You do not want the result. You want the method, the sample size and the error bars.
A backtest that looked clean and a live run that fell well short of it inside a quarter. A data feed that ends at the top of the book. A claim with no n under it.
Every tool you have looked at hides one of two things: the sample, or the method.
Vulcan Trading is a trading intelligence platform built on its own models: order-book reads, candle forecasts, sentiment, regime shifts and portfolio factors across crypto, FX, equities, options, futures, gold and prediction markets.
The card you clicked
Reproduced larger, with its parts named. The sample size sits on the card, the label and the date inside the frame.
This is one calendar day of public index history, computed by us. It is not the product's cell for this date and never claims to be; the source, window and return definition are in the method note.
- the sampleevery session that fell on this calendar day inside the window, counted
- the mean and its error barthe average close-to-close return, with one standard error drawn each side, on the same axis every day
- the distributionthe same sessions, each in its return bucket, so the mean can be seen for what it is
Method note
This card is one calendar day of public index history. The sample is every S&P 500 session that fell on this date in the last thirty years, from January of the first year in the window to the as-of date on the frame, and the return for each is close to close, session over previous session, in basis points on the index level, with no dividends. The number on the card is the mean of those sessions with its standard error beside it; n is how many sessions there were, and weekends and holidays thin some dates. The source is the index's public daily history as published by Yahoo Finance, and the table is recomputed by us on a schedule. What this does not tell you is what this date will do this year: a mean of two dozen sessions with an error bar wider than itself is a description of a small sample, not a forecast and not an instruction.
Show me the sample size or this is astrology.
It is on the card, and it is small: two dozen sessions at most, fewer around holidays. The error bar is drawn at the same scale as the mean, and on most days it is wider. That is the honest shape of calendar seasonality over thirty years, and it is why the card carries n and not a call.
The product runs the same discipline across thirty years of sessions and more than a dozen breakdowns, every cell with its sample size and error bar, and states its method beside each study. This page shows that method on public data you can recompute yourself; it does not preview the product's cells.
Sign in with Google. No card, no trial clock: the Retail view is free for everyone.
One sign-in, two views. The research bench lives in the Pro view.