The best of a bad bunch is not necessarily good. Think about that when considering whether to buy credit card lender

Capital One

(COF) - Get Report

after its recent slide.

The card companies are a sorry bunch. Though apparently cheap,

Providian's

(PVN)

looking very sickly after an earnings warning. Even if

MBNA

(KRB)

weren't overpriced, which it is, investors would be well advised to shun it, due to its poor disclosure practices. Highly-leveraged

Metris

(MXT)

faces huge challenges in this economic downturn.

This leaves Falls Church, Va.-based Capital One. It has a diversified loan portfolio that isn't weighted toward tainted, or so-called subprime, borrowers. The management commands respect on Wall Street. Out of all the consumer lenders that grew exponentially in the '90s credit bubble, this company will almost certainly survive.

But that doesn't mean it's worth buying at $48, the price it closed at Wednesday. If you believe the current recession will take a heavy toll on card companies, a safe entry point for Capital One is as low as $20 to $25: A stress-case model, one assuming a nasty economic downturn, predicts earnings of $1.50 per share next year, half what Wall Street analysts are expecting.

Static and Silence

When looking at Capital One, it's crucial to tune out all the bull-market chatter about the "excellence" of the company's credit-scoring model. People used to talk with the same breathlessness of the model used by credit card lender Providian. Unsurprisingly, these models performed superbly during prosperous times, but now, as the economy falters, they're really going to get tested.

Second, when it comes to disclosure, Capital One's management deserves no premium. Capital One doesn't break down its portfolio according to the creditworthiness of its borrowers. (The three classes of borrower Capital One lends to are: subprime, people with spotty credit histories; prime, people with good credit histories; and superprime, high-income people with excellent credit histories.)

In addition, Capital One refused to tell

Detox how big its average credit line is. That's a key number for Capital One. Why? Because one element in the bull case is that Capital One gives borrowers smaller credit lines than its peers. Ergo, in tough times, stressed borrowers who want to tap out their full credit limit have access to less debt than they would at another lender. The danger of adverse selection is lower at Capital One, it is said.

Mounting Losses?

But a low valuation can compensate for such imperfections -- and this is where Capital One is starting to get interesting for some investors. Analysts currently expect Capital One to make $3.58 per share in 2002, up 23% from forecast 2001 profits. That's unlikely to be achieved. But even if it isn't, Capital One could still be a buy, according to some analysts.

John McDonald,

UBS Warburg's

crack credit card analyst, has built a stress-case scenario that has earnings coming down to $3 per share. (McDonald rates Capital One a buy and Warburg has done underwriting for the lender.) OK, that would be a mere 3% above the $2.91 forecast for this year -- well below the growth investors have come to expect. And sure, it would give the stock, at its current price of $48, a price-to-earnings ratio of 16, which is pricey for low single-digit growth. However, if the stock fell to $30, it'd be trading at 12 times. Not bad for a company that can probably grow earnings at 10% to 15% over the next five years.

But is McDonald's $3 as far down as earnings could go in a tough economic environment? Maybe not. According to the bears, the nightmare scenario for Capital One is if its loan growth slows while bad loans keep rising, pushing up its loan-loss ratio and requiring a higher loan-loss provision. It's additions to the provision that hit earnings.

McDonald's stress model has loan growth slowing to 5% in 2002, from a possible 50% this year. Also, his crisis scenario has fee income and marketing spending decelerating sharply in 2002, and losses jumping 66%.

Even with these grave predictions, the company still makes $3. What's not to like? Well, the 66% jump in losses may be too optimistic. After all, losses are predicted to go up by that much in 2001. If losses were to double in 2002 they could reach $3 billion. And, in turn, if Capital One's provision in the year was 105% of losses, it would be $3.15 billion.

A provision that high could take net income down to around $360 million, or around $1.50 per share.

Of course, this exercise assumes the worst. And, as Tom Brown, manager of the Second Curve Capital hedge fund, points out, Capital One has often proved its skeptics wrong, especially when it came roaring back after the 1998 stock market crisis. (Second Curve is long Capital One.)

But no recession occurred in 1998 or 1999. In fact, we had some of the best economic growth seen in the postwar era. Capital One's multiple says that sort of economy is about to return. It isn't. So, stay away.

Know any companies that the market may be misvaluing? Detox would like to hear about them. Please send all feedback to

peavis@thestreet.com.

In keeping with TSC's editorial policy, Peter Eavis doesn't own or short individual stocks. He also doesn't invest in hedge funds or other private investment partnerships.