Wall Street analysts aren't much use right now if you're trying to figure out what a technology stock is worth. Their earnings estimates are just too fluid, falling hourly on some stocks, it seems. How can you possibly value a stock on these numbers?

But I don't think this is the time to ignore analyst estimates, either. The ebb and flow of their opinions -- particularly something I call "analyst capitulation"-- is a critical guide to figuring out the bottom in individual technology stocks, in my opinion.

And, it turns out, studying when analysts give up on a stock can give investors the solid numbers they need to value a stock on the fundamentals.

At the bottom of a market or in an individual stock, investors capitulate. Some will sell shares at any price just to get out. Others won't buy no matter how cheap the stock gets. That moment -- the moment of maximum despair -- is a bottom.

Before a market or stock gets to that final moment, however, Wall Street analysts have to go through their own ritual of capitulation. Stocks, no matter the price, get downgraded to "neutral" or "hold." No matter how cheap they get, no one issues a buy. Analysts compete with each other to see how deeply they can slash an earnings estimate. Right now, we're about halfway through this process of analyst capitulation for the technology sector.

Analyst capitulation sets the stage for the sustainable rally. All those "hold" and "neutral" ratings become fuel for the market when they're converted into "accumulates" and "buys." Just as downward earnings revisions kicked the stuffing out of stocks on the way to the bottom, upward revisions help support the market on the way up.

The value of identifying analyst capitulation in a specific stock should be pretty clear. If you can tell when the analysts have thrown in the towel on a stock, you'll have a very good indication that the bottom in that stock is near.

A Six-Step Guide for Finding a Stock's Bottom

So how do you tell if a stock is approaching analyst capitulation? And how can you tell how far away that bottom might be?

I think it's pretty easy to ballpark analyst capitulation for an individual stock. I'll use

Intel

(INTC) - Get Report

to show you how my method works, using enough detail so that you can use this example as a template for studying other stocks. And then I'll move on to

Nortel

(NT)

and

Texas Instruments

(TXN) - Get Report

. For the sake of simplicity, I've reduced the process to just six steps.

Step one: Look at How Far Earnings Projections Have Come Down in the Last Three Months.

You can get this information easily on

MSN MoneyCentral

. After looking up a stock quote, simply click on "Analyst Info" on the left side of the page and then click "Consensus EPS Trend." The numbers for Intel show that analysts, at least, have finally started to throw in the towel on Intel's prospects for 2001. Just three months ago, the consensus estimate for 2001 earnings stood at $1.70; today the estimate is $1.04. That's the kind of huge whack to future earnings projections that you want to see here. A high degree of pessimism now colors opinions of Intel's future earnings power -- and that creates the possibility that analysts are reasonably close to a bottom in their opinion of the stock.

Step Two: Look for Signs that Earnings Projections Are Still Falling.

Analysts haven't capitulated until estimates hit bottom and stop falling. When checking stock quotes on Money Central, click on "Advisor FYI" on the left side of the page to see if analysts continue to downgrade the stock and to cut earnings estimates. Click "Recent News" and check the stories there for the same evidence of still-deepening pessimism. For example, note the news story saying that on March 7

Wit SoundView

cut its 2001 earnings estimates for Intel to 95 cents from $1.10. It's clear that analysts are still lowering earnings estimates.

Step Three: Judge the Odds That the Current Consensus Estimate Will Hold.

Check this by studying the spread between the high and low analyst estimate that you can find on MoneyCentral by again going to "Analyst Info," this time looking at "Estimates." In Intel's case you'll notice that there's a huge spread between the high estimate for 2001 of $1.31 a share and the low estimate of 88 cents. That degree of uncertainty among analysts virtually guarantees that we'll see further erosion of estimates. The pack psychology of analysts also argues for further downward revisions. No analyst wants to be 43 cents -- almost 50% -- higher than the low estimate. That's just too far in front to be comfortable.

Step Four: Look Around for Anything That Might Set Off Another Round of Cuts.

