It’s important that investors understand what they are buying. The biggest gains I’ve made over the years were in companies whose business I understood well. The biggest losses have been in companies that I didn’t understand nearly as well.

During the tech stock bubble from 1997 to 2000, a lot of people – including me – made and then lost money on companies we didn’t really understand. I wasn’t sure exactly what JDS Uniphase (NasdaqGS: JDSU) did, but I did know that the talking heads on CNBC and at MSN Money believed that the sky was the limit.

But I didn’t understand its business. Who were the competitors? What were the threats? I didn’t really know and over time it became very clear to me that I was simply gambling, not investing. I was taking the advice of people who in many cases didn’t understand these businesses themselves, but who had an impressive track record primarily because they had been making their recommendations during a bull market for technology stocks. When the stock price of JDSU started to fall, I was uncertain whether to sell, because I didn’t really understand the long-term prospects.

With that in mind, US refiners have been on a hot streak lately. In the past three months the share prices of Valero Energy (NYSE: VLO), Tesoro (NYSE: TSO), and Marathon Petroleum (NYSE: MPC) (all of which have been recommended here) have appreciated by 61%, 44%, and 52% respectively.

But oil refining has historically been a highly cyclical business. Get in at the right time and you can make a lot of money quickly. Get in at the wrong time and you can lose a lot just as quickly. This is why I caution against buying a refiner if you prefer to buy a stock for the next five years. I would have no major concerns about holding a large integrated oil company like Chevron (NYSE: CVX) for that long, but the refiners require much closer monitoring.

So let’s examine the refining industry in some detail. Oil refiners convert crude oil into finished products such as gasoline, diesel, jet fuel, fuel oil and asphalt. But there can be significant differences among refiners. Recently we have seen refiners in the Midwest and on the Gulf Coast making record profits, while those on the East Coast are going bankrupt. The separator is generally the type of crude oil the refinery can process, which is a function of logistics and equipment.

When a refinery purchases crude oil the key piece of information, besides price, is what the crude oil assay – or composition analysis – looks like. The assay gives valuable information on the types of products that can be produced from the oil, as well as the degree of difficulty in refining it. You have probably heard terms like “light sweet”, or “heavy sour”, but how do these qualifiers affect the ability of a refiner to turn these crudes into products?

Let’s look at a pair of typical crude oil assays:

In reality this is only a portion of a crude oil assay. The full assay would include metals concentration, salt concentration, vapor pressure and so on.

The graphic compares an assay of light crude (the higher the API gravity – a measure of density – the lighter the crude) with one of heavy crude. The light crude is also sweet (i.e., it has low sulfur content) and the heavy crude is sour. A heavy crude can in fact be sweet and a light crude can be sour. But the refiners equipped to handle heavy crudes are generally also equipped to handle sour crudes, so that’s what they buy.

An assay is done by boiling the crude and measuring what’s boiled off at various temperatures. This defines the various products and refining stages, also known as cuts. In the sample assay above, when 99°F has been reached, the gas has been boiled off. This is dissolved methane, ethane, propane, some of the butane and some trace higher gases present in the oil. This cut can be further purified for sales or used as fuel gas to help satisfy a refinery’s need for steam.

The next cut is straight run, or natural gasoline. Gasoline is a mixture of hydrocarbons that are characterized by the boiling point, and the gasoline you purchase at the gas station will contain many different blending components. One of these will usually be light straight run gasoline. This cut will contain compounds like butane, octane and every manner of branched and cyclic hydrocarbon that boils in that specific range.

Most gasoline has been subjected to additional processing. The straight run gasoline is what can be expected to be distilled from the crude oil with no additional processing. Typically, straight run gasoline is not a highly desirable gasoline blending component, because the octane rating is usually low. (Octane rating is a measure of resistance to preignition; higher octane ratings can withstand higher pressures before they ignite.)

The next cut is naphtha. Naphtha can be blended into gasoline, but the octane rating is even worse than for light straight run. Therefore, naphtha is usually fed to a catalytic reformer, which boosts the octane rating from less than 40 to more than 90.

We then come to kerosene (also called “jet”), which starts to get into the range of diesel components. This cut has a higher energy content than the earlier cuts, but the boiling point is too high for it to be blended into gasoline. Kerosene is used as fuel for jet engines and is also blended into diesel. It is also used in some portable heaters and lamps.

The sulfur components start to become more concentrated in these heavier cuts, so they are often subjected to hydrotreating. In this step, hydrogen is used to convert sulfur components into hydrogen sulfide, which is then removed.

The next cut is distillate (specifically, No. 2 Distillate; kerosene is sometimes called No. 1 Distillate). Like kerosene, this cut contains sulfur and must be treated (as with all of the heavier cuts). Distillate has two major end uses: as diesel fuel and as home heating oil. In fact, as seen in the assays above, a substantial portion of a barrel of oil ends up as heavy distillate. For the light sweet crude assay above, 32.4 percent ended up as distillate, and for the heavy sour crude the proportion was 19.3 percent.

We then come to gas oil, which is also known as fuel oil or heavy gas oil (distillate also being known as light gas oil). This cut is typically processed in a catalytic cracker to make cracked gasoline. Cracking involves breaking heavy, long-chain hydrocarbons down into shorter hydrocarbons that boil in the gasoline range. Cracked gas is then blended into the gasoline pool.

