What the research says about buying the stocks analysts agree on, how Exent selects Gems, and the checks any track record of analyst picks should pass before you trust it.
The evidence says analyst consensus carries real information, but capturing it is harder than it looks. Using consensus ratings from 1986–96, Barber, Lehavy, McNichols and Trueman (2001) found that buying the stocks with the most favorable consensus ratings and shorting the least favorable earned annual abnormal gross returns of more than 4%. That required rebalancing daily and reacting promptly to rating changes. After transaction costs, the net abnormal return was not reliably above zero.
Two lessons follow. Consensus is a signal, not noise. And a strategy that trades on every rating change gives most of the signal away in costs. Whether a narrower approach, such as listening only to analysts with strong records, does better is exactly what a track record has to show.
Gems are Exent’s list of stocks where several of the analysts with the best track records agree. The process, in order:
| Step | What happens |
|---|---|
| 1. Score every analyst | Thousands of Wall Street analysts are tracked, and each one is scored on how their past ratings played out: price-target hit rate and returns after their calls. |
| 2. Keep the top-rated | Only ratings from analysts with 4 stars or more count; the rest are discarded. |
| 3. Require agreement | A stock qualifies only when several distinct top-rated analysts independently rate it a Buy. One analyst, however good, is not enough. |
| 4. Refresh | The list is recalculated twice a day as new ratings arrive and old ones lapse, so stocks enter and leave. |
| 5. Explain | Each Gem comes with a written thesis summarizing why the analysts are positive. |
Gems are research, not personal advice. This page does not list current Gems, because they change twice a day and a stale list would mislead. More on the idea in what Gems are and on the scoring in analyst star ratings.
Their past records are better, by construction. In Exent’s 7-year backtest of 628,792 ratings by the 1,000 top-ranked analysts (statistics as of Jun 24, 2026):
| Star rating | Analysts | Median price-target hit rate | Median share of ratings up after 12 months | Median 12-month return |
|---|---|---|---|---|
| 5 stars | 141 | 60.8% | 68.6% | +19.9% |
| 4 to 4.9 stars | 853 | 52.8% | 62.8% | +12.0% |
| All ratings pooled | 1,000 analysts | 52.4% | — | — |
| Star rating | 5 stars | 4 to 4.9 stars | All ratings pooled |
|---|---|---|---|
| Analysts | 141 | 853 | 1,000 analysts |
| Median price-target hit rate | 60.8% | 52.8% | 52.4% |
| Median share of ratings up after 12 months | 68.6% | 62.8% | — |
| Median 12-month return | +19.9% | +12.0% | — |
Read this carefully. Star ratings are computed from the same past calls the table measures, so higher numbers for 5-star analysts are partly guaranteed by how the stars were assigned. The window was also mostly a bull market, which flatters any bullish call. The honest test of whether top analysts stay top is out of sample: rank them on one period, then measure the next. The full study and its caveats are in analyst accuracy.
Any record of picks, from an app, a newsletter or a fund, should pass the same checks. If a provider cannot answer one of these, treat its numbers with caution.
| Check | Why it matters | What good looks like |
|---|---|---|
| Live, not only backtested | A backtest knows the answers; live picks do not | Results from picks published before the outcome was known |
| Time-stamped entry | Entry prices chosen afterwards flatter returns | Price on the day each pick was flagged |
| Every pick included | Dropping losers creates survivorship bias | All picks, including those removed from the list |
| Defined holding rule | Flexible exits let you count only the best moments | Exit rule set in advance, such as 12 months or removal |
| Costs and taxes | Frequent changes create real trading costs | Returns shown after a realistic cost estimate |
| Fair benchmark | Most picks look good in a rising market | Compared with the S&P 500 and an equal-weight or sector benchmark |
| Risk, not just return | Higher return with deeper falls is not better | Volatility, maximum drawdown and Sharpe ratio reported |
| Enough history | A year or two can be luck | Several years and at least one market sell-off |
Because the rules are tested on data the designer has already seen. McLean and Pontiff (2016) studied 97 published stock-return predictors and found their returns were 26% lower out of sample and 58% lower after publication. Picks strategies face the same pull: the more rules are tuned to the past, the less they hold up. See backtest overfitting, survivorship bias and out-of-sample testing.
That is why the meaningful record for any picks list, Gems included, is the live one, where each stock is logged with its price on the day it was flagged. This page explains how to judge that record rather than quoting a backtested figure.
Research on 1986 to 1996 found the most favorably rated stocks beat the least favorably rated by more than 4% a year before costs, but net of transaction costs the advantage was not reliably above zero.
Exent shows the live record of Gems inside the app, with each stock logged at its price on the day it was flagged. This page explains how that record should be judged rather than quoting a figure.
Exent scores thousands of analysts on their past accuracy. A stock becomes a Gem when several distinct analysts rated 4 stars or more independently rate it a Buy. The list refreshes twice a day.
Most investors and funds do not beat the S&P 500 after fees, and analyst picks are no guarantee. Consensus from accurate analysts is useful research, but it needs to be checked, sized sensibly and measured honestly.
Backtests are built with hindsight, can leave out failed companies and often ignore costs. Research on 97 published predictors found returns were 58% lower after publication than in the original studies.
Stocks Radar ranks thousands of analysts by their past accuracy and shows the Gems they agree on, with a thesis for each and the live record inside the app.
Educational content, not investment advice. Examples are illustrative; past performance does not guarantee future results.