A warning from a competitor, for example, can lead analysts to lower estimates for a stock that previously seemed to have found a bottom. So, too, can bad news from a major customer or supplier. Bad news from PC makers such as

Dell

(DELL) - Get Report

and

Compaq

(CPQ)

, for example, could lead analysts to slash estimates for Intel yet again.

Step Five: Check the Buy/Hold Scorecard.

How pessimistic have analysts become in the last 90 days? In the case of Intel, not very. Eight analysts now rate the stock a "hold" -- that's just one away from the seven holds of 90 days ago. Twenty-one analysts still rate the stock "buy" or "strong buy." I see a lot of potential downgrades here.

Step Six: Put It All Together.

I believe this approach shows that the estimates for Intel's earnings are likely to fall some more. The huge drop in the last 90 days from $1.70 to $1.04 suggests to me that a lot of bad news is in the estimates. But the big spread between high and low estimates and the continued eruption of new estimate cuts suggests that the trend in estimates is still downward.

Lehman Brothers

, for example, cut its 2001 estimate to 90 cents on March 7. I'll be conservative and say that I expect lowered guidance after both the March and June quarters, and that the low point in estimates for the year will come in below 90 cents, a little more than 13% below the current consensus.

Compare Intel to Nortel Networks right now. To me, Nortel seems to have more downside ahead of it than Intel does. Earnings estimates for Nortel in 2001 have fallen markedly in the last 90 days -- to 72 cents from 98 cents -- but that's only a 27% cut compared to Intel's 40% drop. The stock has seen its share of downgrades in the last 90 days and nine analysts now rate the stock a "hold" -- up from just one analyst 90 days ago. But 27 still call it a "buy" or "strong buy," convincing me that the revision trend still has a way to run. And there's a huge gap between the high estimate of 99 cents and the low estimate of 30 cents for 2001.

In comparison to Intel, I think Nortel's estimates have much further to fall. That opinion gets support from the negative business news still coming out from such Nortel suppliers as

JDS Uniphase

(JDSU)

. I think we could see consensus analyst estimates fall to 50 cents a share for 2001 before we're done. In what I see as a precursor of that general move,

Morgan Stanley

cut its estimate to just that -- 50 cents a share -- on March 7. Analyst estimates for Nortel are likely to fall significantly further before analysts finally throw in the towel.

Texas Instruments, on the other hand, seems to be closer to analyst capitulation than Nortel, if not as far along as Intel. The last round of estimate cuts came at the end of February after the company announced that first-quarter revenue would fall by 20% instead of the previously announced 10%. Analyst estimates have plummeted in the last 90 days to 84 cents from $1.56 a share. Still, the spread of estimates remains huge with $1.14 at the top and 65 cents at the bottom, and while nine analysts rate the stock a "hold," 14 still give it a "buy" or "strong buy." Texas Instruments is probably safe from further estimate cuts and rating cuts until it issues guidance when it announces earnings in April. I'd expect another round of estimate cuts then, however.

A Useful Stock-Picking Tool

Studying analyst capitulation can help you time buys and sells, but it can also give you very useful insight into what stocks to buy and sell. In a recent column, I suggested that, with economic growth unlikely to quickly re-accelerate to the 1999 pace, investors are potentially looking at an extended period of relatively modest growth in corporate earnings. In that environment, growth-stock investors want to be sure they're buying solid growth companies at modest valuations.

Sounds good. Except that it's very difficult to tell if a valuation is "modest" when earnings projections are subject to drastic downward revision. Studying analyst capitulation is one way to get a more accurate read on what the price-to-earnings ratio on a stock will actually look like when all the dust has settled.

For example, at today's consensus earnings estimate of 72 cents for 2001, Nortel Networks trades at 27 times projected 2001 earnings per share. But if estimates continue to fall and 50 cents a share becomes not just the Morgan Stanley estimate, but the consensus projection, then Nortel would be selling at a P/E ratio of 40 on projected 2001 earnings. Quite a difference -- and an important one to any investor trying to compare stocks and buy growth at the most reasonable price.

At the time of publication, Jim Jubak owned or controlled shares in the following equities mentioned in this column: Broadvision, Intel and Texas Instruments.