Monthly Archives: October 2012

Moneyball of Equity Research – Part 2 – mm…So You Can Beat Mad Money Cramer ?

To get access to my world class team who can address your opportunity areas and network with such luminaries,

– Join my group (use the ‘Join this site’ on the right)
– Leave comments, and questions that might help every one else. 



I started this topic in my previous article under ‘Moneyball of Stocks’ – So You Can Beat Mad Money Cramer.  There was such a response and interesting exchange with my followers and new comers to my world, I thought I will bring the additional 10 companies that are likely to grow 100% in the next two years.  In the second step, I am ranking the list so that we know the top three which have the highest chances of achieving a return of 100% in the next two years. 

As always, in the equity market you have to protect your investments leveraging options or derivative market instruments and there are so many instruments that have sprung up in the last 10 years.  The key thing is identify the growth potential of a company, know when to enter, and know when to exit. 

This section is to show that Moneyball big data approach is very much alive and kicking in equity research.

Let me go back to my previous three companies I brought out in discussion.  I will also tell you whether I have investments in which company to  show that ‘I put my money where my mouth’ is to recall a famous quote.

SuperValu Inc – This $7B enterprise value company has only one tenth as market value – fair value is one where the market value = enterprise value.  Its $36B dollar revenue has many holes where the profitability is seriously leaking. What kind of analytics will help to turn this around 4,400 stores conglomerate?  What kind of questions CAO should be asking and how to find the Moneyball components?
Netflix – Under normal circumstances a company revenue and earnings per share like this will be celebrated like crazy and yet the market has scant regard for Netflix. Why? What kind of Moneyball components matter here?  Is the growth period of this organization over forever?  How can it get its next billion dollar revenue and how it get back to its growth path again?

Panera Bread Co – Many analyst missed the explosive growth of this company in the last 12 years, growing on an average annual growth of 100%. Its PEG ratio is showing 1.5 .  Is the hot growth over with this company? Can analytics open up some new opportunities?  What kind of Moneyball questions the CAO should be asking to take this company for another 100% growth, another $4B revenue?

Here are the 10 stocks that gained minimum of 200% in less than 2 years and potential reduced list for ranking that 3 stocks which have the highest expected probability of appreciating 100% in the next two years.

The key questions are: how to identify these? When to enter and when to exit?

Can BIG data help identify these as they start ramp up? Is the growth over or just started? Is the time right to enter?

First of all, YES, BIG data is the way to go, if you are going to use research.  It just strengthens the traditional equity research.

As a starter, let us look at these graphs (courtesy:  The analysis is not just about analysis of trends, which is going to be only one factor along with multitude of factors I mentioned in Part I. Each stock is at different evolutionary level for a 100% increase in 2 years.  At the same time, remember the famous quote among equity research analysts knee jerk reaction to prediction, ‘history is no indication of future’ but I brave that ‘history and current status are indicators of future’.  The more faster you incorporate the current along with the expected consumer lifestyles and market dynamics of future along with the vibes and signals you receive in the unstructured data world, the better the prediction will be.

Out of these, three are ranked high on the basis of BIG data analysis for 100% return with high probabilities in the next two years. Click on the picture to see the full version of the graph.

The simplified list using is 

From Data Monster & Insight Monster

Underutilized Modeling Opportunity – GLM and Its Variations – Auto, Financing, Insurance, HO Mortgage, Education, Luxury, and Political Contributions Markets

To get access to my world class team who can address your opportunity areas and network with such luminaries,
– Joining my group (use the ‘Join this site’ which is a simple way to appreciate)
– Leave comments, questions, and extensions that might help every one else.



In marketing we often build response models, or brand preference models, or product preference models, or media channel preference models, or shopping channel preference models.  We will call these as a class of preference models or propensity models.

Campaigns are created and fielded on the basis of these models.

There is a serious limitation or positively speaking a serious opportunity to be addressed in these approaches.

The preference models (using a target behavior data – actually acted on or said to have acted on using surveys) fundamentally assumes that all consumers are same except for the probability of those activities.  However, that is far from truth.

