ilYzyPKPwhICaq

Discussions specific to trading the stock market.
SwingTrader
Contributing Member
Contributing Member
Posts: 6
Joined: Mon Apr 21, 2003 3:46 pm
Location: Russia

Re: ETF's

Post by SwingTrader »

Dave S. wrote:I'm new to this type of trading - can you tell me about ETF's?
Hi Dave,

This thread could be a place to start:

"Why actively trade ETFs?"
http://www.elitetrader.com/vb/printthre ... eadid=8052
Moodaeng
Senior Member
Senior Member
Posts: 34
Joined: Thu Apr 17, 2003 4:23 am

Post by Moodaeng »

I have done some stock research in the past but have never pulled the trigger. There is one website that has very interesting information on the subject.
http://www.fordinv.com/html/rs_specialStudies.htm
I suggest you read the studies, starting with the oldest one.

I have been doing some basic research on applying long term trend following principles to stocks. It seems to me that they can work reasonably well on the small/ midcap universe where one can really pick up on some outsize moves(ie doubling or tripling of stocks) rather than the large cap stocks which tend to have huge retracements and not provide the same opportunity for profit.

Has anyone else done research on this and did they have similar findings?
rs, the april 30 2002 study seems to agree with you.

do we pick each stock from different industry group.
yp, you should find some information also.

Moodaeng
TradingCoach
Roundtable Knight
Roundtable Knight
Posts: 176
Joined: Thu Apr 17, 2003 9:52 am
Location: Sacramento, CA
Contact:

Rydex and Profunds are leveraged

Post by TradingCoach »

and have short funds available. I am doing research on how to quantify and bring closer to the turtle rules the basic fund switching techniques,
if trend following can be used at the stock market it may be with leveraged index funds and perhaps individual stocks - must be a great stock picker, so I pass.....
- These funds used to trend rather well when we had a long term trend in the markets. i.e. before the death of the Bull.
- Sector funds are available to have long/short pairs
- deversification is built in (via sectors)
I don't know how to account for betsize optimization when some funds
only allow unit to be traded as $10,000 or $25,000 - initial account must
be huge to take a bite that big however exit can be adjusted to percent loss (2 percent or 1)
Now the stock markets trend no more I find my focus back in the futures arena but stocks I think have a void in good long term systems.
trendguy
Full Member
Full Member
Posts: 20
Joined: Fri Apr 18, 2003 11:22 pm
Location: Toronto, Canada

Re: Rydex and Profunds are leveraged

Post by trendguy »

Now the stock markets trend no more I find my focus back in the futures arena but stocks I think have a void in good long term systems.
Tradingcoach,

Funny that you wrote that, over the last six weeks I have caught one of my best uptrends in the stock market in the last 3 years ... I now see that different people have different definitions of trend.

thanks for the insight.[/quote]
rs

Index emulation

Post by rs »

Hi all,

I am a little unclear as to why you would end up emulating an index if you trade many of the strongest stocks out there. If you are picking stocks by a long term trend following methodology then you will stay with the strong stocks and sell the weak ones. Thus concentrating your holdings into the strongest stocks in the index. You would not have exposure to the losers so would not emulate the index.

Perhaps I am missing something but that is my perspective.

rs
blueberrycake
Roundtable Knight
Roundtable Knight
Posts: 125
Joined: Mon Apr 21, 2003 11:04 pm
Location: California

Re: Index emulation

Post by blueberrycake »

rs wrote:Hi all,

I am a little unclear as to why you would end up emulating an index if you trade many of the strongest stocks out there. If you are picking stocks by a long term trend following methodology then you will stay with the strong stocks and sell the weak ones. Thus concentrating your holdings into the strongest stocks in the index. You would not have exposure to the losers so would not emulate the index.

Perhaps I am missing something but that is my perspective.

rs
Think of it this way. Let's say every stock movement has two components. Component one is the underlying movement of the market (or at least that market sector) which moves the stock along, while component two is movement specific to the individual stock. Now just assume each component has a 50% weight in the movement of stocks.

If you buy one stock, then the market movement and the individual stock movement has about equal proportion. Now say you buy a second stock in the same market sector. In this case the market component for the two stocks will be exactly the same, and exert upward/downward pressure on the stocks at the same time. However, the individual components are not perfectly correlated, so they will at times cancel each other out. So the net effect is that the market component now has a greater total impact on your portfolio. This process continues as you add more stocks to your portfolio (assuming they are correlated in the broad market sense).

