those two SMA curves wou
| s********n
发帖数: 1962 | 26 If you use this to argue if diversification works or not, it's kind of
like saying "my scissors works better than your hammer." --- What's the
point to compare anyway?
Now back to your strategy. Can you test one thing for us? You have tested
13/34 EMA cross. Can you also test 12, 13, 14 cross 33, 34, 35 EMA and
show us the result matrix? Then we'll discuss again. Thanks.
【在 g*****g 的大作中提到】
: It's 13 and 34 weekly EMA crosses I used, and it only generated
: 28 trades in about 59 years.
| p******h
发帖数: 1783 | 27 I think he should bash "buy and hold" instead of diversification.
【在 s********n 的大作中提到】
: If you use this to argue if diversification works or not, it's kind of
: like saying "my scissors works better than your hammer." --- What's the
: point to compare anyway?
: Now back to your strategy. Can you test one thing for us? You have tested
: 13/34 EMA cross. Can you also test 12, 13, 14 cross 33, 34, 35 EMA and
: show us the result matrix? Then we'll discuss again. Thanks.
:
| g*****g
发帖数: 34805 | 28 Maybe I don't understand what's diversification. But diversify is to
reduce your risk right? But I show a graph with
most sectors and cap going in the same direction, deep down. So IMHO,
diversification cannot prevent big loss, that's how I define "not work".
I do show a strategy that will prevent big loss, with max drawdown < 20%
and single trade loss < 10%. That's at least working better than
diversification for the purpose of diversification.
【在 s********n 的大作中提到】
: If you use this to argue if diversification works or not, it's kind of
: like saying "my scissors works better than your hammer." --- What's the
: point to compare anyway?
: Now back to your strategy. Can you test one thing for us? You have tested
: 13/34 EMA cross. Can you also test 12, 13, 14 cross 33, 34, 35 EMA and
: show us the result matrix? Then we'll discuss again. Thanks.
:
| s********n
发帖数: 1962 | 29 If there is a medicine for heart diseases, yet you keep saying "if it
cannot cure my liver, it's not a good medicine" then there is nothing
we can say.
Just do the test I mentioned please. Then we discuss. Thanks.
【在 g*****g 的大作中提到】
: Maybe I don't understand what's diversification. But diversify is to
: reduce your risk right? But I show a graph with
: most sectors and cap going in the same direction, deep down. So IMHO,
: diversification cannot prevent big loss, that's how I define "not work".
: I do show a strategy that will prevent big loss, with max drawdown < 20%
: and single trade loss < 10%. That's at least working better than
: diversification for the purpose of diversification.
| g*****g
发帖数: 34805 | 30 You have a heart disease medicine which can prevent you from dying, but
still paralysed nontheless. I show you a medicine which is meant for liver
maybe, but if you use for heart disease, it can prevent paralysis as well.
Then of course I can claim my liver medicine is a better
medicine for heart disease patients.
【在 s********n 的大作中提到】
: If there is a medicine for heart diseases, yet you keep saying "if it
: cannot cure my liver, it's not a good medicine" then there is nothing
: we can say.
: Just do the test I mentioned please. Then we discuss. Thanks.
| | | K****D
发帖数: 30533 | 31 As I said earlier, in your chart everything goes deep down
(even more than S&P) because the author intentionally or
non-intentionally left out some sectors (like consumer staples)
and inserted some sectors that do not belong to US market.
By definition, if everything in S&P goes deep down, S&P should
go down at the average rate.
【在 g*****g 的大作中提到】
: Maybe I don't understand what's diversification. But diversify is to
: reduce your risk right? But I show a graph with
: most sectors and cap going in the same direction, deep down. So IMHO,
: diversification cannot prevent big loss, that's how I define "not work".
: I do show a strategy that will prevent big loss, with max drawdown < 20%
: and single trade loss < 10%. That's at least working better than
: diversification for the purpose of diversification.
| g*****g
发帖数: 34805 | 32 I thought diversification was to prevent big loss, not to get
average loss and feel good about it.
【在 K****D 的大作中提到】
: As I said earlier, in your chart everything goes deep down
: (even more than S&P) because the author intentionally or
: non-intentionally left out some sectors (like consumer staples)
: and inserted some sectors that do not belong to US market.
: By definition, if everything in S&P goes deep down, S&P should
: go down at the average rate.
| K****D
发帖数: 30533 | 33 It does beat all in financials in year 2008. That's the purpose of
diversification.
In fact, getting an average loss will almost always prevent you
from completely wiped out.
Maybe in your eyes, -40% is a big loss. But to me, -40% is much
better than -70%.
【在 g*****g 的大作中提到】
: I thought diversification was to prevent big loss, not to get
: average loss and feel good about it.
| h*******y
发帖数: 864 | 34 You don't understand what the BM was asking, do you?
You kept claiming that you have a better liver medicine, but the method you
used is by applying the historical data with a model containing quite a few
variables, it does NOT count. Say, you group all the liver diseases patient
into two groups (one died with 3 years and one lived beyond three years) and
then you study the all the differences: how do they sleep differently, how
do they eat differently, what drugs they take and how much they exe
【在 g*****g 的大作中提到】
: You have a heart disease medicine which can prevent you from dying, but
: still paralysed nontheless. I show you a medicine which is meant for liver
: maybe, but if you use for heart disease, it can prevent paralysis as well.
: Then of course I can claim my liver medicine is a better
: medicine for heart disease patients.
| K****D
发帖数: 30533 | 35 marked for the hard work of typing, hehe.
you
few
patient
and
how
【在 h*******y 的大作中提到】
: You don't understand what the BM was asking, do you?
