Analyst sell ratings are 5.2% of every rated call on Wall Street

Analyst sell ratings make up 5.2% of every rated call in AnaChart’s records. That is 38,262 sells against 473,654 buys, drawn from 740,965 rated analyst actions reaching back to 2004.

The claim that Wall Street rarely says sell is old enough to be furniture. What it has usually lacked is a number, a list of exceptions, and an answer to the obvious follow-up: if a sell is so rare, what happens after one is issued?

The mix, in one picture

Every rated action in the dataset is mapped to a three level scale. Outperform, Overweight, Strong Buy and their cousins become Buy. Neutral, Equal Weight, Market Perform and In Line become Hold. Underweight, Underperform and Sell become Sell.

Two thirds of the scale is one rating. The bottom third of the scale, the part that tells a holder to get out, is used once in nineteen times.

The ladder is walked, not jumped

There are 99,068 recorded rating changes, meaning cases where we hold both the prior rating and the new one and they differ. Sorting them shows something more specific than reluctance.

All 99,068 recorded rating changes. Buy, Hold and Sell here include their equivalents on each firm’s own scale.

97.2% of rating changes move exactly one rung. A Buy becomes a Hold, or a Hold becomes a Sell. Going the whole way in a single step happened 1,581 times in twenty two years of records, which is 1.6% of changes and roughly 0.2% of all rated actions.

This matters more than the headline scarcity. It means the transition from Buy to Sell is not one decision but two, taken months apart, with a public Hold in between. By the time an analyst arrives at Sell, the position has been signposted twice and a good deal of time has passed.

What happens after a sell

Follow each rating forward to the next change by the same analyst on the same stock, and the sell turns out to be a slower call than its reputation suggests.

Read the last figure again. When an analyst leaves a Hold, four times out of five they go up. The three level scale is not symmetric in how it is used and it is not symmetric in how it unwinds either.

The analysts who will say sell

Restrict the field to the 827 analysts with at least 300 rated actions. The median sell rate in that group is 2.1%, well under the population figure, because the population number is lifted by a minority of heavy sell issuers. Here they are.

Sell rate is the share of an analyst’s rated actions that carry a Sell or equivalent. Minimum 300 rated actions.

Andrew Didora sits at 26.7%, more than twelve times the median for this group. Adam Jonas is the notable one on volume: 1,071 rated actions is one of the largest records in the population, and 259 of them are sells. A sell rate this high is not a rounding artefact from a thin record.

The analysts who never have

The other tail is larger than the first. Of those same 827 analysts, 125 have never issued a single sell rating. That is 15.1%. Lower the bar to 100 rated actions and it becomes 420 of 1,832 analysts, or 22.9%. These are the largest records among them.

Analysts with at least 300 rated actions and zero Sell ratings in the record, ranked by volume.

Heiko Ihle is the extreme case: 1,015 rated actions, 99.9% of them Buy, no Hold to speak of and no Sell at all. James Ricchiuti carries the largest zero sell record in the dataset at 1,150 actions. Michael Sison is the most interesting entry, because at 65.8% Buy he clearly uses Hold as a real signal. He simply stops there.

There is a reasonable defence of the zero sell record, and it is worth stating. An analyst who drops coverage instead of moving to Sell produces the same statistic as one who never goes negative, and dropping coverage is common. That is a limit of what this measurement sees, and it is covered in the method section below.

Sells are not spread evenly across stocks

The scarcity of sell ratings is not uniform. Among the largest and most heavily covered names, it is close to absolute.

Megacap technology names, ranked by sell rate across all rated actions on each.

GOOGL has attracted one sell rating across 1,220 rated actions. AMZN has seven across 1,958. TSLA breaks the pattern completely at 17.8%, which is more than three times the market wide rate and reflects a genuinely split analyst base rather than an evenly negative one.

At the other end, some stocks collect sells the way megacaps collect buys.

Stocks with the highest sell rate among those carrying at least 200 rated actions.

VRTS tops the list at 43.7%, and BYND follows at 39.6%. ED is the one that surprises: a regulated utility with a long dividend record sits third, because a stock bought for yield rather than growth spends a lot of its life trading above what a discounted cash flow model supports.

Is the sell rating disappearing?

Sell ratings as a share of all ratings peaked at 6.5% in 2016 and have fallen since. 2025 came in at 4.0%, the lowest reading in the dense coverage era.