This is obviously a simplification, but you get the idea.

Cheers,
bbc
rs

Post by rs »

I understand the mechanics of portfolio theory and Beta which is basically what you have described. But that applies only to a buy and hold portfolio of diversified stocks. If you trade a long term trend following system you are selling losers and riding winners thus your exposure will be higher to winning stocks and you will have small losses on the weak stocks.

It would not, in my opinion, replicate the index.

rs
stephenm
Full Member
Full Member
Posts: 18
Joined: Sat Apr 19, 2003 1:36 pm
Location: Windsor, UK

Spread Betting and CFDs

Post by stephenm »

I do some trend following on equities with reasonable success using spead betting with contracts for difference. The advantage of these instruments is obviously leverage and the ability to short. It's an accessible and low cost way to trade the equity markets. I'm using CSI end of day charting and the integral charting system to run some moving averages, plus a couple of entry filters. Because there is less volatility in the underlying instruments, it works pretty well on this basis.
Moodaeng
Senior Member
Senior Member
Posts: 34
Joined: Thu Apr 17, 2003 4:23 am

Post by Moodaeng »

Relative Strength or Momentum as Portfolio Selection.

It is a market lore that you should buy "high relative strength" stocks (stocks that have performed best recently).
How has this strategy worked in practice? Take a look at :

http://www.fordinv.com/html/documents/SS0891.pdf

In this article, using data from 12/71 until 09/90, they found that :
- 1 month relative strength is bad : you should sell the stocks that have the highest one month return and buy the stocks that have the lowest one month return.
- 3 months relative strength is bad, also.
- 12 months relative strength is good : you should buy the stocks that have the highest one year return and sell the stocks that have the lowest one year return.

They come up with a formula that optimizes stock selection : you should buy stocks with the highest PRM (price momentum) where PRM is :

PRM = PGY – PGQ – 3 * PGN

With : PGY = 12 month price gain, PGQ = 3 months price gain, PGN = 1 month price gain.


When I first read this article I was suspicious because I suspected curve fitting. But, then, I read this :

http://www.fordinv.com/html/documents/Ss0400_001.pdf

In this article, you see that their momentum model (PMO) has worked remarkably well after their stock selection formula was released
(in 1991)



And I also read this :

http://www.newfinance.com/research/common/commonal.html

In this article, the author finds that a bad one month return and a good twelve month return are very good predictors of future over performance.


Conclusion : there seems to be good historical evidence that :

- short term past underperformance (one month to 3 months) is good for stocks.
- Medium term overperformance (12 months) is good for stocks.

I wanted to know what you thought about these "findings". Have you seen other articles about it ? Is your personal research going the same way? Thanks for sharing your thoughts.
Kiwi
Roundtable Knight
Roundtable Knight
Posts: 513
Joined: Wed Apr 16, 2003 1:18 am
Location: Nowhere near

Post by Kiwi »

It looks awfully like statistical evidence that William O'Neils CANSLIM strategy will work.

CANSLIM (oversimplified) buys stocks that were very strong and have had a cup and handle retracement of some months (retrace, try to breakout, retrace less and then ... ) and then breakout.

John
Forum Mgmnt
Roundtable Knight
Roundtable Knight
Posts: 1842
Joined: Tue Apr 15, 2003 11:02 am
Contact:

Post by Forum Mgmnt »

Our research indicates a 270 day strength filter is very effective for filtering stocks that will be traded using a breakout-based trend-following approach like the Turtle System.

The formula for the one I tested was:

(current close - close 270 days ago) / ATR 270 days ago

This is computed for each stock and then the stocks are ranked and the top 20% to 30% are candidates for trades while the others are excluded.

- Forum Mgmnt
Last edited by Forum Mgmnt on Thu Feb 26, 2004 4:40 pm, edited 1 time in total.
mrscott
Contributing Member
Contributing Member
Posts: 8
Joined: Thu May 22, 2003 9:49 pm
Location: Los Angeles

Re: emulating an index

Post by mrscott »

oops, see below...sorry.
Last edited by mrscott on Tue May 27, 2003 11:30 pm, edited 1 time in total.
mrscott
Contributing Member
Contributing Member
Posts: 8
Joined: Thu May 22, 2003 9:49 pm
Location: Los Angeles

Re: emulating an index

Post by mrscott »

yuri wrote:I have to agree with Gerry, I've been looking into trading stocks mechanically for years. But my experience also dictates that trading many stocks together..begins to emulate an index. :?