: You kept claiming that you have a better liver medicine, but the method you
: used is by applying the historical data with a model containing quite a few
: variables, it does NOT count. Say, you group all the liver diseases patient
: into two groups (one died with 3 years and one lived beyond three years) and
: then you study the all the differences: how do they sleep differently, how
: do they eat differently, what drugs they take and how much they exe
| g*****g
发帖数: 34805 | 36 annual return B&H annual
SPX (since 1950) 13-34 10.03% 6.00%
SPX (since 1950) 12-34 10.12%
SPX (since 1950) 14-34 10.16%
DJI (since 1928) 13-34 9.95% 5.49%
WMT (since 1972) 13-34 12.46% 18.24%
13-34 is just a commonly used TA indicator, I didn't fit it. You can
see 12-34 and 14-34 performing better for SPX, but the result is similar.
The idea behind this: 13 weekly EMA is a mid-term
【在 h*******y 的大作中提到】
: You don't understand what the BM was asking, do you?
: You kept claiming that you have a better liver medicine, but the method you
: used is by applying the historical data with a model containing quite a few
: variables, it does NOT count. Say, you group all the liver diseases patient
: into two groups (one died with 3 years and one lived beyond three years) and
: then you study the all the differences: how do they sleep differently, how
: do they eat differently, what drugs they take and how much they exe
| d*****z
发帖数: 114 | 37 I thought you had the sense of humor to interpret what I was trying to imply
with that example. What I should say but didn't say is, if you are in early
2007, the model should look perfectly working, with backtest against
history data having approved. Now you know it doesn't work, but you didn't
know this in 2007. You could hardly see what would happen in future. Few
people of this generation would think housing market could drop more than 5%
YOY. So the strategy was great in 2007 people's eyes.
【在 g*****g 的大作中提到】
: Certainly not, I show a strategy that works through multiple bear markets,
: the strategy can escape early in 30's depression and today's crisis, while
: yours is wiped out today. Also, since you leverage 1:5, you have to consider
: your margin cost, assume you borrow at prime rate, you pay 20%+ of your
: principle for interest, and only get back 10% in raw return? You are
: under water!
: If there's a permanent bull market, then congratulations, you find an
: arbitrage. In the real world, however, you'd back
| d*****z
发帖数: 114 | 38 This writing is great. I have been thinking hardly how to avoid "pattern
fitting" type of bias when doing fundamental analysis.
you
few
patient
and
how
average
out
【在 h*******y 的大作中提到】
: You don't understand what the BM was asking, do you?
: You kept claiming that you have a better liver medicine, but the method you
: used is by applying the historical data with a model containing quite a few
: variables, it does NOT count. Say, you group all the liver diseases patient
: into two groups (one died with 3 years and one lived beyond three years) and
: then you study the all the differences: how do they sleep differently, how
: do they eat differently, what drugs they take and how much they exe
| g*****g
发帖数: 34805 | 39 OK, I got your strategy. So you can get 7% return and borrow 5%.
How about get 10% return as I did and borrow 5% with 5 time leverage?
At the end of the day, I get 10+ 5 * 5% = 35% annually, more than
double your return, and it's working before and after 2007.
imply
early
5%
people
【在 d*****z 的大作中提到】
: I thought you had the sense of humor to interpret what I was trying to imply
: with that example. What I should say but didn't say is, if you are in early
: 2007, the model should look perfectly working, with backtest against
: history data having approved. Now you know it doesn't work, but you didn't
: know this in 2007. You could hardly see what would happen in future. Few
: people of this generation would think housing market could drop more than 5%
: YOY. So the strategy was great in 2007 people's eyes.
| K****D
发帖数: 30533 | 40 Heartinny is an indexer and likely a basher of all methods, hehe.
I am not an indexer but still a basher of all methods, hehe.
I assumed WMT would not work based on the curve shape. Basically
any curve that makes 13EMA and 34EMA curves cross each other very
often won't work. Let me see if I can find an index that's this
case, hehe.
【在 g*****g 的大作中提到】
: annual return B&H annual
: SPX (since 1950) 13-34 10.03% 6.00%
: SPX (since 1950) 12-34 10.12%
: SPX (since 1950) 14-34 10.16%
: DJI (since 1928) 13-34 9.95% 5.49%
: WMT (since 1972) 13-34 12.46% 18.24%
: 13-34 is just a commonly used TA indicator, I didn't fit it. You can
: see 12-34 and 14-34 performing better for SPX, but the result is similar.
: The idea behind this: 13 weekly EMA is a mid-term
| | | g*****g
发帖数: 34805 | 41 Honestly I am not interested in single stock, no strategy will work
for any stock. It's just like applying LEH on any B&H strategy will
be a disaster.
If you can show me a major market index that will work against this
strategy, I'd be more than happy to admit it.
【在 K****D 的大作中提到】
: Heartinny is an indexer and likely a basher of all methods, hehe.
: I am not an indexer but still a basher of all methods, hehe.
: I assumed WMT would not work based on the curve shape. Basically
: any curve that makes 13EMA and 34EMA curves cross each other very
: often won't work. Let me see if I can find an index that's this
: case, hehe.
| K****D
发帖数: 30533 | 42 It's quite hard now. It's much easier before 2007, hehe.
Basically I need an index that's like a upward sin curve, with
the frequency around 1/(20-25 weeks).
【在 g*****g 的大作中提到】
: Honestly I am not interested in single stock, no strategy will work
: for any stock. It's just like applying LEH on any B&H strategy will
: be a disaster.
: If you can show me a major market index that will work against this
: strategy, I'd be more than happy to admit it.
| g*****g
发帖数: 34805 | 43 I'll take any major index & >10 years period to compare.
It's good to know how much you'll lose in worst condition.
【在 K****D 的大作中提到】
: It's quite hard now. It's much easier before 2007, hehe.
