# The Honest Backtest Census — Release 2026-09-25

> As of September 25, 2026, 25 famous, named trading strategies were each backtested on 36 liquid US stocks & ETFs — 900 strategy-asset combinations, of which 839 had enough trades to judge (61 were too thin and excluded). Of those judged, about 86% failed to beat simply buying and holding the same asset over the same 5-year window — net of costs and slippage, with no look-ahead. This is a survivorship-biased set of well-known assets, measured over the full period with no out-of-sample split, against each asset's own buy & hold, and is not significance-tested.

**Attribution (CC BY 4.0):** Stratsemble, "The Honest Backtest Census: How Often Famous Strategies Beat Buy & Hold" (Release 2026-09-25), CC BY 4.0, https://stratsemble.com/methodology/census/2026-09-25
**Canonical page:** https://stratsemble.com/methodology/census/2026-09-25 · **Methodology:** https://stratsemble.com/methodology

## What this is
An aggregate, openly-licensed snapshot of how 25 famous, named trading strategies performed against
buy & hold across 36 liquid US stocks & ETFs over a 5-year window, as of September 25, 2026.
The headline figure is STOCKS-ONLY (composition-stable). Crypto and "all" rows are context only — a small,
volatile, survivorship-biased sample; an overfitting artifact, not a demonstrated edge.

## Caveats that travel with every number
- **Survivorship-biased** set of well-known assets (labelled, not the whole market).
- **No out-of-sample split** — canonical fixed parameters over the full window; **not significance-tested**.
- Measured against **each asset's own buy & hold** (total-return basis), never a market index.
- **Costs on**: 0.05% per side, next-bar fills, costs and slippage on, no look-ahead.
- "beat" = excess return > 0 by any margin; combinations with too few trades/history to judge are **thin** and excluded.
- Drawdowns are a **negative fraction** of equity and are published as a PAIR (strategy vs buy & hold): a strategy
  that draws down less has usually just sat in **cash** — that is avoidance, not skill, and it still mostly lagged.

## Files
- `by_asset_class.csv` — beat/fail rate + n for stocks (headline), crypto, all.
- `by_strategy.csv` — the 25 strategies (stocks slice): beat/fail rate, n, and a MEDIAN excess (a cross-asset
  central tendency, NOT a forward expectation; `low_sample_flag` marks n_eligible < 10).
- `drawdown_summary.csv` — percentile drawdowns, strategy vs buy & hold, per slice.
- `census-2026-09-25.json` — all of the above plus metadata and the ready-to-paste citation.

## What is NOT here (by design)
No per-trade records, no equity curves, no per-combination rows, no strategy parameters, no custom-parameter runs,
and not the backtesting engine. This is the published ANSWER (aggregate rates), not the raw inputs or the tool.