Highly Correlated Markets

Below is my list of highly correlated markets for my Diversification Risk-Control Rule. (That rule states that I will not invest more than 25 percent of total assets in the same direction - i.e. long or short - in any group of highly correlated markets.) See more on my risk-control rules on my "How It Works" page.
S&P 500: Dow Jones Industrial Average
Dow Jones industrials: BKX Banks Index, S&P 500
NASDAQ 100: n/a
BKX Bank Index: Dow Jones industrials
Nikkei: n/a
Natural gas: n/a
Gold, copper, silver, platinum, heating oil, crude oil (all highly correlated with each other)

Notes:
1) I defined markets as "highly correlated" when their weekly open prices have greater than 0.85 correlation between March 1995 (the start of most of the combined futures and options Commitments of Traders datasets) and the end of 2007 (the end of my backtest data). Note that I haven't updated these correlations with data since 2007, so they may no longer be a reliable guide. I intend to update them at some later point.
2) Markets like the Nikkei and NASDAQ 100 that have the notation "n/a" aren't highly correlated with any other market. These uncorrelated markets are useful for building diversity into my portfolio.

15 comments:

wenli said...

I am very impressed at your system. Can you show me how you arrive at Comp.Equity Indicator of 0.53 given COTs data as of September 18th, 2007? My calculation does not show the same result. For Dows Jones it shows -2.6% (i.e.41.7%-44.3%); for S&P 500, it shows 6.1% (i.e. 26.5%-20.4%); for Nasdaq it shows -21.5% (i.e. 8%-29.5%); for Russel 2000, it shows -19.1% (i.e. 0.5%-19.6%)

Do you use average of S&P 500, Russel 2000, Dow Jones Industries, and Nasdaq to calculate Comp.Equity Indicator? When these four stock index show different net position, will you net them off? Thanks a lot.

About Alex Roslin said...

Hi Wenli,

Thanks for your message. The indicator isn't based on the net numbers, but rather an average of the standard deviation value of each setup (i.e., the number of standard deviations above or below the moving average I use for that setup), adjusted for the trade delay for each of the setups so that the trade would be effective as of the open of next week's trading.

Regards,
Alex

BillGross said...

Trend Trim Trader Says:

Markets can trend in oversold or bought for months or weeks, a swing in sentiment amoungst commercials does not a change in trend or prices make.
I do not doubt that you will be proved right over time, however, time is the traders enemy and must be respected when time is money.

About Alex Roslin said...

Hi Bill,

Thanks for your comment. You make an excellent point. I raise the same issue elsewhere on this blog. (See my "How I Discovered It" page.) This issue is one of the main reasons why I found my initial examinations of the COTs data weren't fruitful. Traders often trended in one direction for a long time without any price change, even at extreme levels.

One day it dawned on me that I could use the concepts of technical analysis to study the COTs data further and see if there were points in trader positioning that had reliably signaled price trend changes. I hadn't seen anyone else take this approach yet. My initial discoveries gave me the idea of going long or short when traders hit specific extremes of bullishness and bearishness.

Certainly, there are other methods of trading off the COTs, such as adding an indicator based on price or simply adding a time element to the COTs-based trade.

However, I found that trading off the COTs alone using my long/short system produced market-beating, statistically robust results.

Regards,
Alex

About Alex Roslin said...

Hi again Bill,

You're not Bill Gross from Pimco, are you? If so, I'm curious to ask your thoughts on my Treasuries results - namely, that the best results came from fading the commercials or trading on the same side as the large specs and small traders. What do you make of that?

Regards,
Alex

investorjack said...

Alex, I'm new to your COT concept--- interesting. How do you arrive at the number for winning/losing trades? Thanks for sharing your system. Regards, investorjack

SURFThru said...

Alex,

I have been a reader for a while not. For my own understanding I documented your calculations and applied these to a spread sheet. I know you need at least 9 weeks of data to get a signal. I attempted to download historical but it appears the CSV files on the .gov site are only as of the current Friday.

Can you point to where I can get all the history? I am interested in mainly the stock indexes for now.

Thank you,
Patrick
www.SURFThru.com

SURFThru said...

Alex,

I have been a reader for a while not. For my own understanding I documented your calculations and applied these to a spread sheet. I know you need at least 9 weeks of data to get a signal. I attempted to download historical but it appears the CSV files on the .gov site are only as of the current Friday.

Can you point to where I can get all the history? I am interested in mainly the stock indexes for now.

Thank you,
Patrick
www.SURFThru.com

Alex Roslin said...

Hi Anonymous,

Check the "Historic Reports" link for the historic data (see top right links).

Regards,
Alex

AUGUST said...

I love the system and its results and appreciate all the work you put into it. Please tell me if S&P index is highly correlated to the DJIA,how can the latter be bullish and the former bearish on your Monday,12/1 report? Thank you.August

Alex Roslin said...

Hi August,

Thanks for your message. Those setups can give opposing signals because the signals didn't come from the same week so aren't based on the same data. In fact, both signals could be right. It depends on when the exit is.

But I don't have a big enough account to justify betting in contradictory ways and hoping to make money on the difference between the returns, so I created my Correlated Markets Rule.

Regards,
Alex

KP Vidya said...

Dear Alex

I was using your copper setups for ordering of our copper requirements. But now you have dropped it. Not understanding the math involved ( i've just ordered the Demark book, by Perl on your recomm) it is now closed for my use. However i put my life savings (not much, about 14 ounces) in gold based on your report in 2006, when I first got to know of Alex and COT.Absolutely no regrets. Can you please restart the copper part as it is of great use economically speaking. Regards Vidyashankar

Alex Roslin said...

Hi KP,

Thanks for your message. I would strongly advise anyone who reads this blog to do their own homework and not to rely on my signals alone. I have confidence in them solely because of all my own work developing the setups. You don't know what that entailed, so you have to do your own research and/or consult a professional advisor. I would never personally put all my assets into any one signal. Proper risk control requires diversification (see my How It Works page for details of my diversification rule: no more than 25% of assets in any highly correlated group of markets), stops and proper position sizing. The last means I never put more than a certain percentage of total assets in any single setup - based on that setup's robustness.

As you can see from my latest signals table, my gold setup has a maximum portfolio allocation too.

I am presently working on a new copper trading setup, which was almost done when my laptop crashed. I had to spend some time recovering a file I was working on at that time. Should be ready soonish.

Regards,
Alex

rk said...

Hi Alex,

I just got directed to your blog from other one. I was relatively a new trader and had a feeling that commodities are very tricky to trade with EW or other technical tools but your system is doing better in those which is awesome.

Do you look in signals on Silver.

I was looking at the Latest Signals spreadsheet, when does Trade Delay comes into play. Is it for information purposes.

Thanks for the blog.

RK.

Alex Roslin said...

Hi RK,

I'd like to develop a signal for silver eventually. Just haven't had the time so far. The trade delay means how many weeks I wait to execute a trade after getting a new signal from a trader group. So in the SPX case, it's a three-week trade delay for both trader groups, meaning if I get a new signal this coming Friday, Nov. 6, the trade is executed on the open of trading on Monday, Nov. 30.

Regards,
Alex