: Basically I need an index that's like a upward sin curve, with
: the frequency around 1/(20-25 weeks).
| K****D
发帖数: 30533 | 44 NASDAQ Biotechnology (^NBI) might have a shot.
Again, you are at a huge advantage now since the market just
went through two huge bear markets, hehe. Before 2000, I think
it should be quite easy to grab some indice that can beat your
method, hehe.
【在 K****D 的大作中提到】
: It's quite hard now. It's much easier before 2007, hehe.
: Basically I need an index that's like a upward sin curve, with
: the frequency around 1/(20-25 weeks).
| g*****g
发帖数: 34805 | 45 I think you have a good point, so I backtest SPX from 1950 to 2000/3,
B&H return is 7.14%, 13-34 return is 12.28%.
【在 K****D 的大作中提到】
: NASDAQ Biotechnology (^NBI) might have a shot.
: Again, you are at a huge advantage now since the market just
: went through two huge bear markets, hehe. Before 2000, I think
: it should be quite easy to grab some indice that can beat your
: method, hehe.
| K****D
发帖数: 30533 | 46 How did you compute the max loss% of buying and holding?
As long as it's comparing with the original purchase price,
I am fine with it.
XLV might also have a shot. (I don't know the index ticker
for this.)
Basically health care survived both bear markets and didn't
have huge drop. In the meanwhile in the bull years indexing
owns the EMA method, hehe.
【在 g*****g 的大作中提到】
: I'll take any major index & >10 years period to compare.
: It's good to know how much you'll lose in worst condition.
| K****D
发帖数: 30533 | 47 BTW: how did you compute the max loss% of your EMA method?
I think to be reasonable, it should be the accumulated loss%
since beginning.
e.g., altogether 5 trades, -10%, -10%, -10%, +30%, +30%, then
max loss% should be -27.1%.
If +30%, +30%, -10%, -10%, -10%, then max loss% should be 0%.
【在 g*****g 的大作中提到】
: I'll take any major index & >10 years period to compare.
: It's good to know how much you'll lose in worst condition.
| g*****g
发帖数: 34805 | 48 I'd use max accumulated loss since portofolio peak as
you don't know when you enter, the number is -29.02%
when applying the 13-34 on SPX.
【在 K****D 的大作中提到】
: BTW: how did you compute the max loss% of your EMA method?
: I think to be reasonable, it should be the accumulated loss%
: since beginning.
: e.g., altogether 5 trades, -10%, -10%, -10%, +30%, +30%, then
: max loss% should be -27.1%.
: If +30%, +30%, -10%, -10%, -10%, then max loss% should be 0%.
| K****D
发帖数: 30533 | 49 S&P was too flat during bull years, hehe.
That's why you only had 20 trades during 50 years.
In order to beat your method, there needs to be a lot of trades
(i.e., EMA got confused, no obvious trend).
【在 g*****g 的大作中提到】
: I think you have a good point, so I backtest SPX from 1950 to 2000/3,
: B&H return is 7.14%, 13-34 return is 12.28%.
| K****D
发帖数: 30533 | 50 Sounds right.
Could you try biotech index and health care index? hehe.
I think your method would still win slightly though. Not sure
about the max loss though.
【在 g*****g 的大作中提到】
: I'd use max accumulated loss since portofolio peak as
: you don't know when you enter, the number is -29.02%
: when applying the 13-34 on SPX.
| | | g*****g
发帖数: 34805 | 51 B&H ^NBI beats 13-34 well, 13.5% vs. 4.17%
【在 K****D 的大作中提到】
: NASDAQ Biotechnology (^NBI) might have a shot.
: Again, you are at a huge advantage now since the market just
: went through two huge bear markets, hehe. Before 2000, I think
: it should be quite easy to grab some indice that can beat your
: method, hehe.
| K****D
发帖数: 30533 | 52 I didn't expect the advantage to be so much. Maybe you should
count from the point when EMA13 crosses EMA34 for the first
time?
Anyway, from intuition, if you expect S&P to have a long L
shape recovery, then the EMA method would eventually be
confused and you have to trade in the bottom for many times
before the trend establishes. That may impair your performance.
If the market recovers in V shape, then it's much better.
Many people are predicting an L shape though.
【在 g*****g 的大作中提到】
: B&H ^NBI beats 13-34 well, 13.5% vs. 4.17%
| g*****g
发帖数: 34805 | 53 L shape is not a big deal, it's
L
L shape that will kill bulls and I'd try to avoid at all cost.
And as I said, I'd use more sophisticated method than this to
reduce noise.
【在 K****D 的大作中提到】
: I didn't expect the advantage to be so much. Maybe you should
: count from the point when EMA13 crosses EMA34 for the first
: time?
: Anyway, from intuition, if you expect S&P to have a long L
: shape recovery, then the EMA method would eventually be
: confused and you have to trade in the bottom for many times
: before the trend establishes. That may impair your performance.
: If the market recovers in V shape, then it's much better.
: Many people are predicting an L shape though.
| h*******y
发帖数: 864 | 54 OK. Can you look at SPX from 1950 to 1980 and tell us what the results are?
【在 g*****g 的大作中提到】
: annual return B&H annual
: SPX (since 1950) 13-34 10.03% 6.00%
: SPX (since 1950) 12-34 10.12%
: SPX (since 1950) 14-34 10.16%
: DJI (since 1928) 13-34 9.95% 5.49%
: WMT (since 1972) 13-34 12.46% 18.24%
: 13-34 is just a commonly used TA indicator, I didn't fit it. You can
: see 12-34 and 14-34 performing better for SPX, but the result is similar.
: The idea behind this: 13 weekly EMA is a mid-term
| h*******y
发帖数: 864 | 55 OK. Can you look at SPX from 1950 to 1980 and tell us what the results are?
