In defense of a quarterly semi-log chart of the DJIA


Back story                    

On Nov. 23, I posted a piece titled, “Happy Thanksgiving – the Dow is going to 4,000.” In the post I provided an overview of the DJIA dating back almost 100 years, highlight a possible top on the semi-long chart.  

On Nov. 26, another blogger and respected author, Michael Harris, was quick to respond with his post titled, ”Dow 4,000? What Technical Analysis Says or Doesn’t Say About Such Possibility.” In his post, Mr. Harris dissected my analysis with a number of critiques.

I get tons of critical comments by email and on the “genius stream of trader consciousness,” otherwise know as the internet. But a critique from another author and blogger grabbed my attention. This became especially true when Josh Brown of TheReformedBroker – an aclaimed trader and blogger whom I have great respect and admiration – applauded Harris’ post.  

This post responds to Harris’ critique of my post on the DJIA. I would like Mr. Harris and Josh to know that this response is not personal, but attempts to address specific items in the critique of my post on DJIA.

My rebuttal to Mr. Harris’ critique

Point 1 – lack of attribution or citation. It is common among scholars or students of a field to provide an attribution or citation to the research of another that is being critiqued. Mr. Harris (and Josh Brown) did not provide a reference to the research analysis (mine) that was critiqued. Josh was focusing on Mr. Harris post, so I can understand his decision not to provide attribution. But as for Mr. Harris, the absence of citation was quite strange and lacks the standard of professionalism that I am certain Mr. Harris has sought in his career.

By way of background, a researcher or analyst normally cites the research he or she is critiquing so that readers can examine the critiqued research for themselves to determine if the critique was fair and balanced. Out of this same respect for professional ethics I am citing Mr. Harris by providing a link to his post in question here. I trust that you, as intelligent and thoughtful students of the markets, can come to your own conclusions on this matter after reviewing my original post, the subsequent critique and now this post.

Point 2 – Mr. Harris did not allude to the many disclaimers I included in my post. In fact, my disclaimers were so clear they might have been written by lawyers responsible for the last 10-seconds of pharmaceutical TV ads. I provided my analysis as a “wild possibility.” Yet, Mr. Harris’ critique (especially because it lacked citation) would imply that my analysis presented a serious possibility. This, in my opinion, is a subtle slight of hand. I am sure this was not Mr. Harris’ intent.

Point 3 – Mr. Harris makes his points, referring to them as “FACTS.” Let’s be honest folks, the real usefulness of technical analysis is to provide a trader with an “edge” by providing an opportunity for an asymmetrical reward/risk profile. Technical analysis of financial markets, at best, is alchemy pretending to be hard science. The label “fact” must be used with great caution when the subject is technical analysis of market prices.

Point 4 – Height of right shoulder in proportion to the left shoulder. Mr. Harris stated that the H&S has already been invalidated because the right shoulder high has exceeded the left shoulder high by 1,000 points. I must admit that this fact has concerned me, and I have stated so, but it does not invalidate the possibility of a H&S top. The Edwards and Magee textbook includes several charts displaying H&S patterns with a right shoulder exceeding the left shoulder. I have seen, and traded, many H&S patterns through the years that have lacked symmetry in shoulder proportion (height and duration).

I will strongly agree with Mr. Harris’ point that chart patterns can provide either a reversal OR continuation function. And I admit that to be too adamant about this possible H&S top is very premature on my part. This is why I specifically stated that a decisive close below the right shoulder low is required to seal the case. This 13-year area of congestion could easily prove to be a continuation pattern. In fact, a decisive close above the 2011 high would confirm a H&S failure with an arithmetic target of near 20,000. If this occurs I hope to be long.

Point 5 – Relative to the discussion of reversal vs. continuation patterns, Mr. Harris makes the following comment:

“I wonder why some analysts see a H&S formation but they do not see a potential inverse H&S formation when looking at the same chart above. The inverse formation has a left shoulder during the fourth quarter of 2002, a head formed during the first quarter of 2009 and a right shoulder under formation.”

If I am correct, the chart below is marked to exhibit Mr. Harris’ interpretation.


I would like to give Mr. Harris every benefit of the doubt on the interpretation he provides above, but the pattern he describes does not qualify as an inverted continuation H&S formation for several reasons, including the lack of an initial left shoulder high, a requirement for a continuation H&S pattern in a bull trend.

