Three days ago, Merrill Lynch analyst Tal Liani decided that all the bad news coming out of

Scientific-Atlanta

(SFA)

was finally reflected in S-A's dwindling stock price.

He was wrong.

Liani, an outspoken bear who has been warning of impending doom at the cable TV equipment manufacturer over the past six months, wavered in his criticism of the company Tuesday, upgrading the stock from sell to hold because the stock price had fallen to a reasonable level.

But after S-A fell way short of analysts' expectations with its Thursday night quarterly earnings release, Liani changed his mind on the stock for the second time this week. In a report entitled "Two Days in Paradise," the Merrill analyst cut Scientific-Atlanta right back down to a sell.

"Despite the fact we're negative" on Scientific-Atlanta, Liani told

TheStreet.com

Friday, "we weren't negative enough."

One Scientific-Atlanta short-seller says Liani should be praised for responding to a changing situation, not criticized for his flip-flop. "I give him high points for his flexibilty," says the short-seller, speaking on condition of anonymity. "But," adds the short-seller with a laugh, "I could see where others say, 'Geez, this guy blows hot and cold every day.'"

Shares in Scientific-Atlanta, down more than 60% from their 52-week highs, traded down 24% from Thursday's close on Friday's opening, but recovered to $11.60 later on, down $1.73, or 13%.

Aside from reflecting the sell side's current hair-trigger approach to ratings changes, Liani's quick turnaround illustrates the difficulties of being an equipment supplier in the cable television business these days. Like their suffering counterparts in the telecom world, S-A and

Motorola's

(MOT)

General Instrument must contend with a limited pool of buyers who are making alarming, well-documented cutbacks in capital spending.

For the Sept. 30 quarter, S-A recorded revenue of $311.6 million, down 24% from the corresponding quarter one year earlier and off 13% from the First Call/Thomson Financial consensus forecast of $358.2 million. Earnings per share -- excluding "special items" such as restructuring charges, bad debt from a customer bankruptcy, and the declining value of equity investments -- amounted to 13 cents, below the consensus of 21 cents.

In his report Friday morning, Liani -- who says the past few quarters' positive suprises caught him off guard -- returned to his thesis that S-A is hobbled by a variety of negative trends. These include a shrinking demand for the company's set-top cable boxes, the likelihood that purchases from major customer

AOL Time Warner

(AOL)

stemmed from an inventory buildup rather than sustainable demand, and falling average selling prices for boxes. Merrill Lynch doesn't do banking for S-A.

Liani

hasn't been shy in his criticisms of Scientific-Atlanta. He slapped the company with a sell rating in April and followed up in July with the bluntly titled report "Six Reasons for a Sell Rating."

Asked how he might have erred in his relatively optimistic report Tuesday, Liani says the target price-to-earnings ratio of 15 that he chose for the stock (excluding financial income) was likely too high for a company that's not growing. Plus, he says, his earnings forecasts for the stock proved to be too optimistic.

Before Tuesday, says Liani, he was hearing criticism from investors for his harsh position on S-A. Given the company's substantial cash on hand, its positive cash flow and its apparently "decent" P/E ratio, he says, investors told him that S-A was a good deal -- relatively speaking, that is, compared to other telecom equipment players such as

Lucent

(LU)

.