Friday, November 30, 2012

Internet heartbeat, the industrial giant approach

A week ago, I wrote Internet heartbeat, the Canadian way.

Someone at General Electric must have read my blogspot, decided to take a time-machine to travel a few months back and told their colleagues to re-align their corporate strategies.

GE released a white paper this week on its industrial internet vision. The news was subsequently picked up and reported by New York Times and InfoWeek, among others. The piece by C|Net provides the best "at a glance" summary.

Again, we really have to give Sun Microsystems the credits. The current cloud computing movement is essentially Sun's "The network is the computer" vision, just re-branded with a silly name. And this "industrial internet" vision is essentially Sun's "internet heartbeat" vision.

Monday, November 26, 2012

Hemptonology: The art of spotting frauds

If watching your downside is the defining characteristics of value investing; if it's more important to avoid costly mistakes than spotting multi-baggers; then spotting frauds is simply the flip-side of value investing.

If there is anything to learn from last week's HP-Autonomy saga, it is that, regardless whether it is a microcap or a large-cap, anyone investing doesn't do his own due diligence is a fool. (In case you missed the story, here is the abridged version: HP now thinks it was conned in its US$11B acquisition of Autonomy  and has written this investment down by a whoppy $8.8B.)

John Hempton, the hedge fund manager based in Sydney who is short HP, shows how you can spot Autonomy's fraudulent accounts in 5 minutes. As a hedge fund manager, Hempton mainly goes long. He says shorting frauds is only his sideline. But throughout the years, he has dissected many frauds on his blog. If Buffett makes value investing look easy, Hempton makes uncovering frauds look like a child's play. But make no mistake. The look is deceiving. As with any intelligent approaches in investing, "it is simple, but not easy". Behind the scene, a lot of brain power and fact finding are involved. If it were that easy, John Paulson who made his name in betting against subprime mortgages wouldn't have been deceived in the Sino Forest fraud.

Every business is unique. Every fraud situation is also unique in its own way. If you look at how Hempton skillfully deconstructs the meaning of low capex here or high receivables there, you see that almost any single number worth scrutinising. There are infinite variations. So, I don't think it's useful to have just a catalog of fraud detection techniques. Instead, it's more important to internalise the underlying principles, the first principles.

Here are principles I distilled from Hempton's blog:
  • Visualise how the business operates from the numbers - Don't fall into the trap treating the numbers as abstracts. When seeing the gross margins going up every year, it's far too easy to tick the "this business has pricing power" box on your checklist and then move on. What is more important is to visualise how the physical goods flow from here to there and how the money flows in the opposite direction. The more vivid it is, the more powerful it is.
  • Cast a skeptical eye on any outliner - Continue from the previous point, always ask the question: "what does that very favourable metric mean in the context of the company's competitors and its industry? Is that economically or physical possible?"
  • The personalities behind a business are as important as the numbers, if not more important - If you catch 100 burglars in a year, how many of them have only done this once? None? Then, track down the histories of the CEOs, the bankers, the lawyers and the accountants!
  • Data point, data point and more data point - Ben Graham says "you are right only if your facts and reasoning are correct". Without the facts, you only have a hypothesis. Look at what Hempton did. He paid "spies" to visit local factories. He took note how often a founder-CEO flew 12 hours to see his mistress living on a different continent.
  • It's all about Bayesian reasoning - How to put the collected anecdotal evidence together? You have the evidence (E). You have the competing hypotheses (H), "this business is a fraud" vs "this business is brilliant". You apply Bayesian reasoning, reasoning things in Occam's Razor style, to find the hypothesis (or the theory) that fits the evidence the best, going from Pr(E|H) to Pr(H|E).
Yes, all these look so common sense and self-evident on paper. But no, it's not easy, it's not easy to have the mentality and emotional strength to apply these consistently in practice.

I'm writing this post not because I'm an expert in fraud detection or in shorting. Exactly the opposite. I'm a novice. I'm writing this as a reminder to myself.

Tuesday, November 20, 2012

Internet heartbeat, the Canadian way

More than a decade ago, Sun Microsystems' CEO Scott McNealy had the vision that every household or commercial electronic device would have an "internet heartbeat". Things ranging from light bulbs to refrigerators would be networked and speak the same language. Of course, in McNealy's vision, the enabling technology was Sun's Java platform.

McNealy was ahead of his time. Sun has since been swallowed by Oracle.

But the vision lives on, And its form has morphed.