Point 6 – Log charts distort chart patterns. Harris states,

“Log charts are very useful in finance, especially in the longer-term analysis of returns and risks of financial assets but when it comes to chart patterns my opinion is that they are not appropriate because the compressions that takes place distort reality.”

Harris acknowledges that log charts are useful in finance, especially for longer-term analysis, and then dismisses stock charts from qualification.

The concept of the log scale is as follows:

  • A stock bought at $50 needs to advance to $100 (or a distance of $50) for a 100% advance.
  • A stock bought at $100 needs to advance to $200 (a distance of $100) for a 100% advance.
  • A stock bought at $200 needs to advance to $400 (a distance of $200) for a 100% advance.

The scaling of log charts reflects these relative relationships. Economists studying long-term trends use log charts all of the time. In fact, it is arithmetic charts, not log charts, that distort reality in long timeframes.

By the way, the DJIA has performed a doubling act more than eight times since the 1932 low. A decline to the 2009 low removes just the last double, leaving seven in place.

I have stacked below the log and semi-log charts of the DJIA over the same time period. The first graph is arithmetic, as proposed by Mr. Harris. The second graph is semi-long, as I displayed in my post. You, as readers, need to decide which of the two charts provides better multi-decade perspective and bests displays the fact the Dow doubled in price five times from the 1932 low to 1984.

Point 7 – It makes no sense to look at long-term charts of the indexes due to the rebalancing of stocks within the indexes. OK, I understand your point, Mr. Harris, and it makes sense in a very elemental way. Yet, I disagree for two reasons. First, at least in the case of the broader indexes, a composite of free market enterprise is represented whether, for example, the transportation companies represented were manufacturers of buck wagons, steam-powered locomotive makers, container ship companies or modern trucking firms. The removal of U.S. Leather Company from and addition of Alcoa to the DJIA in the 1950s was just reflective of the changing nature of the economy. The same argument of rebalancing could extend to ignoring the long-term chart of an individual company simply because the management team has changed or its product lines changed. The Dow is a proxy of a certain slice of the economy, and as such, is subject to the types of mass human buying and selling decisions that can be reflected on a graph.

Second, long-term charts in the Dow have an established history of providing some outstanding trading signals. I will point to just two: the 30-year continuation diamond completed in the mid 1940s and the magnificent 17-year inverted continuation H&S formation completed in 1982.


I have no doubt but that Mr. Harris is a sound market analyst. Yet, on this current subject at hand we disagree. Who is right and who is wrong? I am not sure, because I do not believe technical analysis of stock trends can be adamantly reduced to “fact.”









Members of U.S. Congress skim 6% advantage per year off the stock market from insider trading


Throw the bums out!

That’s what I say. I have been vocal in this blog in being critical toward elected officials, government dumbulators (oh, I mean regulators), the election system, lifetime public skimming (oh, I mean public service), lobbyists that have more power than every last person in the U.S. that has an address on Main Street and the like.

Now we get the insider trading news. So what is really news about members of Congress granting themselves privileges the rest of us don’t have. They have special pension funds, special travel allowances, special health insurance, special everything else.

Those of us who saw 60 Minutes only saw the tip of a very rotten iceberg.

In July, Advanced Trading reported on a study showing how congressmen regularly use information that’s not available to the general public to make profitable stock investments. ( While this could constitute insider trading for an employee of a publically-traded corporation (a criminal offense, by the way), there isn’t any legal barrier stopping lawmakers from doing it.  The study suggests the practice is a popular one on Capitol Hill.

In an analysis of 16,000 stock trades made over a 16-year period,
researchers found that House delegates beat the market by 55 basis points a month or 6-plus percent a year. Just think, your member of Congress could be one of those special MANY skimming 6% off the top. Ever wonder why Congress never wants to limit the access to power of industry and corporate lobbyests?

Following is a special update report posted by this week. After you read this, and dig deeper into this sham activity by our elected officials, perhaps you will do what I did today — Call your members of Congress and demand a straight and immediate up or down vote to correct this misjustice. Demand that the issue be brought up cleanly, not part of some larger bill and not as an amendment. Then, let’s run the bums (and their corporate bribers) out of town.


Is Congress Using Inside Information To Beat the Street?

A recent study claims that members of the U.S. House and Senate have been using non-public information to consistently outperform the market.