The economic value of transaction of people differ even if we assume the simplest case of same preference probabilities.

The CAS(Casualty Acturial Society) professionals address the full range of opportunities that takes in to consideration that there are two parameter variants in the pricing of premium, called pure premium, which takes into consideration both the probability of claim and the amount of claim to readjust premium (pricing) by segments. CAS professionals have been using this concept for almost a century and the method of statistical estimation have become more and more sophisticated in the last 25 years.  The claims analysis involves two key concepts called frequency and severity and generalized linear models affords such qualitatively very different metrics.

In insurance industry the premium models that use data intelligence in the cross-hairs of  claims analysis and loss adjustment expenses in managing the insurance portfolio to file for rate increases and the market competitiveness controls the overall escalation of the rate increases (as long as those rate increases are legally acceptable for social equity).  For a crisp set of slides explaining the methods specifically using the GLM

See: Generalized Linear Model – Statistical Theory and Insurance Examples by Y. Zhang

So the equivalent of that concept for marketing is consumer value ( value at a point in time or life time value).  If one addresses marketing opportunities from this point, a concept of pure marketing value of a consumer similar to pure premium values of an insured, we will benefit by a better micro-marketing customer value concept, that differs not only by the propensity models also by the amount of disposable money. These play well for pricing, better segmentation, and hence better offer and communication.  These are especially valuable for auto financing, home owner mortgage markets, education loan, luxury goods, political contributions markets to name the few other than insurance markets.

The key modeling method that will be used in such circumstances is generalized linear or non-linear models; a broad classes of models are discussed in the following comprehensive collection of slides (this may be skipped by basic readers and concentrate on the above where only linear GLMs are considered)

See: Genearlized Linear and Non-Linear Models for Clustered and Repeated and Longitudinal Data 

This has not been important so far because there is no need of legality of pricing mechanism in pricing of products in general market as it happens in insurance market and differential pricing and/or differential capability modeling based targeting was not considered important as an opportunity area, perhaps.

So far, in marketing it is more about ranking and selection and marketing does not speak about pricing/value concept explicitly and perhaps certain companies do that with certain approaches.  The purpose of this note is to bring out the methods and approaches that are already matured that can be used in marketing also.

With the increased awareness of predictive modeling and competition pushing the envelope for more and more accurate predictions, these methods are getting the attention of executives in all markets.

PROC GLM / REPEATED (balanced data)
PROC GENMOD / REPEATED (balanced or unbalanced)
PROC MIXED (balanced or unbalanced)
PROC GLIMMIX (balanced or unbalnced)

People often build two models separately to attribute expected premium for auto insurance; one is probability of making a claim and the other is estimate of (expected) value of claim.  Use the models to get expected value that an individual will have a certain claim in a year multiply those two and assign that to the individual.

There is a very interesting PROC in SAS called PROC SEVERITY, which does one model for complex non-normal distributions.  PROC SEVERITY handles distributions which is a special class of distributions in generalized linear model where there is positive probability mass at zero support with continuous probability distribution for positive support.  In fact, PROC SEVERITY does more than that.

The quoted is from

  • PROC SEVERITY enables you to fit a distribution model when the severity values are truncated or censored or both. You can specify any combination of the following types of censoring and truncation effects: left-censoring, right-censoring, left-truncation, or right-truncation. This is especially useful in applications with an insurance-type model where a severity (loss) is reported and recorded only if it is greater than the deductible amount (left-truncation) and where a severity value greater than or equal to the policy limit is recorded at the limit (right-censoring)…
  • PROC SEVERITY enables you to define any arbitrary continuous parametric distribution model and to estimate its parameters. You just need to define the key components of the distribution, such as its probability density function (PDF) and cumulative distribution function (CDF), as a set of functions and subroutines written with the FCMP procedure, which is part of Base SAS software. As long as the functions and subroutines follow certain rules, PROC SEVERITY can fit the distribution model defined by them. “”
  •  An example of truncated point mass at zero (or a specific constant) with continuous probability distribution under Tweedie distribution, a special case of exponential family is given in


