The fun part of investing is constructing a portfolio -- assembling a group of stocks that you will have and hold practically forever.

The key is diversification. You don't need a lot of variety, but you're taking a big gamble if you concentrate on a few stocks or a few sectors. The problem is not so much that you will make the wrong choices, but that the volatility -- the ups and downs -- of a highly concentrated portfolio can be extreme, and you'll be tempted to make changes at the wrong times.

A good illustration is Merrill Lynch Growth Fund. The manager, Arthur Moretti, moved heavily into energy stocks in 1997, and the result was disaster. The fund lost 9 percent of its value in the fourth quarter of that year and 24 percent in 1998, trailing the Standard & Poor's 500-stock index by 53 percentage points. Investors redeemed their shares like mad, and in January, Merrill put a new manager, Steve Silverman, in charge.

In late October of last year, Moretti had committed 37 percent of the fund's assets to oil and gas stocks, a proportion seven times the weighting of this sector in the S&P. Silverman jettisoned nearly all the energy stocks in late January, just as these shares were hitting bottom. A short time later, oil prices began to rise and energy stocks soared.

"If he had simply held onto the stocks the fund owned on Oct. 31," wrote Patrick McGeehan in the Wall Street Journal last week, "the fund would be up more than 25 percent -- rather than down 8 percent, which is its performance year-to-date."

Silverman did what any sensible person would do -- tried to correct an imbalance in his portfolio -- and the move backfired.

Now Scott Schoelzel is facing a similar decision. "A Fund's Quick Slide From First to (Almost) Worst," said the headline in the New York Times yesterday. The fund was Schoelzel's Janus Twenty, which returned a whopping 73 percent in 1998 and 23 percent in the first quarter of 1999.

But for the past 2 1/2 months, it has ranked 15th from the bottom among the 4,008 general equity funds followed by Lipper Inc. From April 1 through Monday, the fund was down 5 percent, compared with a gain of 4 percent for the S&P and a gain of 10 percent for the Dow Jones industrial average.

Schoelzel owed his success in 1998 and the first three months of 1999 to a huge emphasis on technology stocks. On March 31, five of his top six holdings were tech firms, and those stocks alone -- America Online Inc. (AOL), Microsoft Corp. (MSFT), Dell Computer Corp. (DELL), Nokia Corp. (NOK) and Cisco Systems Inc. (CSCO) -- represented 38 percent of his total assets.

In the spring, what some analysts saw as a tech bubble started to deflate. AOL, Schoelzel's top stock (16 percent of the fund!), dropped from $167.50 on April 6 to $90.50 on June 14.

What should Schoelzel do? He says he won't change direction, and he's right. As long as he likes the companies, he should keep them. He certainly shouldn't let recent setbacks push him into hyperactivity.

This is not to say that Schoelzel's portfolio is one you should replicate at home. It lacks the diversification that a reasonable portfolio must have in order to avoid wild volatility.

Research shows that, as long as you own about a dozen stocks in different industries, you should achieve performance that's roughly the same as the market as a whole. You won't necessarily beat the S&P, the benchmark for money managers, but you shouldn't get into serious trouble.

A typical, balanced portfolio of 12 stocks of equal market values might include two or three high-techs, two financials, two manufacturers, one or two service stocks, one retailer, one drug stock, and one energy company, real estate investment trust or utility.

In looking for ideas on balancing your portfolio, most mutual funds are too big (they average 100 different stocks) to help. But smaller funds can show how managers try to achieve both safety and growth.

Tom Marsico -- who once ran Janus Twenty (which, by the way, is closed to new investors) and now manages his own fund, Marsico Focus (1-888-860-8686) -- is a master. At last report, his fund held just 25 stocks, a very low number.

Like Janus, Marsico is top-heavy in technology, but not nearly to the same degree. He owns seven tech stocks, representing 37 percent of his total assets. But, as with Janus, the tech companies themselves are solid citizens such as EMC Corp. (EMC), the data-storage specialist; Intel Corp. (INTC), the semiconductor giant; and International Business Machines Corp. (IBM).

Marsico also owns four financial stocks, including Citigroup Inc. (C), with interests in banking, insurance and investing, and Fannie Mae (FNM), the mortgage maker. He owns three consumer companies -- McDonald's Inc. (MCD), Coca-Cola Enterprises (CCE) and Anheuser-Busch Cos. (BUD), the world's largest brewer -- and one retail chain, Home Depot Inc. (HD).

He rounds things out with two airlines, three media and telecommunications firms, two auto companies, one airplane manufacturer, one health company, and one diversified giant. Is this portfolio diversified enough? It seems light on service companies -- Automatic Data Processing Inc. (AUD), the payroll outsourcing firm, would be a good addition -- and on manufacturing and health care. Utilities, real estate and energy are absent.

Marsico, however, has created a good mix. His fund didn't quite have the spectacular returns of Janus Twenty in 1998; still, he was up 51 percent, which isn't too shabby. But he doesn't face the pressure that Schoelzel does to lighten up on high tech. Through Monday, Marsico had returned 13 percent for 1999, compared with 8 percent for the S&P.

But understand that Focus is a large-cap domestic-stock fund. If you create a portfolio like this on your own, you'll need to get small-cap and international stocks somewhere else (perhaps by buying separate mutual funds).

The Inside Scoop

"Growth manufacturers now!" trumpets Elliott L. Schlang in his latest Great Lakes Review (216-621-1330). Schlang, who is normally far more restrained, is great at ferreting out boring Midwest industrial firms with strong balance sheets and a history of increasing their earnings.

In his newsletter, available only to institutional clients, Schlang recommends Kaydon Corp. (KDN), which makes parts for manufacturers and trades at a price-to-earnings ratio of 16; Densply International Inc. (XRAY), the world's largest dental supply company, at a P/E of 17 with annual growth of 15 percent; Hillenbrand Industries Inc. (HB), hospital beds and caskets, at a P/E of 15; and Myers Industries Inc. (MYE), shipping and storage containers, at a P/E of 13 based on 1999 earnings and a growth rate of 14 percent. Also in the June 9 letter, Schlang urged clients to buy Holophane Corp. (HP), but on Monday it jumped 34 percent on a takeover bid.