Tags: Alan Ziobrowski , Georgia State University, Lindenwood University, Florida Atlantic University, Augusta State University, Congress, Business and Politics,

By Justin Grant

Fund managers continually wrack their brains trying to figure out how to generate alpha. Perhaps they could look to Congress for tips on how to consistently beat the market.

According to a study published in the academic journal Business and Politics, a portfolio mimicking the stock purchases of U.S. House members beat the stock market by 55 basis points a month and 6 percent a year. The research suggests those returns are no accident, with congressmen using their access to non-public information to gain an edge on when to buy stocks.

The study’s authors – Alan Ziobrowski of Georgia State University, James Boyd of Lindenwood University, Ping Cheng of Florida Atlantic Universityand Brigitte Ziobrowski ofAugusta State University- analyzed the returns of more than 16,000 stock trades made by nearly 300 House delegates between 1985 and 2001.

“We find strong evidence that members of the House have some type of non-public information which they use for personal gain,” the researchers wrote. “The nature and source of information is unknown, but clearly further research is warranted.”

The study found, however, that while the congressmen used an informational edge to buy common stocks at the most opportune times, they usually dumped the investments before they peaked, suggesting they lacked an insider’s edge on when to sell. “We basically tracked and compared their returns to those of the market to determine whether their profits were superior in statistically significant fashion to those of the average guy,” Ziobrowski tells Advanced Trading.

“We also recognize that members of Congress have information that may be very pertinent to the performance of common stock that nobody else has.”

A Fine Line Between Illegal and Unethical

The practice differs from criminal insider trading since the lawmakers aren’t corporate insiders making trades on potentially market-moving information from within a public company. Nevertheless, the idea of congressmen profiting from dealings on information that’s unavailable to the voting public is disturbing.

The issue was thrust into the spotlight last fall following a Wall Street Journal investigation, which found that congressional aides also reaped the benefits of such transactions. The newspaper reported that 72 aides from both major political parties traded stocks within industries legislated by their superiors. The ensuing fallout brought renewed life to the Stock Act, a bill that’s languished in Congress since 2006, which would ban legislators and their staffers from making investments based on information unavailable to the public.

The proposed bill also would require legislators and their employees to disclose any stock, bond or commodity futures transaction worth more than $1,000 within 90 days. Under the current guidelines, such deals don’t have to be reported until six months or more after they’ve been made.

“What I do find disturbing is the potential for me pushing legislation that would benefit my portfolio. That’s unethical,” says Georgia State’s Ziobrowski.

“On the other hand, if I find out XYZ Company’s tax status is about to change because they’re going to become a kind of preferred industry and I trade on that knowledge, I don’t have a problem with that because you’re not affecting the markets.”

Senate Supremacy

The practice of lawmakers trading on non-public information apparently has not been limited to members of the House. The researchers published a study in 2004 showing that stock purchases by U.S. senators outperformed the wagers made by their House counterparts.

In an analysis of investments made by senators between 1993 and 1998, the researchers discovered that they beat the market by 85 basis points a month, or 10 percent a year. That report also suggested that senators had better information on when to cash in their shares than their House colleagues, since their stocks often declined in value soon after a sale.

“We hypothesize that power is more diluted in the House of Representatives, which is likely to reduce the informational advantages of House members and result in lower excess returns,” the researchers wrote concerning the disparity between House and Senate investment returns.

Meanwhile, the House study found that democrats earned more than republicans between 1985 and 2001. During that 16-year period, democrats beat the market by 73 basis points a month, or 9 percent a year, compared to 18 basis points a month, or 2 percent a year, for GOP members.

“In the late 1980s the democrats had held the House for many decades,” Ziobrowski points out. “A lot of them were deeply entrenched in their positions, and they may have been able to tweak the government in a more serious fashion.”



Traders: Here is your competition

This is part 2 of a series on the winners of last week’s challenge on risk management. Part 1 was posted November 13 (see here).

This post introduces you to the winners and how they reached their answers.

The winners are Jason Szuminski, Jeff Carroll and Kamal Mokeddem. My intent was to provide the winner with a $50 gift certificate to Amazon. Because of the tie, I will provide each with a $25 gift certificate.