【在 g*****g 的大作中提到】
: annual return B&H annual
: SPX (since 1950) 13-34 10.03% 6.00%
: SPX (since 1950) 12-34 10.12%
: SPX (since 1950) 14-34 10.16%
: DJI (since 1928) 13-34 9.95% 5.49%
: WMT (since 1972) 13-34 12.46% 18.24%
: 13-34 is just a commonly used TA indicator, I didn't fit it. You can
: see 12-34 and 14-34 performing better for SPX, but the result is similar.
: The idea behind this: 13 weekly EMA is a mid-term
| g*****g
发帖数: 34805 | 56 You are right, first cross occurs after the price is already 300%.
If it starts with first cross, the B&H return is 4.87%, compared to
4.17%.
【在 K****D 的大作中提到】
: I didn't expect the advantage to be so much. Maybe you should
: count from the point when EMA13 crosses EMA34 for the first
: time?
: Anyway, from intuition, if you expect S&P to have a long L
: shape recovery, then the EMA method would eventually be
: confused and you have to trade in the bottom for many times
: before the trend establishes. That may impair your performance.
: If the market recovers in V shape, then it's much better.
: Many people are predicting an L shape though.
| s**********n
发帖数: 868 | 57 This is a great post.
I was also thinking about how goodbug's TA indicators are different from a
pure parameter fitting, but I do believe there are some underlying reasons
behind it, and the parameters he used do have some robustness.
you
few
patient
and
how
average
out
【在 h*******y 的大作中提到】
: You don't understand what the BM was asking, do you?
: You kept claiming that you have a better liver medicine, but the method you
: used is by applying the historical data with a model containing quite a few
: variables, it does NOT count. Say, you group all the liver diseases patient
: into two groups (one died with 3 years and one lived beyond three years) and
: then you study the all the differences: how do they sleep differently, how
: do they eat differently, what drugs they take and how much they exe
| s**********n
发帖数: 868 | 58 From a scientific researcher point of view, I don't buy any empirical
conclusion before I have confidence in its underlying theory.
Market is not a random walk. We all know it's not.
The theory behind buy & hold is human productivity always improves over time
, and the productivity will be reflected in corporate profits over the long
run.
The theory behind your method, as far as I can understand, is the existence
of economic cycles, and an estimate of its time frame:
1. A typical bear market ind
【在 g*****g 的大作中提到】
: annual return B&H annual
: SPX (since 1950) 13-34 10.03% 6.00%
: SPX (since 1950) 12-34 10.12%
: SPX (since 1950) 14-34 10.16%
: DJI (since 1928) 13-34 9.95% 5.49%
: WMT (since 1972) 13-34 12.46% 18.24%
: 13-34 is just a commonly used TA indicator, I didn't fit it. You can
: see 12-34 and 14-34 performing better for SPX, but the result is similar.
: The idea behind this: 13 weekly EMA is a mid-term
| K****D
发帖数: 30533 | 59 So, do you think now the 34-week should be longer or shorter?
the 13-week should be longer or shorter?
On the other hand, we are more confident in theory 1. The chance for thoery
1 to be falsified seems much smaller than theory 2. That's why we have
confidence in buy&hold. The time frame of economic cycles and market
response may or may not be affacted by a lot of factors (other than human
nature-greediness). For example, I may argue, information is spreading
faster, the market becomes globally
【在 s**********n 的大作中提到】
: From a scientific researcher point of view, I don't buy any empirical
: conclusion before I have confidence in its underlying theory.
: Market is not a random walk. We all know it's not.
: The theory behind buy & hold is human productivity always improves over time
: , and the productivity will be reflected in corporate profits over the long
: run.
: The theory behind your method, as far as I can understand, is the existence
: of economic cycles, and an estimate of its time frame:
: 1. A typical bear market ind
| g*****g
发帖数: 34805 | 60 As an engineer, I buy empirical conclusion as long as it shows statistical
significance and robustness. I believe the indicator helps because
1. It shows better return than B&H in all broad market indices I found
2. It shows less max drawdown
3. It shows slightly worse but still positive return than B&H in cyclic
market like retailers and BIO, with lower max drawdown. Exactly what you
try to achieve in diversification.
4. What I didn't mention yesterday, it shows a positive 4% annual return
by g
【在 s**********n 的大作中提到】
: From a scientific researcher point of view, I don't buy any empirical
: conclusion before I have confidence in its underlying theory.
: Market is not a random walk. We all know it's not.
: The theory behind buy & hold is human productivity always improves over time
: , and the productivity will be reflected in corporate profits over the long
: run.
: The theory behind your method, as far as I can understand, is the existence
: of economic cycles, and an estimate of its time frame:
: 1. A typical bear market ind
| | | m******t
发帖数: 2416 | 61
Because
Wait, now I don't buy _that_ either, for the same reason I don't
buy bug's "holy grail" - you are too certain.
If bug has backtested the strategy with historical data that
goes back far enough, then I would believe that empirically it has
some better odds to work in the future. "Better odds" doesn't
mean that it will definitely work, but only that it will more likely
work than a random strategy. And it would become "no better
odds" only when in some future the market shows otherwise.
【在 h*******y 的大作中提到】
: OK. Can you look at SPX from 1950 to 1980 and tell us what the results are?
| g*****g
发帖数: 34805 | 62 There's no holy grail, as one mentioned, if it's China in 1949,
we all wiped out. But I show a strategy that still works if the market
will mirror the history, only on totally opposite way. Moreover,
it will get the same return if you do both long and short. That by
my definition, is more robust and making less assumption.