Would like to hear other view on this.

Are we talking about stocks that have a low correlation to sp ?
is there an initial criteria that we choose prior to picking the stocks ?

do we pick each stock from different industry group.

I've been looking into this for a while now...but would like to hear other's opinion on preliminary selection for picking stocks to include in the mix ?

Forum Mgmnt, if you can expand on your prior posts regarding correlation of stocks vs. futures..it would be great.

thanks.
I would suggest reading Michael Lauer's interview in Stock Market Wizards, he's done his homework on this subject.
Loggerhead
Contributor
Contributor
Posts: 4
Joined: Mon May 26, 2003 5:25 pm
Location: Houston

Post by Loggerhead »

Forum Mgmnt,

When you refer to the 270 day RS filter for stocks, are you using the actual RS developed by WW? Or are you using a 270 day breakout (RS) as a filter?

thanks,
Loggerhead

MODERATOR'S NOTE: c.f. edited his reply above to be more clear.
lperepol
Senior Member
Senior Member
Posts: 30
Joined: Thu Feb 26, 2004 12:50 pm
Location: Castlegar, BC, Canada
Contact:

Equity Trading Parameters and results

Post by lperepol »

This system uses a channel rule for trend following starting amount is $1,000,000.00
Trading Period(3-Feb-98,24-Feb-04)

Set of Equities (NASDAQ, NYSE, AMEX) all. Selection Random Walk.

Long Side:
Brkout mnStp mxStp lKly wins ttl rtio avePft Cash LakeRatio
284 226 238 0.001 1428 2439 0.585 0.326 1414274.625 0.062

Short Side:
Brkout mnStp mxStp lKly wins ttl rtio avePft Cash LakeRatio
159 113 157 0.001 533 1068 0.499 0.116 1040255.750 0.090

Brkout = N Day Breakout
mnStp = N Day Quick Stop
mxStp = N Day Stop
lKly = Fractional Kelly (not optimal)
wins = Number of winning trades
ttl = Total number of trades
rtio = Wins/TTL
avePft = Average Profit per trade (not very meaningful)
Cash = Cash at the end of period
LakeRatio = Ed Seykota's Lake Ratio.


Ironically it does not do too well on the short side. Volatlity is the major factor.
I have not completed the system. I need to think over some design detials. It is written in C++.
I have some graphs created using R. I can post those if there is further interest -- histograms and charts.

Lawrence
JDV
Contributing Member
Contributing Member
Posts: 5
Joined: Thu Feb 26, 2004 4:07 pm
Location: Dallas, TX

Post by JDV »

Dear lperepol,

You stated that ironically your model didn't work well on the short side.

My day job is as a forensic analyst uncovering shorting opportunities, working for one of the larger short sellers in the market.

I'd just make a couple of observations that might help.

1.) The performance of long stocks seems to behave differently than short positions. Almost all of the money I've made on shorts has come very quickly rather than the stock slowly declining in value. My guess is that this is because when a company warns investors of an earnings short fall, for example, people run for the exits fast and then "figure it out" later. So, you'll get some huge moves very quickly.

This means that the long model shouldn't look like the short model. If I understand what you posted correctly, you used different time periods. But, possibly you need to go further in tweaking the short model to capture the bigger profits that tend to happen more quickly.

2.) Since bull periods seem to be more prevalent than bear periods (using the normal business cycle it would be 3:1), you're shorts using your model probably shouldn't perform nearly as well. So, possibly some sort of filter needs to be added to determine if the market as a whole is more conducive for being long or short and then taking the positions accordingly? For example, William J. O'Neill of Investors Business Daily has stated that in bull periods, 85% of the stocks will trend up (I'm going from recollection here, the actual numbers may be different).

3.) You should short stocks anyway because the risk of being in the market during a down period is considerably greater than being out of the market during a bull period. So, don't be tempted to just be "long only." A lot of professors talk about why market timing doesn't work because if you were out of the market the 5 best days of the year, you wouldn't make any money (or something similar to this). What they don't tell you is that if you're in the market the 5 worst days, your results are much, much worse. So, you should short anyway.