Jason Szuminski

Jason has quite a resume. He is the only MIT graduate ever to be drafted into major league baseball. As you can see from his picture, he is a San Diego Padres fan, having pitched in a nationally televised game in his MLB debut appearance. Jason is an active Air Force reservist with the rank of Captain and went on from MIT to obtain his MBA from Stanford. He is presently an entreprenuer.

Following is how he explains his process of answering the challenge:

 “I estimated the end/initial bankroll to be   A-1.9,  B-3, C-5.8,  D-1.42.  Then I assumed one trade per day, 250 trading days per year, for 5 years.  I took the 1250th root of the bankroll estimates, minus one, to get an expected gain per trade. 

A-.0514%,   B-.0879%,   C-.1407%,   D-.0281%

 This average gain is a function of x=bet size and p=probability of win, where E(Gain)= 2x(p-.5). I used x=3% as a max boundary condition and charted some possible values of x and p for each case.

 Lastly, I judged the order of bet size based on variance in the graphs, with greatest to least being DABC. 

These numbers are completely dependent on the number of iterations, or bets.  Trades per day is roughly an inverse function to the expected gain equation, so increasing to 2 trades per day would only require half the bet size, or half of the win ratio above 50%, to build the same end bankroll.  Overall, to minimize risk, the frequency of bets is just as important as the win ratio. 



































































    • Plan A: Win ratio 51%, bet size  3%
    • Plan B: Win ratio 52%, bet size 2%
    • Plan C: Win ratio 57%, bet size 1%
    • Plan D: Win ratio 50.5%, bet size 3%


Kamal Modeddem

Kamal is a research at a hedge fund. He is proof that getting into MIT takes significant brain power. Kamal as a friend of Jason Szuminksi.

I wrote a quick monte carlo simulation in C++ where I could plug in a win percent and bet size and simulate a portfolio. The program would do 1000 simulations and then average the high, low, and maximum drawdown of all the portfolios. I changed the parameters by hand until it matched the graph’s ending value, high, low, and maximum drawdown percent reasonably well (source code attached). A big unknown was the number of trades represented by the graph, I used 1000 since it is a nice round number close to the number of trading days in 5 years. Of course a different number of trades will give different results, but short of having hard data to do MLE I think this is the best you can do. 

[I understand] one of my good friends, Jason Szuminski, is one of the co-winners. A mutual friend emailed us both your blog post. We used different methods that gave similar results, the biggest difference between our two answers was that he tried to eyeball the variance (hard to do on non log scaled graphs), whereas I used the maximum drawdown percentage as a proxy for variance. My method gives B the lowest variance, while he judged it to be C. 

Editor’s note: Kamal had very little variance between his high and low answers for the win ratio and bet size. The average of his win ratio answers was 52.6%, the average of his bet size answers was 2.2%.


Jeff Carroll

Jeff, is a full-time trader from St. Joseph, IL (focusing on the ES and the EURUSD). He spent 20 years as a software development project manager and in risk management — both career experiences that have translated to his trading career. He has educated quite a few managers and project stakeholders about the differences in simple, complex, and chaotic systems, which applies to pieces to software development and how they affect the ability to predict outcomes.

“Here’s my reasoning:

I felt there wasn’t enough info given about profit targets (and other money management) for the four trading plans to make an exact prediction, so my first guess was that Peter was making a point about how drawdowns and run-ups can vary significantly given identical %win-loss rates. Since a win-loss ratio of 50% with varying profit target and stop loss management could give these results, that seemed the most “shocking” choice. The 5% risk estimate was based both on a guess that the risk was 2, 5, or 10%, and upon a visual guess that the worst series of sequential losses equaled a 40% drop. With a 50% win-loss ratio and 100 trades, 4 losers in a row will happen 97% of the time, 8 losers in a row 17%, and 20 losers far too rarely, so 5% it was. Really, though, I had to make too many assumptions so the reasoning behind the guesses probably matters more than the exact numbers.”

Jeff Carroll
Twitter: @becomingatrader



Jason and Kamal approached the problem primarily as a  statistical exercise. Jeff approached the problem logically through observation and reasoning. Jeff would if he went with his 2.5% risk choice, but I liked the way he concluded I was attempting to make a point that identical metrics can produce different outcomes.