【在 m******t 的大作中提到】
:
: Because
: Wait, now I don't buy _that_ either, for the same reason I don't
: buy bug's "holy grail" - you are too certain.
: If bug has backtested the strategy with historical data that
: goes back far enough, then I would believe that empirically it has
: some better odds to work in the future. "Better odds" doesn't
: mean that it will definitely work, but only that it will more likely
: work than a random strategy. And it would become "no better
: odds" only when in some future the market shows otherwise.
| p******h
发帖数: 1783 | 63 could u calculate the annual return for doing long in bull and short in bear?
【在 g*****g 的大作中提到】
: There's no holy grail, as one mentioned, if it's China in 1949,
: we all wiped out. But I show a strategy that still works if the market
: will mirror the history, only on totally opposite way. Moreover,
: it will get the same return if you do both long and short. That by
: my definition, is more robust and making less assumption.
| g*****g
发帖数: 34805 | 64 I think long only is 10%, short only is 4% and total is about 12%.
I can double check the number tonight.
bear?
【在 p******h 的大作中提到】
: could u calculate the annual return for doing long in bull and short in bear?
| K****D
发帖数: 30533 | 65 B&H does not assume more than what you assume:
It also assumes in 10-30 years the market will repeat
its history -- yield positive.
Just like your assumption:
In 10-30 years the market will repeat its history -- does not
turn cyclical (basically any shape that involves rapid cycles,
high spike, and low absolute long-term yield).
Actually I believe that there is no guarantee that S&P won't
turn cyclical, provided the current economy.
Overall it works like a MA, if you buy when 200MA is moving up
【在 g*****g 的大作中提到】
: I think long only is 10%, short only is 4% and total is about 12%.
: I can double check the number tonight.
:
: bear?
| g*****g
发帖数: 34805 | 66 If SPX goes all the way from here to 26 points in 50 years, my strategy
still makes money, how about B&H? And when you talk about cyclic, you have
to define the frequency. I can call the last 10 years cyclic too, and my
strategy still beats B&H. I admit every stategy has limitation, but compare
the strategy with B&H
1. The EMA cross strategy will make money if history repeats trending,
no matter up or down. B&H will make money if history repeats trending up,
and make less if that's the case.
2.
【在 K****D 的大作中提到】
: B&H does not assume more than what you assume:
: It also assumes in 10-30 years the market will repeat
: its history -- yield positive.
: Just like your assumption:
: In 10-30 years the market will repeat its history -- does not
: turn cyclical (basically any shape that involves rapid cycles,
: high spike, and low absolute long-term yield).
: Actually I believe that there is no guarantee that S&P won't
: turn cyclical, provided the current economy.
:
| K****D
发帖数: 30533 | 67 If SPX goes to 26, it does not qualify the "cyclical" I mentioned.
Your method fear high VIX pig market, an example would be biotech
during 2003-2007.
I think as long as the frequency is near 13 weeks but far away
from 34 weeks, your method gets defeated. I don't have a tool
to simulate though, that's just my engineer's intuition, hehe.
Basically in that case I would imaging EMA13 to go up and down
very sensitively, while EMA34 keeps comparably flat, thus making
EMA13 cross EMA34 quite often.
【在 g*****g 的大作中提到】
: If SPX goes all the way from here to 26 points in 50 years, my strategy
: still makes money, how about B&H? And when you talk about cyclic, you have
: to define the frequency. I can call the last 10 years cyclic too, and my
: strategy still beats B&H. I admit every stategy has limitation, but compare
: the strategy with B&H
: 1. The EMA cross strategy will make money if history repeats trending,
: no matter up or down. B&H will make money if history repeats trending up,
: and make less if that's the case.
: 2.
| g*****g
发帖数: 34805 | 68 You are right. But
1. Even Retailer and BIO wasn't this cyclic.
2. There is no strategy that will work for any market.
3. No strategy can consistently win in random walk.
4. In real world, I'd add other TA indicators such as
higer high, higher low to help. It may reduce the overall
return in backtesting history data but will reduce false
signal in pig market as well.
If you believe stock market is a probability game as I do.
I'd put my bets on MA cross over B&H.
【在 K****D 的大作中提到】
: If SPX goes to 26, it does not qualify the "cyclical" I mentioned.
: Your method fear high VIX pig market, an example would be biotech
: during 2003-2007.
: I think as long as the frequency is near 13 weeks but far away
: from 34 weeks, your method gets defeated. I don't have a tool
: to simulate though, that's just my engineer's intuition, hehe.
: Basically in that case I would imaging EMA13 to go up and down
: very sensitively, while EMA34 keeps comparably flat, thus making
: EMA13 cross EMA34 quite often.
| K****D
发帖数: 30533 | 69 ft, I admit my maths is bad. If a sin curve has frequency of
1/13 weeks, the EMA13 would be flat, while EMA34 goes up and down
slightly. But your method still won't work well though.
【在 K****D 的大作中提到】
: If SPX goes to 26, it does not qualify the "cyclical" I mentioned.
: Your method fear high VIX pig market, an example would be biotech
: during 2003-2007.
: I think as long as the frequency is near 13 weeks but far away
: from 34 weeks, your method gets defeated. I don't have a tool
: to simulate though, that's just my engineer's intuition, hehe.
: Basically in that case I would imaging EMA13 to go up and down
: very sensitively, while EMA34 keeps comparably flat, thus making
: EMA13 cross EMA34 quite often.
| K****D
发帖数: 30533 | 70 I have confidence that your method would still beat B&H in
the near future. However, its advantage might decrease, due
to the higher-than-before VIX level and lower-than-before
big waves (more likely to be pig market, not roller-coaster).
But nobody knows, maybe the market would go up 100% than go
down 50% again in 3 years. Then EMA kills.