From Data Monster & Insight Monster

Function of a Chief Analytics Officer

Companies are scrambling to figure out how to use analytics to get the competitive advantage that would give them the extra head start because they could:

  • Understand the winning strategies which have not gotten the attention of their competition, 
  • Figure out the products that are crying to be invented because the market has not fulfilled the product gap
  • Configure mix of products and services that has not been articulated in the board rooms or executive lunches
  • Sell the concept of strategic acquisition of ‘forward developers’ of their own products and services, which they missed seeing them coming – blind-sided (acquisition and mergers)
  • Acquisition and retention of the targeted segment market
  • Be the company’s tug-boat leader in helping the senior management see the opportunities and buy into testing and validating big decisions

The organizations understand this and are willing to go to the next step of hiring process to get that right person. 

The safe bet is go and hire who was already a Chief Analytics Officer (CAO), or some one who was Joint Chief Analytics Officer, or VP of Analytics.  Definitely there is a reason why they are in those positions, especially the exceptional skills in building teams in a traditional way, and that is how many companies will approach.

Are there alternatives? What can organizations supposed to do beyond that safe path but would benefit by additional attributes that would in good likelihood will result in big wins?

Who is the right person? What kind of plan he is supposed to bring?  What kind of questions he should be asking that would be considered smart, forward thinking, and can get to key issues in 90 days; how does he corral the resources to dig out the most important insight?

The best combination is one which complements exceptional managerial team building skills with hands on experience on creation of strategic data intelligences that would fundamentally shift the operational processes and product developments of companies.  This requires some one who had their hands dirty in data in addition to exceptional team building skills.

The person should have command of data, measurements, data sources, understand/ask right questions that will dig for and lead to the right moneyball questions.

What does money ball question mean to that person in that vertical?  After all there is no one single question that will fit in every situation.

Granted the above bullet points are highly strategic in nature.  In deed that is what the Chief Analytics Officers are supposed to do – think about big questions which are highly strategic.

They help organizations identify the next billion dollar opportunity either in revenue or in cost leveraging opportunity.  I bet such opportunities are not going to come because an analyst built a logistic regression or did a machine learning algorithm with out the full picture; it will eventually result to such tactical part of execution.  Building large number of models can be done if the situation warrants.They are the easiest part.

After all, the Moneyball did not build 100s of models.

They built one critical measurement which can be called a strategic measurement and figured out the whole operation.

The CAO position could be unearthing billion dollar opportunities; and so should have the training, background, capability and confidence to talk to CXOs and learn very quickly the opportunity areas and rank them by its dollar values in 90 days.  Depending on the companies and the evolution of their status of data, data intelligence teams, vision, mission, and how big they are, these 90-days plan could be any where from a plan to chart out the key targets to figure out high impact areas to actual implementation plan including some test plans.

These top talents know for every billion dollar or so opportunity,

  • the key conceptual model (algorithm), 
  • how to translate it in a data based form, 
  • what data need to be acquired, 
  • how to build it (with the right team members), 
  • how to interpret its implications that fits into the organization, market, consumer segments, and other operational metrics
  • how to facilitate the organizational changes based on those results for the company wide redefined operations.

Fundamentally, the CAO, should think and boldly ask ‘What are the Moneyball Components Here?’.

Those components are invariably abstracted into the following situations broadly. Which one matters most depends on the vertical and at what stage the company is in terms of their analytical competitiveness to understand and address the inefficiencies and gaps.

– Labor market inefficiency (if it is a labor intensive organization)
– Manufacturing and market delivery disconnect (if the organization relies on movement of goods heavily and hopelessly dangling with various process inefficiencies between production and the retail consumers)
– Product market gaps (if organization is not actively understanding how consumers use their product or listen to the consumer communications)
– Product mix market gaps (if product does not work alone and depends on certain other products or offers to get the best out of the product)
– Blind spots of forward developers growth
– Not knowing the strategic consumer segments (this is really a identity crisis )
– Confusing market strategies where the marketing communications, marketing channels, and marketing segments are not fully aligned (hopelessly disconnect among departments/divisions inside the organization)

Can you identify other items that defines some generic functions of a Chief Analytics Officer? – Thanks

Just for the fun of above understanding, can you identify for the following companies, which one of the above one could be the most likely reasons. I will be quoting few companies from the list of stock companies as there are lot of research material available about these companies.  It is such a fun, I can list 100s of companies that can give ideas on how to pursue with Moneyball approach.

