Price targets are numbers, ratings are labels, and the gap between them is a signal most feeds throw away. A Buy held while the analyst quietly trims the price target is one of the more reliable conviction-loss patterns in sell-side data, and you can only see it if the record keeps every revision in order.
AnaChart’s analyst price target dataset is built for exactly this: 661,383 price targets and 759,654 ratings, point-in-time, across 7,191 analysts and 9,686 tickers, back to 2004 with deep coverage from 2013.
Where the factor comes from
Three fields do most of the work. The rating and the price target give you direction and magnitude. The price at issue lets you compute the implied return the analyst was calling on the day they called it, not with hindsight. Line those up over time and the revision path, not the current level, is the signal.
- A rising target under a steady Buy is momentum an analyst is willing to underwrite.
- A falling target under an unchanged Buy is quiet conviction loss, the pattern that leads downgrades.
- A target set far above the price at issue, from an analyst whose targets usually land, is a high-conviction call worth more than a crowd of me-too Buys.
Weight it by who is calling
The reason to keep the accuracy fields next to the ratings is that not every analyst deserves the same weight in a factor. The met ratio and performance score let you scale each contribution by the analyst’s real record on the name, so a signal leans on the analysts whose targets reach. That is the difference between an analyst-sentiment factor and a headcount of Buys. See how the accuracy field works for the mechanics.
Built for a backtest
Every row is point-in-time and survivorship-free, including tickers that later delisted, so a factor built on this data sees each call as it stood on its issue date. The full query that assembles the signal is on the use-cases page, and it lands in your stack as a native Snowflake or BigQuery share.
Try it on one name first
The free Apple sample carries the ratings, targets, price-at-issue, and accuracy fields for every analyst on AAPL, enough to prototype the factor on a single name. Request the AAPL sample, or ask about a custom cut.
FAQ
What fields do I need to build an analyst-sentiment factor?
Rating, price target, and price at issue give you direction, magnitude, and the implied return on the call date. Add met ratio and performance score to weight each analyst by track record.
Is the data point-in-time?
Yes. Every target and rating is stored as it stood on its issue date, survivorship-free, so a backtest has no look-ahead bias.
How is it delivered?
As a native Snowflake or BigQuery share, or a custom one-time file if you do not run a warehouse.