The lesson to us, as traders, is that competition for profits is stiff. Particularly in forex and futures, which are less than zero-sum games, it is somewhat intimdating to know that Jason, Jeff and Kamal represent the new breed of competitors. I understood the challenge because I created it. I understand the risk model used for the challenge because I was its architect. But there is no way I could have solved the challenge had I been a contestant without inside information. I might have guessed that all four graphs represented the same performance metrics because I understand that identical metrics can result in different performances. But I would not have been able to identify the metrics as well as did our co-winners.

Thanks to everyone who particpated in this challenge.






Silver is set up for a major league hammering


Confluence of a major down trend, a 7-week rising wedge and a 3-week symmetrical triangle could take the shine off Silver ($SLV, $SI_F).

The Silver chart is turning into a potential perfect storm. Since Sept. 26 the Silver market has been struggling to correct the decline that took place during a brief two-day period (Sept. 23 through early in the session on Sept. 26).

The entire rally has taken the form of a rising wedge, normally a bearish chart pattern. Note how the upper and lower boundaries are converging. This is a sign the market is reaching a point of termination. A symmetrical triangle has been forming since Oct. 28.

A decisive close below the Nov. 10 low of 3313 would complete the triangle and the wedge. The first target would be the September low. However, if one takes this wedge to be a half-mast pattern, then the decline from the Oct. 28 high should equal the decline from the Sept. 2 high to the Sept. 26 low. This decline was $17.35. Thus, a further target of  $18.35 can be established.

Remember, a decisive close below the Nov. 10 low is required to trigger this sell signal. But also remember that classical charting is not a perfect science. The market could breakout to the downside and fail. The present wedge pattern could morph into any variety of larger patterns. All we have, as chartists, is trigger points that carry a certain reward to risk profile. Trading is about reward and risk and possibilities and probabilities, not about certainties and certainly not about being bullish or bearish.


Disclosure: I am long SLV with a stop below the Nov. 10 low.





And the winners are….


Congratulations to Jeff Carroll, Jason Szuminski and Kamal Mokeddem as the co-winners of the recent risk management challenge.

There are some really smart people in the investment world. I was strikingly reminded of this fact on Thursday when I sat down to review the entries to the risk management challenge I posted on my blog last Monday.

My amazement at how close several people came to the correct answers was vastly surpassed when I inquired of each as to how they reached their conclusions. As I said, there are some very smart people in the investment business. Declaring a single winner is not possible – these three men had the answers completely smothered with very little statistical difference from each other.

This is a two part post. In this first installment I will provide the background of the challenge and its practical importance to trading and present a tabular summary of their answers. In the second installment I will introduce the winners and have them present in their own words (with some slight editing on my part) how they solved the problem.


In the December 7 post (click here) I presented the five-year net asset value graphs of four trading schemes. A total of 226 respondents attempted to provide two data points for each of the four graphs (for a total of eight data points).

  • % win ratio
  • % of capital risked per trade

The four graphs are shown below:

My starting assumption was that a half the respondents would make wild guesses and be wildly wrong, a quarter would spend some time working through alternatives to provide thoughtful responses, but would miss the most important theme of the problem, and the final quarter would approach the problem very analytically – with a small subset of the last quarter taking a shot at replicating the model which produced the graphs — and in the process this small subset would come strikingly close to sticking their landings.

I stated in the original post that the answers would surprise many people – in a real sense the problem contained a trick. I guessed that the small subset mentioned above pick up the scent of that trick in solving the problem.

My assumptions were not too far off. However, my real delight came in discovering the diligent, creative and highly analytical ways used by about a dozen people to attack the problem.


Over the years I have developed some robust computerized risk models (with the help initially from the Carlson Management School at the University of Minnesota and then the Leeds School of Business at the University of Colorado). I can feed into these models three to five years of benchmark variables from existing trading programs or the assumed/retro-determined variables of a system under development. The models, in turn, produce and plot numerous alternative asset curves for any particular set of variables. The variables entered for simulated runs (sims) include:

  • % win ratio
  • Ratio of the avg. $ value of winners vs. losers
  • % of capital risked per trading event
  • Number of trades equating to a certain period of time

The actual models are more complex than the simple framework described above, but we will use the above framework for the sake of this discussion.

I knew the typical respondent to the challenge would be deceived by the fact each equity curve looked quite different – therefore, they would assume the answers for each graph had to be vastly different.