The 2016 to 2020 plateau above 6% covers a period of unusual dispersion in the market, and the decline since 2021 tracks a long equity advance. Whether that is a change in analyst behaviour or simply what a rising market does to a rating scale is not something these records can settle.

How to read a rating once you know this

A Hold is doing more work than it looks like. With sells at 5.2% and the ladder walked one rung at a time, Hold is where negative opinion actually lives. An analyst moving to Hold has taken the first of the two steps to Sell, and 19.5% of Holds do eventually go there.

The rating scale is analyst specific, not universal. A Hold from Ravi Shanker, who issues sells on 21.2% of his actions, means something different from a Hold from an analyst who has never issued one. The distribution behind the person is the context for the call.

Never treat a rating as a standalone data point on a crowded stock. On the most heavily covered names, the base rate of a sell is under 3%. The information is in the deviation from that base rate, not in the label.

How we measured this, and what it cannot tell you

The population is every recorded analyst action in AnaChart’s dataset that carries a rating, which is 740,965 of the 833,361 total events. Records reach back to 2004 with dense coverage from 2013 onward. Ratings are normalised from each firm’s own vocabulary onto a three level Buy, Hold and Sell scale, so Outperform and Overweight are counted as Buy, and Underweight and Underperform as Sell. Analyst identities are consolidated to a canonical name, so an analyst who changes firm keeps a single record.

Rating changes are counted only where both the prior and the new rating are present in the record, which gives 99,068 changes. Duration is measured to the next rating change by the same analyst on the same stock, so any rating still open at the snapshot date is excluded from the duration figures rather than being counted as short.

The main limitation, stated plainly. Dropping coverage looks identical to never going negative. When an analyst stops publishing on a stock rather than moving to Sell, this method records no sell, and it cannot distinguish that from genuine conviction. Some portion of the zero sell records above is quiet exit rather than permanent optimism. Anyone quoting the 15.1% figure should quote that caveat with it.

A second limitation. Normalising every firm’s scale to three levels loses real information. A firm running a five level scale uses its middle rating differently from one running three, and mapping both to Hold flattens that. The alternative, comparing raw labels across firms, is worse.

Related reading: who moves price targets first and who follows, why price targets follow the stock rather than lead it, and why the name on the call matters more than the firm.

Frequently asked questions

How rare is an analyst sell rating?

Across 740,965 rated analyst actions in AnaChart’s records, 5.2% carry a Sell rating or its equivalent. Buy and its equivalents account for 63.9% and Hold for 30.9%. That works out to 12.4 buy ratings issued for every sell rating.

Do analysts ever go straight from Buy to Sell?

Almost never. Of the 99,068 recorded rating changes, only 1,581 went from Buy directly to Sell, which is 1.6%. The reverse move, Sell straight to Buy, happened 1,167 times or 1.2%. 97.2% of all rating changes move exactly one rung, which means the path out of a Buy runs through Hold.

How long does an analyst hold a sell rating?

The median Sell rating lasts 139 days before that analyst changes it on that stock. When it does change, 75.9% of the time it moves to Hold and only 24.1% of the time straight to Buy. For comparison, the median Buy rating lasts 147 days and exits to Hold 93.5% of the time.

Which analysts issue the most sell ratings?

Among the 827 analysts with at least 300 rated actions, Andrew Didora has the highest sell rate at 26.7% of 303 actions, followed by Steve Byrne at 25.1% of 487 and Adam Jonas at 24.2% of 1,071, the largest record in the group. The median analyst in that population sits at 2.1%, so these are roughly ten times the norm.

How many analysts have never issued a sell rating?

Of the 827 analysts with at least 300 rated actions, 125 of them, 15.1%, have never issued a single Sell. Widen the net to analysts with at least 100 rated actions and it is 420 of 1,832, or 22.9%. Heiko Ihle has 1,015 rated actions of which 99.9% are Buy and none are Sell. James Ricchiuti has 1,150 with no Sell among them.

Which stocks get the most sell ratings?

Sells cluster on a small set of names. Among stocks with at least 200 rated actions, Virtus Investment Partners tops the list at 43.7%, then Beyond Meat at 39.6% and Consolidated Edison at 38.1%. At the other end, Alphabet has one Sell rating across 1,220 rated actions and Amazon has seven across 1,958. Tesla is the outlier among megacaps at 17.8%.