PASSED

Moving average crossover (20/50) · exit by standard deviation f=2.06

A trend technique delivers its result despite the trades that turn against it, not because of them. Cutting each trade as soon as price moves far enough from its own best point takes the loss before the technique itself notices the turn — and what remains is better than having no stop at all, provided the level adapts to volatility instead of being a fixed amount.

Measured in crypto — Binance spot · 540 pairs, delisted ones included · 0.2% per round trip

Net per trade

+2.27%

after fees

The fee is charged on both legs: every trade pays to open and pays to close.

Weakest point

placebo

survived all six controls (t=2.35 against a threshold of 2)

Worst drawdown

−45%

608 days underwater

How far the account fell below its own best previous moment.

Timeframe

1 day

The time grid this was measured on. The same technique on a coarser or finer grid is a different measurement, and can earn a different verdict.

invariance
costs
placebo
benchmark
out of sample
multiple testing

Equity curve

Risking 1.0% per trade, marked to market every day — not only when a trade closes. The line at 1.0 is the capital it started with.

At its worst the account was worth 45% less than its own best previous moment, and it spent 608 days below that peak.

The technique against chance

Same number of trades, same holding time, same assets — only the dates were drawn at random.

The average trade and the typical one

The mean sits above the median: the lottery signature — the typical trade loses and a rare handful pays for everything.

What was measured

We coded the rule exactly as it is described and let it trade on its own from 2017-08-17 to 2026-07-28. It found trades in 540 coins, 14,865 in total. Each one enters and exits at the price that existed on that day — the program never sees what comes next, which is the most common mistake people make testing strategies at home. Wherever the description was ambiguous we took the reading least favourable to the technique; whatever was left open is stated in the hypothesis, at the foot of this page.

How many of those trades actually count

Only 29 of the 14,865. Cryptocurrencies rise and fall almost in unison, so the same rule fires on dozens of coins on the same day — and that is one bet repeated, not dozens of different bets. Counting them all separately is the trick that makes a bad test look impressive.

What it paid, before any deductions

+2.47% per trade. This is the number the technique claims, and the only one on this page that has not yet been through a single control.

Why it passed

It survived every mundane explanation we tested:

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 would have happened to the money

Risking 1.0% of capital per trade — the most widely taught rule — across the whole period: The capital would have ended at 5.24× what it started with. At its worst the account was worth 45% less than its own best previous moment, and it spent 608 days below that peak. In none of the 12 paths tested did it fall to less than half of what it started with.

The account would not fit every signal

28% of the trades had to be turned down: by the time they appeared, the money was already tied up in other positions. That matters because the average return per trade the technique claims is computed over trades that nobody could have taken all of.

What this result does NOT say

Clearing the six controls is not a promise of future profit, and it is not advice to trade. It means only this: over this period and in this universe, we did not find the mundane explanation that knocked the others down. An edge that existed may be running out — this archive holds a technique that passed and whose most recent slice of time is already negative. The difference matters, and it is the rule of this house: the card confronts the claim, never the person who made it.

What this card does not measure

Every verdict holds for the conditions it was measured under. These are this card's — and outside them the result does not apply.

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 technique itself (held until the opposite signal). 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 14,865 trades, but only 29 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 29 episodes, what the data supports is a range from +0.38% to +4.56% 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	2.22	2.05	29
placebo	passed	2.35	2.05	29
benchmark	passed	—	—	—
out of sample	passed	—	—	—
multiple testing	passed	—	—	—

invariance14865 signals across 750934 barsconcentration: +38.18% of the profit sits in the top 5% of trades — with the dates shuffled, +37.83% (fails above +50.00%)
costsgross +2.467% · cost 0.200% · net +2.267% (t=2.22)
placeboactual +2.267% · placebo -0.251% · excess +2.518% ± 1.070% (t=2.35 against a threshold of 2.05, 29 real groups, 743,250 sham dates, draw error ±0.025%)
benchmarktechnique +2.27% · buy and hold (same horizon) +0.22% · excess +2.05%
out of sampleasset half A: +2.121% (t=2.02, 28 episodes) · asset half B: +2.415% (t=2.31, 29 episodes) · liquid half (>= US$ 2,066,613/day): +2.115% (t=2.38, 29 episodes) · illiquid half: +2.442% (t=1.44, 27 episodes) · period 1/4 (2017-10-11 a 2022-07-07): +5.715% (t=3.45, 16 episodes) · period 2/4 (2022-07-08 a 2023-12-20): +0.446% (6 episodes — too small, does not count) · period 3/4 (2023-12-21 a 2025-04-29): +1.833% (5 episodes — too small, does not count) · period 4/4 (2025-04-30 a 2026-07-27): +1.081% (5 episodes — too small, does not count)
multiple testing3 variation(s) tested · t=2.22 across 29 episodes (equivalent to t=2.13) · p≈0.0335 · false positives expected by chance ≈ 0.10

Equity — outside the six controls, and here is why

The t of the trade series is invariant to bet size: 0.5%, 1% and 3% agree to the sixth decimal. Nothing here moves the verdict — it moves what the account would have lived through.

