The consensus you can buy today tells you what analysts think now. It cannot tell you what they thought on a Tuesday in 2015, which is the only thing a backtest cares about. Rebuild history from a live feed and you inherit two silent errors: revisions that overwrite the past, and tickers that quietly dropped out when they delisted.
AnaChart’s analyst price target dataset is stored the other way around: point-in-time and survivorship-free. 660K+ price targets and 750K+ ratings, each kept as it stood on its issue date, across 7K+ analysts and 9K+ tickers, active and delisted, since 2008.
Why point-in-time matters
Every target carries its issue date and the price at issue, and the previous target is kept rather than overwritten. So when you ask what consensus looked like on any past date, you get the answer as it stood then, not as it was later revised. That is the difference between a backtest you can trust and one that has quietly read tomorrow’s newspaper.
Why survivorship-free matters more
The bigger leak is the one you cannot see. A universe that only contains today’s live tickers has already dropped the companies that failed, merged, or delisted, which is exactly the set a strategy needs to be tested against. AnaChart keeps delisted, renamed, and acquired names, so a backtest sees the losers as well as the survivors. For a name’s track record across firm moves, see the accuracy data.
How to join it to your history
Join on ticker and action_date against your own price or holdings history and every row lines up without look-ahead bias. The full query is on the use-cases page, and the data arrives as a native Snowflake or BigQuery share so there is no pipeline to keep in sync.
Test the shape on one name
The free Apple sample includes the point-in-time fields for every analyst on AAPL, enough to check the structure before you license the universe. Request the AAPL sample, or ask about the full history.
A consensus as it stood on the day
Point-in-time data lets you rebuild a consensus on any past date. Take Meta Platforms in October 2022 and keep each analyst’s latest target set in the previous twelve months:
| As of | Analysts | Median target | Lowest | Highest |
|---|---|---|---|---|
| 26 Oct 2022 | 47 | $230.00 | $116 | $466 |
| 31 Oct 2022 | 48 | $217.50 | $80 | $466 |
Five days moved the median by $12.50, and the lowest target from $116 to $80. The $466 in both rows was set by Ivan Feinseth on 7 July 2022 and was still his latest target in the window. A feed that stores only the current consensus cannot give either row. One that restates history would show the October 2022 consensus with targets set later.
FAQ
What makes analyst data point-in-time?
Each target and rating is stored with its issue date and the price at issue, and prior values are kept rather than overwritten, so any past date can be reconstructed as it stood then.
Does it include delisted stocks?
Yes. Delisted, renamed, and acquired tickers are kept, which is what makes a backtest survivorship-free.
How far back does it go?
Records reach back to 2008.
Can I rebuild a consensus for a past date?
Yes, from the dated rows. On 26 October 2022 the median of 47 analysts’ latest Meta targets was $230.00. On 31 October 2022 it was $217.50, across 48 analysts.
Get the rows, or ask a question
Every scored target on one stock is a $29 CSV, no subscription: S&P 500 and Nasdaq-100 names plus 607 Canadian and 184 German stocks. The Apple file is a 100-row preview with no sign-up, and the full file sent to a work email address on request. The first question you ask is answered free, and further queries start at $149.
Studies from the same data: how often analyst price targets get hit, price target hit rate by bank and analyst price targets by year issued.