I have no relationship with the CAO of these companies and I do not know them or even whether they have such a position.  My interest in success and failures of companies in general brings up such interesting examples, as quoted below.  Have fun.

SuperValu Inc – This $7B enterprise value company has only one tenth as market value – fair value is one where the market value = enterprise value.  Its $36B dollar revenue has many holes where the profitability is seriously leaking. What kind of analytics will help to turn this around 4,400 stores conglomerate?  What kind of questions CAO should be asking and how to find the Moneyball components?

Netflix – Under normal circumstances a company revenue and earnings per share like this will be celebrated like crazy and yet the market has scant regard for Netflix. Why? What kind of Moneyball components matter here?  Is the growth period of this organization over forever?  How can it get its next billion dollar revenue and how it get back to its growth path again?

Panera Bread Co – Many analyst missed the explosive growth of this company in the last 12 years, growing on an average annual growth of 100%. Its PEG ratio is showing 1.5 .  Is the hot growth over with this company? Can analytics open up some new opportunities?  What kind of Moneyball questions the CAO should be asking to take this company for another 100% growth, another $4B revenue?

Can you point out any company that you think will benefit a very specific Moneyball type opportunity discussion?  What is the company and what do you see are the Moneyball components?

From Data Monster & Insight Monster

BIG data – Moneyball of Stocks, mm.., Can You Really Predict Stock Market!?

So you can beat Mad Money Cramer?
It has happened to me and just to put my thoughts how to go about using analytics for moneyball of stocks, I wrote this piece.

Also, finally I figured this and becoming successful; just the beginning of a great time I want to have with people who are interested in stock market. Show that after all prediction is not that complicated, even in stock market and enjoy the education and make fun money on the way.  I sold apple last week because the management is yet to prove that they are at least 50% as good as Steve Jobs was.  If iPAD mini is not a grand success then the previous statement gets more weight for the credibility of the Sr. Management.

You believe in the analysis of data and the power of moneyball type success. What is not to believe in, you question yourselves?

When it shows in party discussions – in passionate chit chats – how the data intelligence is becoming a defining factor of daily life, the folks (who are lead dogs in the parties) look at you and ask you the question, “So You Are damn Good, mm.., Do You Think You Can Succeed in Stock Market Prediction?” with a twinkle and a squinting for every one else see, that possibly results in a loud or smirky or smug laugh depending on the dress and the cool demeanor you have on that day.

“Listen to what I say, and let me know whether you agree or not”, I talk to them as an analyst.

Do you know the top broad measurement areas you have to analyze to succeed in stock market? It is truly a big data problem.

I was smiling with out opening the mouth at the twinkle and with an up and down shaking of head looking at every one around and say, yeah, here they are, when every one, kind of showing their lips and eyes as if in a puzzle land.

Consumer Lifestyle: Look at the trend of life style changes of targeted consumers by the companies and see whether the trends support the vertical and the company you are investing in 

Management: Look at the management and see whether they are the bunch of people with passion and clarity in product/service innovation and their capability to take their vision to fruition, proven by every little activity, like a great sculpture that comes out a dull rock; a simple ratio of number of business positive measures vs. business negative measures.  These dynamics become lot more complex when a new management change happens.  AAPL management has not proven it has the same tenor as Steve Jobs.

Business Cycle: Are we in boom or bust business cycle? This is a general dynamics of all the factors that have been coming together of over heating of the economy or new trajectories of innovations and up beat of the sentiments due to demographics, not only in USA but also globally; this is becoming all the more important as BRIC countries are becoming powerful consumers. 

Is the Business Acquiring Its Forward Developers:  These are companies, if left alone might grow faster and possibly gobble up the business of the company if you are investing in or at least will fragment the brand of the company that you are investing in or chip away a significant portion of the revenue path that fits the brand of the company that you are investing in.