Many novice traders falsely assume that an approach with a 50% win ratio will play itself out in a predictable fashion such as:


If it was only that easy! A trading system with a benchmark win ratio of 50%will experience an even split of total Ws and Ls over an extended number of trades (Ns). But, smaller sequences of Ns can do some very weird things. Further, the exact sequencing of Ws and Ls within the random distribution of outcomes can create havoc to assumptive predictions.

Assume that a win ratio of 50% is run through a random-results generator with N=6,000. In this case the total Ws will equal the total Ls with a very small difference (one or two tenths of a percent). This series of N=6,000 could be run through a random-results simulator all day long with insignificant variations.

Next, assume that ten N=600 random-results generation runs are conducted and then averaged. The average of the ten runs of N=600 would be insignificantly different from the single run of N=6,000.

Next, assume that thirty N=200 random-results generation runs are conducted and averaged. Again, the average of the thirty runs of N=200 would be insignificantly different from the single run of N=6,000. However, and here is where it gets interesting, there would be more variation in the W/L ratio among the thirty runs of N=200 than among the ten runs of N=600.

Next, assume that 100 N=60 random-results generation runs are conducted and averaged. Again, the average of the 100 runs of N=60 would be insignificantly different from the single run of N=6,000. But, the variation in the W/L ratio among the individual 100 runs of N=60 would begin to spread out further on an equally incremented bell curve. There is a mathematical formula to express the variations from the exact 50 win ratio based on the size of the Ns. This formula is unimportant for the scope of this discussion. The fact variations can increase as N sizes decrease is the important point.

Within an N=600 random-generation run there are sub-series with quite different %W ratios. The equity curve can vary significantly from one N=600 random-results generation run to the next even though each random generation run is based on a 50% win architecture. The variation from one equity curve to the next will depend upon the random sequencing of W and Ls and periodic bunching of each.

To illustrate the “bunching” concept, for this article I ran a random generation of the numbers 1 through 100 wherein certain numbers were considered to be Ws and certain numbers were considered to be Ls with a construct of a 50% win ratio. Within the N=1,000 sequencing, there were 12 separate series of five or more consecutive Ls. That equals a minimum of 70 Ls, or 7% of the total N, that appeared as Ls in sequence with four or more other Ls.

These sequences of five or more consecutive Ls were interspersed throughout the N=1,000, although not equal distant. Three of L-series equal or greater than 5 appeared within a sequence of 229 Ns. Such a concentration is a significant outlier, yet significant outliers exist in trading results.

Am I losing you yet?

To create the problem for the blog post I did about 20 random-results generations of N=600 using a 50% win ratio, a bet size of 3.2% of assets and a yet unidentified $W/$L ratio. I selected four of the series that appeared to have distinct asset curve profiles. The real variable, then, became the sequencing of Ws and Ls. Incidentally, the most extended outlier W/L % ratio of the four samples was 51.7/48.3. The other three samples were tighter to the model’s 50/50 ratio.

The 50% win ratio and 3.2% risk were not arbitrarily chosen. Several weeks ago I conducted two completely unscientific polls – one asking respondents for their % win ratio and the other asking respondents for the percentage of capital they risked per trade. I statistically examined the results of those polls to determine that the mean % win ratio was approximately 50% and the mean risk per trade was 3.2%. Of course, there are the issues of self-selection, honesty of self-reporting and sample sizing of the polls, but I needed to start with some data sets.

By the way, a risk of 3.2% is purely crazy idea. This is way too much risk. I believe that a risk aobve 2% is too high. I seldom risk more thant 1% of capital.

As I mentioned, the challenge was rigged – all four equity curves were based on an N=600 model with an identical % win ratio, $W/$L ratio, and % risk per trade.

I was hoping that some of the brightest respondents would pick up on this theme quickly, understanding from their own academic and professional experiences that sequencing or very slight changes in the variables can make an enormous difference in the outcomes; and, that with this theme understood they could come very close to the actual variables used.

The second and final post on this challenge will feature the winners and their methodology.

Practical implications of these results to the average speculator

I am a trader. I have no interest in modeling for modeling sake. I am interested in the practical implications of the modeling process.

I can only imagine that some of you traders (if you have lasted this long into the story) are saying to yourselves, “What the heck does all of this statistical probability stuff have to do with trading….my job as a trader is just to find good trades?”