【在 g*****g 的大作中提到】
: You are right. But
: 1. Even Retailer and BIO wasn't this cyclic.
: 2. There is no strategy that will work for any market.
: 3. No strategy can consistently win in random walk.
: 4. In real world, I'd add other TA indicators such as
: higer high, higher low to help. It may reduce the overall
: return in backtesting history data but will reduce false
: signal in pig market as well.
: If you believe stock market is a probability game as I do.
: I'd put my bets on MA cross over B&H.
| | | g*****g
发帖数: 34805 | 71 The chance of wipsaw in bear market is actually lower, as 13EMA
rarely turns. As you can see here.
http://stockcharts.com/def/servlet/Favorites.CServlet?obj=ID2393449&cmd=show[s155827289]&disp=P
【在 K****D 的大作中提到】
: I have confidence that your method would still beat B&H in
: the near future. However, its advantage might decrease, due
: to the higher-than-before VIX level and lower-than-before
: big waves (more likely to be pig market, not roller-coaster).
: But nobody knows, maybe the market would go up 100% than go
: down 50% again in 3 years. Then EMA kills.
| K****D
发帖数: 30533 | 72 I was exactly saying that there will likely be no more bear
market starting from today -- that is, another 30% drop from
today.
Therefore, wipsaw chance is higher. That's just my opinion.
【在 g*****g 的大作中提到】
: The chance of wipsaw in bear market is actually lower, as 13EMA
: rarely turns. As you can see here.
: http://stockcharts.com/def/servlet/Favorites.CServlet?obj=ID2393449&cmd=show[s155827289]&disp=P
| g*****g
发帖数: 34805 | 73 In bull market VIX is low, my friend. Long lasting pig market with
5-10% up or down in turn every 3-6 months is rarely seen in history.
From conspiracy point of view, MM will have hard time to beat retailers
in that kind of market.
【在 K****D 的大作中提到】
: I was exactly saying that there will likely be no more bear
: market starting from today -- that is, another 30% drop from
: today.
: Therefore, wipsaw chance is higher. That's just my opinion.
| s**********n
发帖数: 868 | 74 For EMA to work, you need history to repeat trending, and its trending time
frame should not differ too much from that in history. For B&H to work, you
don't need market pattern. The only bet is we will live in a more advanced
world when you retire, and the market doesn't deviate from the economy for
more than a few decades.
How can you view the stock market purely as a number game? On the time scale
of a few years, it's really the economy that matters.
compare
【在 g*****g 的大作中提到】
: If SPX goes all the way from here to 26 points in 50 years, my strategy
: still makes money, how about B&H? And when you talk about cyclic, you have
: to define the frequency. I can call the last 10 years cyclic too, and my
: strategy still beats B&H. I admit every stategy has limitation, but compare
: the strategy with B&H
: 1. The EMA cross strategy will make money if history repeats trending,
: no matter up or down. B&H will make money if history repeats trending up,
: and make less if that's the case.
: 2.
| K****D
发帖数: 30533 | 75 实际上如果市场接下来1-2个月收盘价水平浮动（哪怕浮动很大，只要
平均收盘价跟现在差不多），你的EMA34就要cross down EMA13了, you
will buy then.
到那时候如果市场继续水平上下浮动（biotech curve), 你就危险了。
这并不是没有可能，至少2009年。
【在 K****D 的大作中提到】
: I was exactly saying that there will likely be no more bear
: market starting from today -- that is, another 30% drop from
: today.
: Therefore, wipsaw chance is higher. That's just my opinion.
| g*****g
发帖数: 34805 | 76 The loss will be small, usually < 10%, it will take years of pig market
to accumulate a big loss. And I can always combine other indicators to
reduce noise.
People see profit in bull market, opportunity in bear market, but only
boredom in pig market. A long lasting pig market is the last thing MM
wants to see.
【在 K****D 的大作中提到】
: 实际上如果市场接下来1-2个月收盘价水平浮动（哪怕浮动很大，只要
: 平均收盘价跟现在差不多），你的EMA34就要cross down EMA13了, you
: will buy then.
: 到那时候如果市场继续水平上下浮动（biotech curve), 你就危险了。
: 这并不是没有可能，至少2009年。
| g*****g
发帖数: 34805 | 77
time
you
The EMA is only assuming the trending time is not too small, e.g. longer
than
6 months. And if market indeed reflects economy, for EMA to not work,
economy
must be up and down like SIN wave every 6 months and something must be
seriously wrong. And for your argument, I am not sure Japanese people will
break even when they retire if they buy around 1990s, 20 years' gone by
already. For any strategy, I'd check 30 or 40 years (use your retire time
here) worst return as bottom line, and B&H
【在 s**********n 的大作中提到】
: For EMA to work, you need history to repeat trending, and its trending time
: frame should not differ too much from that in history. For B&H to work, you
: don't need market pattern. The only bet is we will live in a more advanced
: world when you retire, and the market doesn't deviate from the economy for
: more than a few decades.
: How can you view the stock market purely as a number game? On the time scale
: of a few years, it's really the economy that matters.
:
: compare
| K****D
发帖数: 30533 | 78 Well, since the DOW 10-year return became negative, whether
the world would advance in 10 years has become an assumption
too, hehe.
Some people will retire in 10 years. They need the assumption
to continue to buy and hold.
time
you
scale
【在 s**********n 的大作中提到】
: For EMA to work, you need history to repeat trending, and its trending time
: frame should not differ too much from that in history. For B&H to work, you
: don't need market pattern. The only bet is we will live in a more advanced
: world when you retire, and the market doesn't deviate from the economy for
: more than a few decades.
: How can you view the stock market purely as a number game? On the time scale
: of a few years, it's really the economy that matters.