Trend in revenue, debt, profitability, ROE(on equity), ROA (on assets),, and performance of competitive companies, operating margins, and profit margins

– Trend in P/E,  PEG, Price/Sales, Price/Book

– Crowd intellligence such as and reports/tid-bits from analysts like Cramer; have your eyes and ears open when you visit stores of the vertical in which you have invested

The rate of new innovations that strengthen the products and services of the company you invest in

Consumer Acceptance and Satisfaction Measures Favorable Every Quarter:  The consumer stories that are circulating talking about the products and services – satisfaction measures

Integrity of Products Supported:  The product attributes are elevated to new and better functionality and satisfaction in an integrated way; the only difference is that product differentials defined by its defined functionality –  MAC, IPAD, IPHONE, ITOUCH all have the same quality of functions whatever the functionality are assigned to its platform and increasing product ownership does not cause any additional learning or product failures.  Microsoft was very good in the early growth stages.  This requires a strong visionary who is not only thinking in terms of integration or simplification of the product design but also a fast adapter of technology and life styles of consumers.

– Design beauty and exceptional service for your segment or have proof that your invested company bets against failure:  Check how well the organization is following this.  This is basically positive way of seeing ‘every dog has its price’.  There is a segment of 5-20 Million people who will pay extra for the exceptional design, well defined functionality, and commitment for the promise of delivering the every time the product or service used; the resulting market size depends on the products.  There is a market for FEDEX too and it had its greatest growth times too.

– Macro economic measures: Certain measures are more important to certain verticals vs. others.  This is different from the overall prediction of business cycles for a minimum of 100% in 1-3 years.

Hear the signals in various ways the hedge fund manager behaviors; they are the lead indicators what is going to happen in general in equity market, more specifically in which vertical, and much sharply which stocks.

Note that a good ROE for Warrent Buffet is not just beating S&P but around 16%, I read in an article.

I listen to Cramer but factor that in the above list of measurements. He recommends 100s of stocks and diversification is a great protection in stock market.  I would identify one stock a quarter to identify for 3 to 5 year investment and protect with stock options.  I may not even come out with a choice in a particular quarter and that is ok in my approach.

Often, it is not that difficult; you know it when the trends are getting picked up by the market and more news about the company is heard by ordinary walk of people (as against investment pros).

Ok, smarty can you give examples?

Of course yes:  WMT, MSFT, INTC, GOOG, AAPL, PCLN, AMZN, IBM, NFLX, BIDU, CHPT and all the dow 30 are playing by these.  I can give hundreds of those examples; their crazy growth times and moderate growth times and decimation time.  In the last 12 months, NFLX got decimated, at time when the NASDAQ and DOW were performing very well.  Now it is back on growth track.  The challenge is any time, Hulu Plus can emerge as a serious threat and I am sure, last year wreak would have given enough soul searching to refocus and re-strategize the organizations path.
These stocks as examples, all, had great time growing like crazy and some will continue to do moderately and new ones in the market will sprout. Will AAPL continue to grow?  The management team is not proven yet and they are getting into quality issues – something Steve Jobs will go crazy about.  Even if it does, it does not belong to the case of 300%-500% in 3-5 years group, barring some extraordinary thing to happen for which the immediate credibility, market leadership, and market domination has to occur day in day out.

So why are you not rich ?  – Well, that is a private conversation.

So do you have a ranked list of stocks on the basis of these measurements?

Yes, that is another topic – right now I want to show it is after all looking at the right measurement and now with BIG data, right full integration of BIG data, and not making any emotional judgments!

So you think you can beat Mad Money Cramerica?

Mad Money Cramer has lot of equity market experience with a great following and I love his entertainment approach.  (It was surrealistic when he gave the reason to sell APPLE as fiscal cliff and then today he articulated well the reasons why APPLE needs to prove ahead with its continuous innovation…)

The investment principles and logic are different.  I think the investment approach should follow the big data analysis of metrics that you can collect in the above list of guidelines and invest for long term; 3 years minimum.  Of course, if things change drastically unfavorably in the metric, then you have to move on.  Apple was an example to that effect for now and also it will still handily pass 20% easily per year depending on when you get out and when you get in.