The final cut, residuals, or just plain “resid,” is of greatest interest when we talk about the economics of heavy crudes versus light crudes. Note that in the assay above less than 5 percent of the barrel of light crude ends up as resid. Meanwhile, the heavy crude yields over 28 percent resid.

Resid is sold as asphalt and roofing tar and is not a desirable end product from an economic standpoint. So more refiners are installing cokers, which can crack resid into additional gasoline, diesel and gas oils. The economics are usually very attractive given the historical price spread between light oil and heavy oil. A coker can turn over 80 percent of the resid into valuable liquid products.

Heavy sour is cheaper than light sweet for a couple of reasons. The most obvious is that heavy and sour crudes are more difficult to refine, and require more capital investment. Refineries will only make these capital expenditures if the heavy and sour crudes are discounted. But many refineries in the world are not configured to handle heavy and/or sour crudes, which tends to ratchet up the price of light, sweet crude.

The gasoline derived from any type of crude is interchangeable and priced as a single commodity. But thanks to the steep discount, there is more money to be made with heavy sour crudes as long as a refinery is configured to handle them.

Watch That Crack Spread
The crack spread refers to a refinery’s profit margin on the fuels it produces. It is, in other words, a gauge of profitability.

Let’s compare two hypothetical refineries.

Refinery A has no coker and is thus restricted to either buying light crude or buying heavy crude and selling a lot of low-value asphalt and roofing tar. Refinery A pays $100 a barrel for the generic light, sweet crude illustrated by the graphic. It will convert that barrel into 0.909 barrels of liquid fuel product, which let’s say has a value of $120/bbl (in reality, the volume would be slightly greater due to something called refinery processing gain; simply stated the volume “swells” as it is processed). Per the light assay above, 4.4 percent ends up as gas and 4.7 percent as resid.

Refinery A has therefore grossed $120*0.909 – $100, or $9.08 a barrel before we consider the value of the asphalt and the gases. The price of asphalt has risen sharply in recent years as a result of higher oil prices, and also because more refiners are turning their asphalt into higher value products, reducing the asphalt supply. Let’s say that our refiner receives $0.25/lb for its asphalt. A barrel of crude weighs around 300 lbs, so with a 4.7 percent asphalt yield, the barrel yielded 300*0.047 = 14.1 lbs of asphalt worth $3.53. Let’s value our gases at the value of propane (about $0.20/lb on the spot market), and we get a value of 300*.044*$0.20 = $2.64 for the propane. Our gross profit, or crack spread (before operating costs, taxes, etc. are considered) is then $9.08 + $3.53 + $2.64, or $15.25 per barrel for the light crude.

Now consider Refinery B. Instead of buying light, sweet at $100/bbl, it can obtain a heavy Canadian crude for $70/bbl (the current discount of heavy Canadian crudes to West Texas Intermediate is actually about $40/bbl.) Again, its barrel of oil weighs some 300 lbs, and as we can see from the assay above the resid yield may be in the range of 28 percent. So, of the 300 lbs, 84 lbs ends up as resid. But with a coker, the refinery can convert 80 percent of that resid into high-value products, and only 20 percent (16.8 lbs) ends up as low-value coke (a coal substitute).

Therefore, the overall yield from the heavy crude amounts to the sum of the cuts up to resid (71.6 percent), plus the resid that was turned into products (80 percent of 28 percent, or 22.4 percent), minus the gas cut (3.4 percent), for a total of 90.6 percent. The overall liquid yield is almost the same as for the light crude, and yet the refinery paid much less for its heavy.

So, the economics for Refinery B look like this: For the liquid fuels, it grossed $120*0.906 – $70 = $38.72 a barrel. This is four times Refinery A’s liquid fuel profit from the light crude, but Refinery B has slightly less propane yield than the previous example. The value of propane is $2.04. Finally, Refinery B ends up with 16.8 lbs of coke, which is worth only about $0.03/lb (about $0.50 total). The total gross profit then is $48.72 + $2.04 + $0.50 = $51.26.

It should be no surprise that refiners are rushing to install cokers in order to be able to process the discounted crudes. As light sweet supplies deplete, refiners will increasingly turn to heavy sour crude. This also explains why refiners are willing to bring in crude by rail, even though it is more expensive than via pipeline. With discounts this steep on the heavy crudes, the processing economics overwhelm the cost of transportation.

Of course the catch is that a coker is a major capital expense (hundreds of millions of dollars), and it is only part of the equation. I have focused here only on processing heavy crudes, but not on sour crudes. The story is similar to that for the heavy crudes. Sour crudes trade at a discount to sweet crudes, and the refiners need additional processing equipment to handle them.

But the economics currently favor installing the cokers and hydrotreaters to handle the heavy sour crudes, and will continue to do so as long as they trade at a substantial discount to light sweet crudes.

As an investor, you want to own shares in refiners that are configured to run the heavy, sour crudes – as long as these crudes continue to trade at a significant discount to Brent crude. If that discount starts to meaningfully shrink, that will serve as a warning signal for investors in refineries.