:
: compare
| K****D
发帖数: 30533 | 79 ?
take 30-year as an example, what's the worst % on DOW
starting any time from 1900-1979?
I don't think it ever went negative ah?
(I thought you computed in another way.)
【在 g*****g 的大作中提到】
:
: time
: you
: The EMA is only assuming the trending time is not too small, e.g. longer
: than
: 6 months. And if market indeed reflects economy, for EMA to not work,
: economy
: must be up and down like SIN wave every 6 months and something must be
: seriously wrong. And for your argument, I am not sure Japanese people will
: break even when they retire if they buy around 1990s, 20 years' gone by
| g*****g
发帖数: 34805 | 80 Depends on what you are comparing to. e.g.
Bond has 40, 60 and 80 years overperforming dji.
It makes more sense to use bond as the bottom line
since you have inflation.
And only take dji and not Nikkei is another way
of making more assumption. How do you know dji will
not be like Nikkei? The last 10 years look like it.
【在 K****D 的大作中提到】
: ?
: take 30-year as an example, what's the worst % on DOW
: starting any time from 1900-1979?
: I don't think it ever went negative ah?
: (I thought you computed in another way.)
| | | K****D
发帖数: 30533 | 81 What about the number in your simulation? negative double-digit?
What's that?
【在 g*****g 的大作中提到】
: Depends on what you are comparing to. e.g.
: Bond has 40, 60 and 80 years overperforming dji.
: It makes more sense to use bond as the bottom line
: since you have inflation.
: And only take dji and not Nikkei is another way
: of making more assumption. How do you know dji will
: not be like Nikkei? The last 10 years look like it.
| g*****g
发帖数: 34805 | 82 Which number?
【在 K****D 的大作中提到】
: What about the number in your simulation? negative double-digit?
: What's that?
| K****D
发帖数: 30533 | 83 Max. system % drawdown
【在 g*****g 的大作中提到】
: Which number?
| g*****g
发帖数: 34805 | 84 It means the max drawdown from your portofolio peak ever.
e.g. If you buy and hold any time before 2007/11 and sell
any time after 2009/3 it will be -57%+ even if you close
at 10000 in the future.
The strategy has about -30% backtesting SPX since 1950, and
it's still too high for my taste. As I said, most investors
here can afford much higher risk than me, I am always scared
on how bad a hit B&H can be, and I learned my lession in a hard
way.
【在 K****D 的大作中提到】
: Max. system % drawdown
| K****D
发帖数: 30533 | 85 So what does this max drawdown value indicate in real life?
For a B&Her, having -60% in the middle doesn't matter much
as long as he/she doesn't retire at that time. Same thing for
EMA'er. Why are you afraid to see -30%?
【在 g*****g 的大作中提到】
: It means the max drawdown from your portofolio peak ever.
: e.g. If you buy and hold any time before 2007/11 and sell
: any time after 2009/3 it will be -57%+ even if you close
: at 10000 in the future.
: The strategy has about -30% backtesting SPX since 1950, and
: it's still too high for my taste. As I said, most investors
: here can afford much higher risk than me, I am always scared
: on how bad a hit B&H can be, and I learned my lession in a hard
: way.
| g*****g
发帖数: 34805 | 86 You may be forced to liquidate and not get back into market
due to any reason. Like retire, like you need the money to
do house downpay. etc. So that's always a risk. Bond, CD, you
can liquidate any time. That's why you want a premium buying
stock and deriative like put can help smooth it out.
It's always important to know how much money you can afford to
lose before you are in the market, not after you already lose it.
【在 K****D 的大作中提到】
: So what does this max drawdown value indicate in real life?
: For a B&Her, having -60% in the middle doesn't matter much
: as long as he/she doesn't retire at that time. Same thing for
: EMA'er. Why are you afraid to see -30%?
| h*******y
发帖数: 864 | 87 Could you calculate the return of your method for SPX between 1950 to 1979
v.s 1980 to 2009 please? I would like to see if it works better in the normal
economy or in the asset bubble economy that we had for the last 30 years.
【在 g*****g 的大作中提到】
: There's no holy grail, as one mentioned, if it's China in 1949,
: we all wiped out. But I show a strategy that still works if the market
: will mirror the history, only on totally opposite way. Moreover,
: it will get the same return if you do both long and short. That by
: my definition, is more robust and making less assumption.
| K****D
发帖数: 30533 | 88 Well, I personally would only buy and hold in retirement accounts.
All other stuffs you mentioned do not have a fixed time frame
unless the risk are taken care in other ways (say the investment
is completely 闲钱).
In reality I think the retirement accounts (not necessarily
401k or IRAs, as long as the purpose is for money after
retirement) are the biggest chunk and buy-and-holders
are talking about that.
If invest for house downpay, I think even die-hard indexer
Ringer won't dare to do B&H.
【在 g*****g 的大作中提到】
: You may be forced to liquidate and not get back into market
: due to any reason. Like retire, like you need the money to
: do house downpay. etc. So that's always a risk. Bond, CD, you
: can liquidate any time. That's why you want a premium buying
: stock and deriative like put can help smooth it out.
: It's always important to know how much money you can afford to
: lose before you are in the market, not after you already lose it.
| s********n
发帖数: 1962 | 89 Max drawdown is a measure of risk. It's the most important measurement
for a mechanical strategy.
One usage of max drawdown is to determine how much you can leverage
up your strategy. If you have a strategy with 1% yield and 2% max drawdown,
you might feel comfortable to leverage it by 10 so you can get 10% yield.
But if you have a strategy with 15% yield but 50% max drawdown, you'd
better think twice if you ever want to use it.
The other usage is to function as a warning/stop signal. If you his
【在 K****D 的大作中提到】
: So what does this max drawdown value indicate in real life?