It has EVERYTHING to do with trading. A trader with an historic benchmark metric of 50% wins may think his next trade will be a winner, but a “think” is not a reality. If a 50% trader actually knew whether the next trade in a series was going to be a winner he would not be a 50% trader.

In reality, a trader with a 50% benchmark (or any other % win metric for that matter) has no idea if his next trade (or brief series of trades) will be a winner or loser. There is great significance to this fact. The significance is that a trading event is nothing more or nothing less than the next data point in a series of data points subject to random distribution and governed my statistical probability. Like it or not, this is truth.

A trader may engage a trading system that truly has a reliable win ratio of 50%, but the trader has no control over the sequencing that will occur. There are too many implications to this concept to unpack here. Yet, even an excellent trading system destined for solid long-term performance can come out of the gate with a sharp drawdown and not recover for up to two years. This is the importance of the concept of sequencing due to random distribution of events subject to statistical probability.

Traders, you may think you have a great way to select winners and sound risk management protocols, but welcome to the world of mathematics.

Congratulations to some very smart men

There were about a dozen or so respondents to the challenge that seemed to have picked up the scent. But there were three men who pursued the trail to its eventual conclusion. They were not spot on, but they were very close, as shown in the table below.

As a general observation, on average and in mean, the winning three were low on % risk and slightly high on % winners.

Jeff, Jason and Kamal and their respective solutions to the challenge will be introduced in the next post.




Applying trade probabilities to reward/risk ratios

On Sunday I posted a chart analysis of Apple INK (not Apple Computer). The chart set up I featured was based on a possible island top. In the post, I laid out a trade with a notional reward to risk profile of about $11 to $1.

Tag teaming off my analysis of $AAPL, my friend and fellow blogger Kid Dyamite posted a wonderful piece today on the importance of applying the probability of outcomes to notional reward/risk ratios. I highly recomend KD’s post, as well as his insightful blog. See the post here. KD’s blog should be on your “regular read” list.

KD’s point, well taken and articulated, is that all reward/risk ratios must be adjusted by the probability of outcomes. I generally do not do this in my blog  posts when I mention reward/risk profiles, but in my trading I adjust these profiles by estimated probabilities. 

Based on the prompting of KD’s post, I would like to revist my Apple INK piece, correctly applying probability of outcomes in order to calculate the Expected Value (EV) of the trade I outlined.

I have traded chart patterns since 1980. I have seen it all. I have a general idea of the probability a particular chart structure will lead to a certain outcome — I use the phrase “general idea” in the most loosely possible way because charting is not an exact science. I reject the data provided by some analysts/authors on the probability that different patterns will “work,” however defined.

So, let’s return to the reward/risk profile for Apple I listed at about $11.0 to $1. Specifically, I cited my profit potential as $50,700 and my risk as ($4,600).

In my opinion, the probability that the island top will not be filled and that the next trend in Apple will carry prices back to $300 is about 12.5% (or 1 in 8). Apple bulls would probably place the odds much lower. To calculate the Expected Value (EV) of the trade we adjust the reward to risk profile as follows:

  • A 12.5% chance of success (.125 x $50,700) = $6,337
  • An 87.5% of failure (.875 x $4,600) = ($4,025)
  • EV = $2,312

Expressed as a ratio, my EV-adjusted reward to risk ratio is 1.6 to 1. While this is a far cry from 11 to 1, it is still very acceptable EV-adjusted odds for a trade.

Thanks, KD, for bringing the matter of probability into the discussion of reward/risk ratios. It is a vitally important component of trading.




How much do you really know about risk management? (First in a series)


For many traders, risk management is about using protective stops.

Unfortunately, the vast overwhelming proportion of traders, especially relative newcomers to market speculation, do not have a clue about the finer points of risk management needed for consistent trading success.

This is the first in a series of posts about risk management intended to display how the conventional wisdom and assumptions on trading performance and risk are dangerously unsophisticated.

Let’s take a specific example.

The four graphs below display the equity value of four trading approaches during a five year period, labeled Plans A, B, C and D . Your assignment, if you dare to undertake the challenge, is to guess the win/loss ratio and the % of total trading capital risked per trade for the four trading plans.




