Moving average crossover (20/50) + weekly filter
The higher time frame's context is the view of the trail from above: trading with it, instead of looking only at the short chart, greatly raises the trade's probability.
Measured in crypto — Binance spot · 540 pairs, delisted ones included · 0.2% per round trip
- ✓invariance
- ✓costs
- ✓placebo
- ✓benchmark
- ✓out of sample
- ✓multiple testing
What was measured
How many of those trades actually count
What it paid, before any deductions
Why it passed
- the profit spread across the trades instead of coming from a handful
- the edge survived the broker's fee
- it paid more than entering on randomly drawn dates
- it paid more than buying and holding for the same stretch
- the result came back in the separated pieces of the data
- it survived the discount for having tested many versions
What this result does NOT say
- One market, one universe
- Measured on 540 spot cryptocurrency pairs, delisted ones included. It says nothing about futures, equities or indices, nor about how the same technique behaves in another market.
- One window of time, not every window
- The measured period runs from 2017-08-17 to 2026-07-28. A market moves through regimes, and a technique can work in one and fail in another — the card measures the regimes that fit inside this window, not the ones still to come.
- One cost structure
- The cost charged is 0.10% per leg, in and out. Anyone paying more than that gets a worse result, and anyone paying less gets a better one — the verdict holds for this fee.
- One exit rule
- The trade was closed by: the auditor's fixed horizon (10 bars). The same entry measured with a different exit is a different strategy, and can earn a different verdict — it happens in this archive.
- The number of trades is not the sample size
- There are 10,529 trades, but only 301 independent market episodes: a single move fires the technique across dozens of assets at once, and counting those as separate observations inflates any result. It is the smaller number that governs the arithmetic. With 301 episodes, what the data supports is a range from +1.09% to +3.38% per trade — the published average is the centre of it, not the exact measurement.
- Daily bars
- Measured at the daily close. Nothing here measures what happens inside the day, and an intraday technique is not auditable with this data.
Numbers and reproducibility
The six controls
| control | status | t | threshold | episodes |
|---|---|---|---|---|
| invariance | passed | — | — | — |
| costs | passed | 3.50 | 1.97 | 301 |
| placebo | passed | 3.86 | 1.97 | 301 |
| benchmark | passed | — | — | — |
| out of sample | passed | — | — | — |
| multiple testing | passed | — | — | — |
- invariance10529 signals across 750934 bars
- costsgross +2.236% · cost 0.200% · net +2.036% (t=3.50)
- placeboactual +2.036% · placebo -0.240% · excess +2.276% ± 0.589% (t=3.86 against a threshold of 1.97, 301 real groups, 526,450 sham dates, draw error ±0.026%)
- benchmarktechnique +2.04% · buy and hold (same horizon) +0.03% · excess +2.01%
- out of sampleasset half A: +2.044% (t=3.12, 288 episodes) · asset half B: +2.026% (t=3.15, 284 episodes) · liquid half (>= US$ 2,066,613/day): +1.869% (t=2.78, 297 episodes) · illiquid half: +2.225% (t=3.10, 264 episodes) · period 1/4 (2017-10-11 a 2022-04-27): +2.109% (t=2.15, 147 episodes) · period 2/4 (2022-04-28 a 2023-11-09): +1.334% (t=1.21, 56 episodes) · period 3/4 (2023-11-10 a 2025-05-05): +2.808% (t=2.10, 55 episodes) · period 4/4 (2025-05-06 a 2026-07-18): +1.892% (t=2.58, 45 episodes)
- multiple testing1 variation(s) tested · t=3.50 across 301 episodes (equivalent to t=3.49) · p≈0.0005 · false positives expected by chance ≈ 0.00
Reproducibility
- period
- 2017-08-17 to 2026-07-28
- assets that traded
- 540
- variations tested before this one
- 1
- gross per trade
- +2.24%
- net per trade
- +2.04%
- exit rule
- the auditor's fixed horizon (10 bars)
- horizon bars
- 10
- fee per leg
- 0.001
- episode days
- 10
- fast average
- 20
- slow average
- 50
- filter weeks
- 4
Binance spot klines (delisted pairs included) · collected from 2026-07-27 23:31 to 2026-07-28 14:29 · 540 assets · 750,934 bars · 2017-08-17 to 2026-07-28
Seed not recorded: audits before 2026-07-29 used 20260728 by convention in the code, and the value is not in the record.
Hypothesis, filed before the result
The central thesis of the multi-timeframe transcription, isolated and testable: using the direction of the HIGHER timeframe to filter trades on the lower one improves the result. Tested by adding the weekly filter to techniques already in the corpus. A specific, falsifiable prediction, written before any number: the filter does NOT improve — it will cut the n and worsen the error bar, like the volume filter (cut 73% of the n, worse gross, error 42% larger) and the risk-reward one (varying from 1.0 to 4.0, it moved the gross by 0.06 percentage points). If I am wrong, this is the corpus's strongest finding: it would be the first time a guru's filter separates anything.
filed on 2026-07-28, before the number existed
The original, as it was filed
A tese central da transcrição multi-tempo-gráfico, isolada e testável: usar a direção do tempo gráfico MAIOR para filtrar operações no menor melhora o resultado. Testada acrescentando o filtro semanal a técnicas que já estão no corpus. Previsão específica e falsificável, escrita antes de qualquer número: o filtro NÃO melhora — vai cortar o n e piorar a barra de erro, como o filtro de volume (cortou 73% do n, bruto pior, erro 42% maior) e o de risco-retorno (variando de 1,0 a 4,0, moveu o bruto em 0,06 ponto percentual). Se eu estiver errado, este é o achado mais forte do corpus: seria a primeira vez que um filtro de guru separa alguma coisa.
Pre-registration exists to keep prediction apart from rationalisation: written after the number, every hypothesis is right.