risking 1.0% per trade: ×5.24
worst drawdown from the peak: −45%
days below the previous peak: 608
signals refused for lack of capital: 28%
paths where the account halved (out of 12): 0

Reproducibility

period: 2017-08-17 to 2026-07-28
assets that traded: 540
variations tested before this one: 3
gross per trade: +2.47%
net per trade: +2.27%
exit rule: the technique itself (held until the opposite signal)
median duration bars: 10
mean duration bars: 12.7
max duration bars: 84
fee per leg: 0.001
episode days: 112
seed: 20260728
fast average: 20
slow average: 50
dev f: 2.06
dev period: 20
base: Moving average crossover (20/50)

Binance spot klines (delisted pairs included) · collected from 2026-07-27 23:31 to 2026-08-03 10:49 · 540 assets · 750,934 bars · 2017-08-17 to 2026-07-28

Hypothesis, filed before the result

True range over 2 days; its mean and standard deviation over 20 periods; DDEV = mean + f × deviation; the stop for a long is the TRADE'S HIGH minus DDEV. The source gives three levels of f — 1, 2.06 to 2.25 and 3.20 to 3.50 —, and says the larger ones «correct the skew and allow more risk»: these are three risk appetites of the SAME rule, and all three will be measured, at the lower bound of each pair (1 · 2.06 · 3.20). This stop BACKS AWAY: measured over the base's trades, it loosens on 37.7% of the bars and on 99.2% of the trades, with a maximum retreat of 51.7%. The engine refuses a stop that loosens (`SetupComGestao`), and the source explicitly allows volatility stops to back away — so this enters as an EXIT RULE, not as a managed stop, which is how the source itself describes it («liquidated if the trend changes or the stop is touched, whichever comes first»). Family-level prediction, recorded before measuring: (1) the stop IMPROVES concentration and worsens the mean — it cuts the right tail along with the left, and control 1 is where the 20/50 crossover already dies; if any card in this family changes its verdict relative to the base, I expect it to be by clearing the invariant, not by gaining return; (2) the VOLATILITY stops will do better than the fixed percentage, because that is what the source claims («the ones most likely to work need to adapt to volatility, and not to a fixed value in money or percentage of price») and the claim is testable; (3) the 12 × average true range of 252 will almost never be touched — it is far too wide a stop for crypto, and will measure practically the same thing as the base with no stop; (4) the profit TARGETS will do worse than the stops, because the source's own table shows that the best result for the slow average was WITHOUT profit taking; (5) none survives the family's Benjamini-Hochberg.

filed on 2026-08-03, before the number existed

The original, as it was filed

Amplitude real de 2 dias; média e desvio-padrão dela em 20 períodos; DDEV = média + f × desvio; o stop da compra é o TOPO DA OPERAÇÃO menos DDEV. A fonte dá três níveis de f — 1, 2,06 a 2,25 e 3,20 a 3,50 —, e diz que os maiores «corrigem a assimetria e permitem mais risco»: são três apetites de risco da MESMA regra, e serão medidos os três, pelo limite inferior de cada par (1 · 2,06 · 3,20). ⚠️ Este stop RECUA: medido sobre as operações da base, ele afrouxa em 37,7% das velas e em 99,2% das operações, com recuo máximo de 51,7%. O motor recusa stop que afrouxa (`SetupComGestao`), e a fonte autoriza explicitamente que stops de volatilidade recuem — então esta entra como REGRA DE SAÍDA, não como stop gerido, que é como a própria fonte a descreve («liquidada se a tendência mudar ou o stop for tocado, o que vier primeiro»). Previsão da família, registrada antes de medir: (1) o stop MELHORA a concentração e piora a média — ele corta a cauda direita junto com a esquerda, e o controle 1 é onde o cruzamento 20/50 já morre; se algum card desta família mudar de veredito em relação à base, espero que seja por sair do invariante, não por ganhar retorno; (2) os stops de VOLATILIDADE irão melhor que o percentual fixo, porque é isso que a fonte afirma («os que têm mais chance de funcionar precisam se adaptar à volatilidade, e não a um valor fixo em dinheiro ou percentual do preço») e a afirmação é testável; (3) o 12 × amplitude real de 252 dias quase nunca será tocado — é um stop largo demais para cripto, e vai medir praticamente a mesma coisa que a base sem stop; (4) os ALVOS de lucro irão pior que os stops, porque a própria fonte mostra na tabela dela que o melhor resultado da média lenta foi SEM tomada de lucro; (5) nenhuma sobrevive ao Benjamini-Hochberg da família.

Quotations from the source were translated from the Portuguese record and back into English — they are not the author's exact words.

Pre-registration exists to keep prediction apart from rationalisation: written after the number, every hypothesis is right.

The same technique in the other market

The verdict is different there. The same technique, measured on independent data, came out the other way — and that says more about it than either card says alone. Read the cost and the detection floor before concluding: they are not the same in the two markets.

cryptoPASSED← this one
forexFAILEDopen that card

Earlier audits of the same technique

Each variation an author teaches enters as its own test, so that whatever might work in the strategy gets covered. The verdict held in all of them.

this measurement →PASSED
2026-08-03PASSEDopen ↗