What is the biggest one principle you would advise on money and investing? (or we are stacked against)

Institutional investors are hard to beat not S&P; they can even bend these at least temporarily and perhaps restart of the markets at different levels because some times they interfere too much with innovation and greed.

They invest billions of dollars and you have to figure out a way to enter the market before the hedge fund managers enter and leave the market before the hedge fund managers leave the market.

I believe that BIG data can help; for me that is what BIG data means.  It is not just the volume(pun intended).  It is about variety, velocity.

So what can you say about GOOG? 

It is not in trouble though it dipped almost $100 in the last one month. This is a correction needed for its unproven revenue adjustments for major activity categories that are connected to new consumer lifestyle trends – mobile and eWallet; however, it has a very strong base to get Warrent Buffet ROE; that is not a problem.

Will it double the latest max of around $700 to $1,400 in the next 2 or 3 years?.   The chance is hardly 30%.  But this estimate will change significantly if its android OS and eWallet show enormous success during the holidays.

But the overall trend of earnings in the market place in the last week does not bode well; it is an indication of some general softening spot.

If fiscal cliff is addressed in a meaningful way, then we are on to beyond  Dow 15,000 in the next 4 years! (I also believe it will happen irrespective of who comes in to power, so to say)

See: why apple’s product lines are getting messed up with the new iPAD mini.

Update: 29th Oct, 2012

iOS guru and Retail head leaving Apple – More reasons to sell and hold Apple.

If the stock goes below 550, it is a buy.

Update 8Nov2012

Only yesterday and today, Mad Money Cramer asserted the above points and I see fund managers are getting out of APPLE, and I will move down the entry point (for the fun of owning Apple is $450 to $500, since I have not seen the full impact of the money managers yet) –  There are better stocks than this and I am avoiding APPLE for now.

You may like to read the following:  Moneyball of Equity Research – Part 2 – mm…So You Can Beat Mad Money Cramer ?

From Data Monster & Insight Monster

Architecting Analytically Competitive High Performance Teams and Processes

The premise ‘know your consumer’ is always true in every walk of life, and equally in an important way in building high performance analytics teams, so that one can be relevant to one’s organization and their external customers.

Organizations are at different evolutionary stages in terms of the maturity of their analytics teams, what the team creates, how well defined is the team’s relationship with their client engagement managers, and what is the 1 year, 3 year, and 5 year plans of the organization.

We as creators of high performance analytics teams are coming in only to fit in to the process which is already cranking, and yet take them to higher performing standards.

In terms of measurements the above understanding will be shaped by

– the volume of data to be processed,
– the amount of models to be built and scored
– the amount of campaigns to serve (email campaigns are lot more compared to snail mail campaigns)
– the time latency allowed for analytics to be plugged into in some form of automated processing (with least intervention on the production side), and
– the reports including hard evidence of campaigns with ROI reports

Architecure of an Analytics Platform

Click on the picture which will expand it as a bigger picture making it easily readable.

The systems and software combination does not have to involve a 30 days testing period.  One can make a definitive decision based on the expected demands, regarding R or SAS.  The above discussion is for situations that are not clear.

From Data Monster & Insight Monster

Comedy Central Jon Stewart’s Interview of Nate Silver – Signal from Noise in Our Understanding of Politics

Nate Silver, a brilliant all star statistician who is famous for his Five Thirty Eight, the blog that keeps us more scientifically than our daily news and real times news feeders is with comedy central’s hilarious friend Jon Stewart. 

Interesting thing is Nate knows the politics of information sharing, poker, and our political parties.

How do you take hardly 0.001% of the population and predict in almost real time the usefulness of survey and it WORKS for the whole population.  Data Scientists or Decision Scientists or Statisticians, you have to know this stuff as to why it is successful.

The science of political forecasting and understanding what is signal from the noise.

Nate Silver has published a new book – The Signal and Noise.

The Signal and the Noise: Why So Many Predictions Fail-but Some Don't

I am buying it.  Encourage you to do if you are interested in prediction.

From Data Monster & Insight Monster