Stock forecasts and price targets

A consensus price target is an average of opinions, and averaging hides the thing worth knowing. Two stocks can carry the same consensus while one has every analyst within a few percent of each other and the other has half predicting a double and half predicting a halving. Those are completely different situations described by the same number.

A wide spread almost always means the outcome hinges on something specific and unresolved. That is useful, because it tells you what to go and read about. A narrow spread means analysts agree, which is worth less than it sounds: agreement is common right up until the moment it is wrong.

Two cases produce a blank. Some companies, mostly smaller and newer ones, simply have no meaningful analyst coverage. Others have coverage that has gone quiet, and a target published against facts that have since changed is worse than no target at all.

So the figures are withheld past a freshness window rather than shown with a caveat. A stale number gets quoted; a missing one gets questioned. The page always says which of the two situations applies.

An analyst builds a model of the business, projects revenue and margins out several years, discounts the resulting cash flows back to today, and arrives at a number. Or they take an earnings estimate and multiply it by a multiple they judge appropriate. Either way the output inherits every assumption that went in, and small changes to growth or margin assumptions move the answer a great deal.

That is not a criticism of the method. It is the reason the number should be read as one informed argument rather than a measurement. Two analysts modelling the same company from the same public filings routinely arrive at targets 40% apart, and neither has made an arithmetic error. They have made different judgements about the future.

The horizon is almost always twelve months, which is short enough that most of the value in the model sits in a terminal assumption about what happens after it. That is worth knowing when a target is quoted as though it were a forecast of next year.

Analyst targets carry three well-documented tilts. They cluster near the current price, because a target far from where a stock trades is a career risk and a target close to it rarely is. They are revised after news rather than before it, which makes the consensus a lagging summary rather than a leading indicator. And they skew optimistic across the market as a whole.

The correction is not to ignore them. It is to read the distribution rather than the average, and to treat revisions as more informative than levels. An analyst cutting a target after holding it for a year is telling you something changed in their model; the absolute number they landed on is telling you much less.

The rating mix is subject to the same tilt. Buy ratings substantially outnumber sells across the market, so a stock with a majority of holds is closer to genuine scepticism than the label suggests.

Implied upside is the gap between the consensus target and today’s price. Because targets are revised slowly and prices move instantly, the gap widens most sharply when a share price falls hard. A screen sorted by implied upside is therefore, quite reliably, a list of companies that have recently disappointed.

Sometimes that is exactly the opportunity, because the market has overreacted and the analysts are right to hold their view. Sometimes it means the analysts have not finished cutting yet. The two look identical on the screen and are distinguished only by reading why the price fell.

So implied upside is best used as a prompt to investigate rather than as a ranking. The pages below show it alongside the spread and the recent actions, which together say far more than the single percentage does.

The number of analysts covering a company is itself information. Heavy coverage means the obvious questions have been asked publicly and the price probably reflects the answers, which makes finding an edge harder. Thin coverage means fewer people are looking, which cuts both ways: less competition for insight, and less scrutiny of the things that might be wrong.

Coverage also tends to arrive and leave for reasons unrelated to the business. Banks initiate on companies they might do work for and drop coverage when an analyst departs and is not replaced. A company losing coverage has not necessarily deteriorated; it may simply have become less commercially interesting to cover.

Where we hold no meaningful coverage, the page says so rather than presenting a thin consensus as though it were a broad one. An average of two opinions is not a consensus.

A price target published against last year’s facts is not a weak signal. It is a misleading one, because it looks exactly like a current target and gets quoted as though it were. Once a company has reported twice since the note was written, the number describes a business that no longer exists in that form.

The convention on most data sites is to show it anyway with a date beside it. Our judgement is that the date does not do enough work: the figure gets copied and the date does not travel with it. So figures older than the freshness window are suppressed, and the page states plainly that the most recent note is too old to quote.

The trade-off is real. Some pages will show less than a competing site shows for the same company. A missing number gets questioned; a stale one gets believed, and being quietly wrong is the worse failure for a page people use to make decisions about money.

Searches for a stock’s price in 2030 or 2035 are among the most common things people look for, and the pages that answer them confidently are almost all extrapolating a growth rate rather than reporting anything. Sell-side coverage does not extend that far. The published horizon is twelve months, occasionally two years, and beyond that the honest answer is that no analyst has committed to a number.

What can be said usefully about a long horizon is structural rather than numerical: what the business would have to keep doing to justify its current price, how much of today’s valuation already assumes growth rather than current earnings, and what would have to break for that to fail. Those are answerable from the filings and they age far better than a figure.

So the pages below carry the published targets with their dates and firms, and describe the longer arc in terms of what the price currently assumes. An invented 2030 number would rank perfectly well and be worth nothing to the person reading it.

A change in rating is a discrete event with a date, and it tends to move a share price more sharply than a quiet revision to a target, because it represents someone changing their mind in public rather than adjusting a spreadsheet.

The direction of travel across several firms matters more than any one action. A stock collecting successive downgrades from different banks is being repriced by the market’s collective view, not by one analyst’s. That is why each page lists recent actions with the firm named and the date attached, rather than folding them into a single sentiment score that hides who said what and when.

The most common misuse of a price target is treating it as an instruction. Even taken at face value, it says something about one company in isolation and nothing about the position it would occupy in your account.

A stock with strong consensus upside that duplicates exposure you already hold three times over is a worse addition than a modest one that fills a genuine gap. Conversely, a downgrade on something that represents 2% of your portfolio matters far less than the same downgrade on something representing 20%, and the second case is common precisely because winners grow into it without anyone deciding.

That question, how a view about one company interacts with everything else you own, is the one these pages deliberately leave open. It cannot be answered without seeing the account.