The answers to this quiz will be provided later this week, or when I get to it. The answers will shock you. Until then…

If you are up to the challenge email your answers to

  • Your email’s subject line should be: Blog test Nov. 7, 2011
  • Your answers should be in the following format for each of the four trading plans:
  • Plan A: Win ratio __%, bet size  __%
  • Plan B: Win ratio __%, bet size __%
  • Plan C: Win ratio __%, bet size __%
  • Plan D: Win ratio __%, bet size__%

Deadline is 5 PM MST, Wednesday, Nov. 9. Only one entry per person. Please do not add extra explanation or rationale for your answers. Let’s see how well you understand the risk management component of trading.

Traders who have attended the Factor/EWI Boot Camp are not eligible to enter. You already understand the nature of the challenge and the implications for your trading . [Note: A seat to the December 1-3 Orlando Boot Camp has become available. If you are interested click here.]


Scoring rules: Respondents will be rank ordered as to closest on their answers to each of the 8 questions. The combined total of the rank ordering for each respondent will determine the winner.





Four charts to be watching

There are four charts I have my eyes closely on: Live Piggies, Silver (two graphs) and Corn.


I have discussed this chart structure before. The advance on October 25 completed a 5-week symmetrical triangle bottom pattern and set up the possibility of a massive 6-plus month flag. This interpretation would have an upside projected price of $50 to $55.

There is still the possibility that this recent up thrust is part of a 7-week up flag that is serving as a retest of the massive wedge completed on September 22. This pattern, if completed, would switch me back to a short-side bias. The Silver chart indicating a decline is shown below.

Charts are like living and breathing entitites, constantly evolving. What a chart trader should mainly be looking for are set-ups (long or short) that provide a trigger point and extremely favorable reward to risk profiles.


The December Corn chart displays a clear possible H&S top pattern. Symmetry between the two shoulders of the H&S pattern is most common, but not necessary. The right shoulder does not need to rally further to comply with the requirements for the H&S construction. In fact, the present right shoulder shows some topping behavior.  A close below 625 may indicate that the right shoulder high has been established.

Live Hogs

My favorite chart set up at the current time is in the Hog market. This is a market that I trade very infrequently. The dominant pattern is a rectangle top. The market attempted to break out of the top boundary of this pattern in mid October. The attempt failed. The market now exhibits a possible 5-week H&S top formation. This smaller pattern could launch the completion of the rectangle. The target of the rectangle is just above 70 cents.

The CFTC COT data are also very negative in the Hog market. Commercials have a record short position, funds have a record long position. Liquidation of long positions by MF Global could be the spark that starts this decline.

Futures traders who follow seasonal patterns believe that there is a strong upward bias in Hog prices in the fourth quarter. In actuality, during the past 20 years the price of the February Hog contract has been as likely to go down as go up during the fourth quarter. In fact, some of the largest trends in Hogs during the fourth quarter have been price declines.

$HE_F, $SLV, $SI_F, $ZCZ



Update on MF Global

I have had numerous emails about MF Global from traders who had money with the firm. Below is a press release dated October 31, 2011 from the Commodity Futures Trading Commission. I will make some comments after the press release.


October 31, 2011

CFTC-SEC Statement on MF Global

Washington, DC – The Commodity Futures Trading Commission and Securities and Exchange Commission today made the following joint statement:

“For several days, the SEC, CFTC and other regulators had been closely monitoring developments affecting MF Global, Inc., a jointly registered futures commission merchant and broker-dealer, in anticipation of a transaction that would include the transfer of customer accounts to another firm. Early this morning, MF Global informed the regulators that the transaction had not been agreed to and reported possible deficiencies in customer futures segregated accounts held at the firm. The SEC and CFTC have determined that a SIPC-led bankruptcy proceeding would be the safest and most prudent course of action to protect customer accounts and assets. SIPC announced today that it is initiating the liquidation of MF Global under the Securities Investor Protection Act (SIPA).”


My understanding has always been that segregated funds were protected by government insurance or by the clearing firms of the exchanges. I have no reason to doubt that this is not the case, even if the “segregated funds” had been intermingled with MF Global’s other activities.

The collapse of Refco in 2005 was similar — although Refco collapsed due to financial fraud by the CEO and MF Global’s bankruptcy seems to be related to bad investments. In the case of Refco all holders of segregated accounts were guaranteed and received their funds back fully. However, at the time the forex markets were unregulated and holders of RefcoFX accounts only recovered about 38 cents on the dollar.

I would advise MF Global segregated account holders to monitor the CFTC web site for updates. The CFTC web site is