And its fundamental concept is so simple: You just need to make really good guesses of the future cashflows and discount it back to the present and what you get is the intrinsic value.

Assuming this public company is valued at $1 billion but based on your inside knowledge of the company's projected cashflows, you derive an intrinsic present value of $5 billion - it's a screaming buy.

Later, as time goes by and the company meets your previous cashflow projections, the market will adjust their valuation to your initial calculation and voila, you're in the money.

Other models like P/E and dividend don't work well. Earnings and dividends can be manipulated SO EASILY. Imagine some dying company borrowing lots of cash in order to increase their dividend. Based on the dividend model, its valuation increases.

So the only thing you can back your life on is the cashflow. You can't just manufacture cash.

DCF can help you explain several phenomenons. Eg. why doesn't Salesforce crash despite its high P/E? Because the bulk of Salesforce customers pay upfront. So there's a lot of cash coming in and that cash has value.

So here's a fun exercise for you to do today: Project Facebook's cashflow for the next 10 years and discount it back to the present and compare it with Microsoft's $15 billion valuation. Then you can tell people whether Microsoft overpayed.

Happy DCFing!

Yes, you can. Read up on Enron's repo agreements. They raised cash by selling assets near the end of the quarter, and promising to buy them back at the start of the quarter -- it was underhanded, but it still showed up as operating cash flow. They did this for t-bills, Nigerian barges, and everything in between. Cash flow quality is higher than earnings quality, but it is by no means perfect.

They raised cash by selling assets near the end of the quarter, and promising to buy them back at the start of the quarter

So there is an initial cash inflow at the end of the quarter and a cash outflow at the start of the next quarter. Just add that in your DCF analysis.

When I say you can't just manufacture cash, I'm talking about free money with no strings attached. In your Enron example, there is a string attached - they have to buy it back in the future, so there is a projected cash outflow in the future. No company can manipulate the books to inject $100 million from the thin air. That $100 million has to come from somewhere - from debt, equity investment, asset sales etc.

Earnings is so unreliable. Eg. this company is projected to earn $1 million with annual growth of 10% from Year 1 to Year 10. But they have a major debt ($100 billion) that is due on Year 11. A DCF model would value this company correctly as bankrupt while an earnings model would not be able to value this company correctly.

But you don't see it! All they have to do -- all they did -- was repo a little more at the end of the next quarter. RJR did this with cigarettes a while back: they offered discounts and 100% refunds to customers who bought just before the end of the quarter, and they ended up with extra cash for their 10-Q even though these actions hurt the business.

A few other scenarios a cash flow analysis misleads you on:

* Gradual liquidation: if a company sells off its inventory and equipment over time, it can show positive, growing cash flow (and a rapidly growing cash flow as a proportion of capital!) even though the business is falling apart.

* A company with a few long-lived assets, like an airline with just a couple planes, will show low earnings and high cash flow when it's not replacing a plane. This is one of those situations in which high cash flow is deceptive and earnings are not.

this company is projected to earn $1 million with annual growth of 10% from Year 1 to Year 10. But they have a major debt ($100 billion) that is due on Year 11

If the debt is normal debt, that shows up on the balance sheet, and the interest payments. If it's a zero-coupon bond, it shows up in the income statement but not the cash flow statement (although the tax effect shows up in both).

The point of earnings is to show what kind of value has been added to the business over time. We might call this Platonic version Earnings. Cash flow is a better way to approximate Earnings when companies are manipulating their earnings, but not when they're manipulating both. In the long run, given a perfect accounting system, DCF will equal Tangible Book + Discounted Earnings. Since accounting is imperfect, and a business can be manipulated to appear better than it really is by any measure that people actually use to value companies, picking just one kind of valuation is folly. I use free cash flow, but I use it judiciously -- I try to make it as much like Earnings as possible.

DCF deals with this. They get upfront cash but since these actions hurt the business, it hurts cashflow in the future. So it's not like as if they got off scot-free with this strategy. Like I said, DCF requires that you have really good estimates about future cashflows. If you say But you don't see it!, then you are not making good estimates.

Note: My point is not that DCF will give you a 99.99% accurate valuation. No model will since all models depends on the accuracy of the inputs. My point is that given the amount of knowledge you have about a particular company, the DCF model provides the best framework to reaching a fair valuation.

A company with a few long-lived assets, like an airline with just a couple planes, will show low earnings and high cash flow when it's not replacing a plane. This is one of those situations in which high cash flow is deceptive and earnings are not.

It is not deceptive. If the company does that, you just assume their future cashflow from operations will decline from the aging planes.

While other valuation models are useful in supporting your initial valuation calculations, DCF is the fundamental block.

Therefore, the price is what someone is willing to pay in hopes that when they but it now there will be someone willing to pay more in the future.

In any case, a company will be worth more if it doesn't pay out dividends since there is no cash outflow from the company.

With a DCF model, you are evaluating the cash inflow and outflow from the company perspective, not the shareholder perspective.

There is nothing concrete to which you can actually tie the current value.

There is - the cashflows of the company.

Imagine a company with net cashflow of -10 million for the first 5 years and +10 billion thereafter. You just discount back the cashflow and you'll get the intrinsic value of the company.

The idea of DCF is to split up the value of an investment into chunks of future cash flow, and ask yourself how much you'd pay for each chunk. For a simple example, if you have a business that's sure to pay you $100, once, a year from now, you ask yourself: how much would I have to put in the bank, now, to get $100 on that date? That amount is the present value of that cash flow.

Now consider a business that will pay that same certain $100, but after two years. The principle is the same -- how much would you put into a bank account now to get that same $100 on the same date?

To further complicate things, imagine that you're betting on a coin flip: two years from now, you will get either $100 (heads) or $0 (tails). To figure out the net present value, you'd first determine the average outcome ($50), then decide how much you'd have to put in a risk-free account now to get that amount in two years.*

Put these together, and you can understand the DCF framework. Let's say you have a business that earned $100 last year, and that you expect to earn about 5% more each year thereafter. But in any given year, there's a 10% chance that the business will go under. The discounted future value is that same procedure, repeated for each year: the price you should pay is how much you'd invest in a bank account now for a 90% chance of $105 in a year, plus an 81% chance of $110.25 in two years, etc., or sum(100 * 1.05^n * .9^n * [1 - risk-free interest rate]^n).

So now all you have to do is 1) figure out what the business will earn every year from now until the end of time, and 2) figure out the intrinsic value of a given sum of money to be delivered at a given future date. These are both, of course, impossible. But rough estimates get you pretty close to where you need to be, and it provides a good way to compare two stable-growth businesses in the same industry (how much should you pay for a soft drink company growing at 3% each year, versus an otherwise identical company growing at 5% each year, for example?).

* This assumes you have an infinite tolerance for risk. But a one in a billion chance of one billion dollars is probably not worth a dollar -- or, rather, it's worth more than a dollar if you enjoy gambling, and less than a dollar if you intend to retire on it.

Edit: replaced a second '$105' with the correct number, $110.25.

The important thing is whether a certain number maps to a mechanism. The mapping is the thing people are really copyrighting when they copyright IP, and that is not a number.

(Or maybe I could write it and someone else would be willing to upload it onto the store?)

We're (I have 4 kids) moving to the other side of the continent in a few weeks, so things are likely to be on hold until we're done.

I have no connections to Apple or any Apple development companies. So it looks like they are finally allowing normal devs in.

I wonder if someone could make an auction site for that? Or is that explicitly forbidden?

I signed up for the developer program yesterday. Any idea how long it takes to get accepted?

No Apple conspiracy I guess :-(