Guide · Markets
Analyst Upgrades and Price Targets Explained
An analyst upgrade raises a broker's rating on a stock, and a price target is its estimate of the share price in about 12 months. AMD rose 13.26% on November 9, 2026, after an Oppenheimer upgrade.
On Monday, November 9, 2026, Oppenheimer upgraded AMD to outperform from market perform and set a $600 price target, about 18% above where the stock would close that day. The note argued that data-centre chips could make up about 85% of AMD's revenue by 2027. AMD was already up about 4% before the open and finished the session 13.26% higher at $508.58. It was not alone, though: Broadcom rose 8.78% and Applied Materials 11.37% on the same day without an upgrade, which is a useful reminder that a rating change is one input to a move, not the whole of it.
Sell-side analysts, who work for brokers and investment banks, publish a rating and usually a price target for each stock they cover. The labels vary by firm but fall into three buckets: positive (buy, outperform, overweight), neutral (hold, market perform, neutral, equal-weight) and negative (sell, underperform, underweight). Words such as outperform and overweight are relative: they mean the analyst expects the stock to beat its sector or the market over roughly 12 months, not that it will necessarily rise. The price target is the analyst's estimate of where the shares should trade in that time, usually built from a multiple of forecast earnings or a discounted cash-flow model.
Changes matter more than levels. Most ratings on large companies are positive and outright sells are rare, so a stock sitting on a buy tells you little, while a move from one bucket to another says the analyst's view has shifted. Research going back to Kent Womack's 1996 study in the Journal of Finance found that prices kept drifting in the direction of a recommendation change for weeks afterwards, more strongly after downgrades. The effect tends to be larger for smaller companies followed by few analysts, and for analysts whose past calls have been right.
Conflicts are part of the picture. In 2003 ten Wall Street firms paid about $1.4 billion under the Global Research Analyst Settlement after regulators found that research had been shaded to win investment banking business. Today FINRA Rule 2241 requires brokers to disclose conflicts such as banking relationships and to separate research from deal-making, and the SEC's Regulation FD, adopted in 2000, stops companies from giving analysts material news before the public. Disclosures sit at the end of every research note and are worth reading.
Price targets deserve extra caution. They often follow the price rather than lead it: after a big rally analysts tend to raise targets so that the implied upside still justifies their rating. The consensus target is an average of many targets, and the spread between the highest and lowest says more about uncertainty than the average does. A target 18% above the price, like Oppenheimer's on AMD, is a statement about valuation under the analyst's assumptions, not a forecast you can trade on by itself.
When an upgrade hits the tape, ask four questions. What is new in the note: higher earnings estimates, or only a new label? How much of the move happened before the open, when most notes are published? Did the rest of the sector move the same way, as chips did on November 9? And is the analyst moving with the crowd or against it? An upgrade that changes the earnings numbers and goes against consensus carries more information than one that follows a rally. Treat it as a catalyst and plan where the idea would be wrong before acting on it.