Skip to the record
RVB Partners

Public record · current to 08 Oct 2026 · 209 chained entries

Research

How much was searched to produce what is published, and the bar every result is measured against. Figures as of 02 Sept 2026.

Why publish the search

The denominator, not just the numerator.

A strategy cannot be evaluated apart from the number of ideas tested before it. Run a thousand backtests and the best of them will look remarkable whether or not anything was found, so a result quoted without its search is a result that cannot be judged.

RVB therefore publishes not only what survived, but the scale of the search that produced it — and corrects every result for it. The figures below are that search. They are not an achievement; they are the denominator the rest of this record is measured against.

How much was searched

528,527
Backtests recorded
Every backtest, parameter sweep and grid cell, across 6,762 entries of an append-only ledger.
829
Strategies researched
Each with its own committed returns, report and verdict.
319
Idea families
Strategies grouped by correlation and by shared code.
206,005
Effective independent trials
The backtest count after collapsing repeated ideas: the number the correction below uses.
517
Effective independent strategies
The same collapse, counted in strategies.
42
Presented as an edge
13 promoted and 29 conditional.

The bar

Search enough and something will look significant by chance. So every result is measured twice: on its own, and against the whole catalogue’s search. The second is the bar a strategy must clear to be promoted, and clearing the first is not, by itself, evidence of an edge.

  1. Strategies researched829
    Every strategy with committed returns, a report and a verdict.
  2. Clear the standard bar (α = 0.05)284
    Significant measured on their own results alone.34% of the catalogue
  3. Clear the whole-catalogue bar13
    Still significant after correcting for every trial recorded.1.6% of the catalogue
About 10,300 false positives are what chance alone would produce at this threshold among 206,005 effective independent trials. That figure is counted in trials, not in strategies, so it is not drawn on the scale above.

224 of the rows in this correction use a Sharpe ratio before the risk-free rate is subtracted, because their committed records predate that adjustment.

Strategy selection

How the catalogue is graded, tier by tier, and what that grading presents, is set out under strategy selection. Both pages read the same published file.

Source

Every figure on this page is read from research.json, generated 02 Sept 2026.