4.) The "forensic" in forensic analyst means that I look at accounting shenanigans to improve the odds of finding good short candidates. If you're fishing in the right part of the lake, you may have a better chance of actually catching something. Typically, this type of analysis yields much higher probabilities of finding companies that announce an earnings shortfall or an SEC investigation. This improves your chances of finding a good short idea. A study has actually been done on this. I would need to dig up the statistics though. I am not sure how you could create this in a model though.

Best,

John D.
Bernd
Roundtable Knight
Roundtable Knight
Posts: 126
Joined: Wed Apr 30, 2003 6:39 am

Post by Bernd »

:wink:
Last edited by Bernd on Sat Apr 19, 2008 6:36 am, edited 1 time in total.
dctag
Roundtable Fellow
Roundtable Fellow
Posts: 50
Joined: Tue Apr 15, 2003 11:55 pm
Location: Salt Lake

Post by dctag »

JDV,

I agree that trading on the short side is helped greatly by looking for troubled companies. I have found that even just going to the "potential" problem companies can help significantly. Like just screening for high debt and downwards earnings drift. Simple fundamental screens can really improve your odds on the downside. Obviously this is a lot lighter then the work that you do but it does up the odds on the short side and are easily screened with most stock screeners. And then you can throw the screened stocks into a technical trading system.

One thing that I have found to be true as well is stocks go down faster then up. I ratchet down my stops a lot faster on the short side then on the long side due to the drastic moves for shorts.

-David
lperepol
Senior Member
Senior Member
Posts: 30
Joined: Thu Feb 26, 2004 12:50 pm
Location: Castlegar, BC, Canada
Contact:

Re: emulating an index

Post by lperepol »

mrscott wrote:
yuri wrote:I have to agree with Gerry, I've been looking into trading stocks mechanically for years. But my experience also dictates that trading many stocks together..begins to emulate an index. :?

Would like to hear other view on this.

Are we talking about stocks that have a low correlation to sp ?
is there an initial criteria that we choose prior to picking the stocks ?

do we pick each stock from different industry group.

I've been looking into this for a while now...but would like to hear other's opinion on preliminary selection for picking stocks to include in the mix ?

Forum Mgmnt, if you can expand on your prior posts regarding correlation of stocks vs. futures..it would be great.

thanks.
I would suggest reading Michael Lauer's interview in Stock Market Wizards, he's done his homework on this subject.

I do not see a strong correlation between ETFs and trend following a large number of stocks on the stock market. :shock:

Here is what I see as correlations for the period of 21-Dec-00 to 9-Feb-04

<TABLE border=1><TR><TD> </TD><TD>TF</TD><TD>DJI</TD><TD>GSPC</TD><TD>IXIC</TD><TD>TMW</TD></TR><TR><TD>TF</TD><TD>1.00000000</TD><TD>0.2793499</TD><TD>0.04953071</TD><TD>0.2083971</TD><TD>0.1887970</TD></TR><TR><TD>DJI</TD><TD>0.27934993</TD><TD>1.0000000</TD><TD>0.95262837</TD><TD>0.8967725</TD><TD>0.9743610</TD></TR><TR><TD>GSPC</TD><TD>0.04953071</TD><TD>0.9526284</TD><TD>1.00000000</TD><TD>0.9378182</TD><TD>0.9885429</TD></TR><TR><TD>IXIC</TD><TD>0.20839709</TD><TD>0.8967725</TD><TD>0.93781822</TD><TD>1.0000000</TD><TD>0.9583489</TD></TR><TR><TD>TMW</TD><TD>0.18879695</TD><TD>0.9743610</TD><TD>0.98854288</TD><TD>0.9583489</TD><TD>1.0000000</TD></TR></TABLE>

TF -> Trend Following
DJI-> Dow Jones
GSPC-> S&P 500
IXIC-> Nasdaq
TMW-> Wilshire 5000 Total

TF is the least correlated of all.
Last edited by lperepol on Thu Mar 04, 2004 8:06 pm, edited 2 times in total.
JDV
Contributing Member
Contributing Member
Posts: 5
Joined: Thu Feb 26, 2004 4:07 pm
Location: Dallas, TX

Post by JDV »

Hi David,

Yes, there are some simple screens that can help you locate troubled companies. And, you're right, it's not nearly the heavy lifting that I do.

But, it also seems like this is difficult to backtest though.

In real-time, I'm sniffing around the Enron type companies. I don't always find them, but I'm usually in the ball park.

However, to backtest this in a model seems difficult because of 20/20 hindsight.

Best,

John D.
Post Reply