The internet heartbeat beacon is now carried by Android. At the moment, Android is still just the software which powers mobile phones. But it's slowly changing. Android is turning up in unexpected places. If you don't believe me, take a look at this array of (weird) espresso machine and washing machine , this platform for medical devices, this latest and greatest camera made by Samsung, or this Android-powered satellite built by NASA.

Okay, I've stretched my argument a bit too far. There is no cellular network in the space. NASA doesn't put Android into the satellite in order to connect it to the cloud. But my point is, Android is becoming the common denominator for everything electronics. Every electronic device requires a real-time operating system (RTOS), be it a dumb firmware burnt into the circuit or a smart one. If you are a manufacturer and someone is giving away a free RTOS which has baked in wireless cloud-connectivity, is proven to run well on embedded architectures like ARM and offers a whole array of other bells and whistles, it's an offer too good to refuse.

To give McNealy more credit, Android is actually a descendent of Java. (That's why Oracle sued Google for intellectual property infringements.)

However, I've been blind to one thing until I spotted a writeup on SeekingAlpha today. There is another contender to proliferate the internet pulse: Blackberry 10. Or more precisely, QNX.

When I decided to invest in RIMM, my focus was the downside protection. The beauty of this approach is there are more than one way the upside can unfold. High uncertainty, yes. But you don't need to predict precisely which way it goes. And out of all the possible paths, apparently, a sustainable BB10 ecosystem will deliver the true multi-bagger result. To achieve this, RIMM doesn't need to unseat Android or iPhone from the market. It just needs to have enough critical mass. This in turn requires a solid BB10 OS. So, as long as I knew QNX is a solid RTOS with good reputation, I stopped investigate further. (If you don't know what QNX is, it's enough to say that it's been powering cars, medical devices and even nuclear power plants.)

What has escaped my mind is the possibility RIMM has more far-fetched strategic plans for QNX. What will the possibility be if QNX is married to RIMM's secure network on which BBM and BES currently run? You get an end-to-end "smart grid" solution with some interesting mission-critical applications in vertical industries.

Will it be successful? I don't. In a world where everything is converging towards TCP/SSL, do we need a solution based on proprietary network? Again, I don't know. But I won't sweat it. I'm more than happy to have an additional way for the investment thesis to work out.

(Disclosure: Long RIMM, ORCL)

Sunday, November 18, 2012

CNRD 2012Q3 Results - Many things are going right

When I initially invested in CNRD 20 months ago, the investment thesis was a simple one: CNRD, a well-run shipyard located at the Gulf of Mexico, was adversely affected by the Deepwater Horizon oil spill. I reasoned it would only be a matter of time oil/gas exploration in the region would return to normal and CNRD's business would recover. This is a classic case where investment opportunities are created by temporary industry-wide downturns.

Since then, instead of passively waiting for tide to turn, CNRD has proactively replaced its revenues from the energy sector with revenues from other commercial and government clients. Now, from their third-quarter filing, we can see a few tailwinds are in play:
  • Backlog is growing healthily. If we assume the run rate of first 9 months continues, annual earnings will come in at about $3/share. Shipbuilding is a cyclical business. Anecdotal evidence suggests we are now entering the up-cycle in inland water transport. CNRD is now riding this high tide. (For anyone interested in intels in the barge industry related to CNRD's operations, check out the Credit Bubble Stocks blog.)

  • CRND is preparing to submit claims to the BP Settlement Fund. The claim amount is expected to be around $22-23m. Remember that CNRD's market cap is about $110m. So, we are looking at an potential one-off 20% (pre-tax?) boost to its intrinsic value.

  • Oil/gas exploration at the Gulf is recovering but not in full swing yet. While revenues from oil/gass industry has improved to 12% of its total revenues, it's still far below the 27% before the oil spill or 40%+ before the GFC. Gross margin from its repairing and maintenance segment has improved to 15%, but is not yet back to long-term average of 20-25%. When oil/gas operations in the region are back in full force, there will be a continuous supply of repairing/maintenance work. But vessels which have just been moved into the Gulf won't need maintenance immediately. There will be a time lag.

  • The company has repurchased about 2.5% of its shares in 2012 so far.

  • Since Jr Conard took over the CEO role in 2004, this is the first time the company has engaged a financial adviser "to assist in its evaluation of strategic initiatives in order to determine potential alternatives that will enhance shareholder value and provides us with flexibility to respond to potential future business opportunities and risks." This can mean anything. It can also mean nothing. One possibility: Senior Conrad is in his 90's, well past his retirement age. The likelihood that CNRD will put itself up for sale is higher than ever. 
I am patiently waiting for the plots to unfold.

(Disclosure: Long CNRD)