: For a B&Her, having -60% in the middle doesn't matter much
: as long as he/she doesn't retire at that time. Same thing for
: EMA'er. Why are you afraid to see -30%?
| s********n
发帖数: 1962 | 90 you should expect what your this "fine-tuning" can do.
Your original simple strategy probably fails every 50 years.
After your "fine-tuning", this "better" strategy most likely
will fail every 50 months.
| | | d*****z
发帖数: 114 | 91 But as long as your max draw is >15%, you are wiped out with 1:5 leverage.
Right?
【在 g*****g 的大作中提到】
: OK, I got your strategy. So you can get 7% return and borrow 5%.
: How about get 10% return as I did and borrow 5% with 5 time leverage?
: At the end of the day, I get 10+ 5 * 5% = 35% annually, more than
: double your return, and it's working before and after 2007.
:
: imply
: early
: 5%
: people
| d*****z
发帖数: 114 | 92 I agree with what you said. It looks to me that bug's strategy does have a
not-so-simple implied assumption about how trending pattern works.
time
long
existence
years
【在 s**********n 的大作中提到】
: From a scientific researcher point of view, I don't buy any empirical
: conclusion before I have confidence in its underlying theory.
: Market is not a random walk. We all know it's not.
: The theory behind buy & hold is human productivity always improves over time
: , and the productivity will be reflected in corporate profits over the long
: run.
: The theory behind your method, as far as I can understand, is the existence
: of economic cycles, and an estimate of its time frame:
: 1. A typical bear market ind
| d*****z
发帖数: 114 | 93 The last conclusion should be changed to:
MA strategy assume bull or bear with a trending pattern within a range. B&H
asume bull only when timeframe is long enough, without trending pattern
constrints.
compare
【在 g*****g 的大作中提到】
: If SPX goes all the way from here to 26 points in 50 years, my strategy
: still makes money, how about B&H? And when you talk about cyclic, you have
: to define the frequency. I can call the last 10 years cyclic too, and my
: strategy still beats B&H. I admit every stategy has limitation, but compare
: the strategy with B&H
: 1. The EMA cross strategy will make money if history repeats trending,
: no matter up or down. B&H will make money if history repeats trending up,
: and make less if that's the case.
: 2.
| g*****g
发帖数: 34805 | 94 Not if my ass is covered by Fed, and can borrow from tax payer for
free. :-)
【在 d*****z 的大作中提到】
: But as long as your max draw is >15%, you are wiped out with 1:5 leverage.
: Right?
| g*****g
发帖数: 34805 | 95 15.70% from 1950-1979, 8.31% from 1980-2009. 10.03% combined.
It's worth mentioning the money is in the market for 2/3 of the
time. The majority of this 1/3 is bear market and those can last
multiple years. The strategy can certainly get some return from
bond or CD in those periods.
normal
【在 h*******y 的大作中提到】
: Could you calculate the return of your method for SPX between 1950 to 1979
: v.s 1980 to 2009 please? I would like to see if it works better in the normal
: economy or in the asset bubble economy that we had for the last 30 years.
| p******h
发帖数: 1783 | 96 直观上也很好理解，做大波用长周期，中波中周期，小波短周期
have
【在 K****D 的大作中提到】
: If SPX goes to 26, it does not qualify the "cyclical" I mentioned.
: Your method fear high VIX pig market, an example would be biotech
: during 2003-2007.
: I think as long as the frequency is near 13 weeks but far away
: from 34 weeks, your method gets defeated. I don't have a tool
: to simulate though, that's just my engineer's intuition, hehe.
: Basically in that case I would imaging EMA13 to go up and down
: very sensitively, while EMA34 keeps comparably flat, thus making
: EMA13 cross EMA34 quite often.
| p******h
发帖数: 1783 | 97 看来要确定13，34这几个参数，要作傅立叶变换了
【在 K****D 的大作中提到】
: ft, I admit my maths is bad. If a sin curve has frequency of
: 1/13 weeks, the EMA13 would be flat, while EMA34 goes up and down
: slightly. But your method still won't work well though.
| p******h
发帖数: 1783 | 98 另外，你手头有黄金的数据么？能不能对黄金来个backtest, 牛市做多，熊市做空
以及对DOW/GOLD曲线，SPY/GOLD曲线用一下你的strategy.
【在 g*****g 的大作中提到】
: I think long only is 10%, short only is 4% and total is about 12%.
: I can double check the number tonight.
:
: bear?
| g*****g
发帖数: 34805 | 99 哪里有黄金的数据呀？GLD年头太短。
【在 p******h 的大作中提到】
: 另外，你手头有黄金的数据么？能不能对黄金来个backtest, 牛市做多，熊市做空
: 以及对DOW/GOLD曲线，SPY/GOLD曲线用一下你的strategy.
| p******h
发帖数: 1783 | | | | g*****g
发帖数: 34805 | | p******h
发帖数: 1783 | 102 传个下好的excel 文件
【在 g*****g 的大作中提到】
: 这个太不方便了，还得分30年下。
| g*****g
发帖数: 34805 | 103 按第一次EMA买入为开始
EMA annual return B&H return EMA max drawdown
6.40% 6.22% -49.10%
小beat B&H，但这个max drawdown很厉害。
【在 p******h 的大作中提到】
: 传个下好的excel 文件
| p******h
发帖数: 1783 | 104 good info.
BTW, can u also calculate the curve of DOW/GOLD or SPY/GOLD ? interested to
know the return rate measured by gold standard.
【在 g*****g 的大作中提到】
: 按第一次EMA买入为开始
: EMA annual return B&H return EMA max drawdown
: 6.40% 6.22% -49.10%
: 小beat B&H，但这个max drawdown很厉害。
|
|
