FAILED

Gate by the equity curve trend (10 days, 80-day average)

A trend system's stream of results has a trend of its own: good and bad stretches come in blocks, not at random. Applying to the accumulated equity the same moving average you apply to price lets you stop trading when the curve turns down and resume when it turns up — avoiding the bad runs without predicting anything about the market.

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

Net per trade

+0.74%

after fees

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

Died at

invariance

the result is not physically plausible

Worst drawdown

−67%

1,501 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 67% less than its own best previous moment, and it spent 1,501 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, 57,833 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 57,833. 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

+0.94% 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 did not pass — it died here · where the result came from

We look at where the result came from: whether it is spread across the trades or packed into a handful of them. It concentrated. That is the shape of a lottery: the typical trade loses, and a rare handful pays for everything. Anyone following the rule for real would face a long run of losses before any gain, and would almost certainly quit before it arrived. To give the measure: 58% of the profit came from the top 5% of trades, and the same technique with its dates shuffled would concentrate 42%. The control fails above 50%.

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 3.15× what it started with. At its worst the account was worth 67% less than its own best previous moment, and it spent 1,501 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

47% 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

For an edge to be assertable here it would have to reach 4.54% per trade — that is the size that survives this archive's multiple-testing correction, and it rises as the archive grows. So FAILED means “we found nothing above that size”, and never “it cannot possibly work”. The instrument is far more sensitive than that: on synthetic data, with a clean effect, it separates from 0.69% upwards. The distance between the two numbers is the price of a real market and the price of publishing many claims. 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 57,833 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 -1.13% to +3.01% 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	failed	—	—	—
costs	inconclusive	0.73	2.05	29
placebo	inconclusive	0.75	2.05	29
benchmark	passed	—	—	—
out of sample	passed	—	—	—
multiple testing	inconclusive	—	—	—

invariance58% of the gross profit comes from 2891 trades (5% of the total) — lotteryconcentration: +58.00% of the profit sits in the top 5% of trades — with the dates shuffled, +41.79% (fails above +50.00%)
costsgross +0.941% · cost 0.200% · net +0.741% (t=0.73) · the range runs from -1.330% to +2.811%
placeboactual +0.741% · placebo -0.030% · excess +0.770% ± 1.022% (t=0.75 against a threshold of 2.05, 29 real groups, 2,891,650 sham dates, draw error ±0.011%)
benchmarktechnique +0.74% · buy and hold (same horizon) -0.01% · excess +0.75%
out of sampleasset half A: +0.752% (t=0.70, 29 episodes) · asset half B: +0.729% (t=0.76, 29 episodes) · liquid half (>= US$ 2,066,613/day): +0.929% (t=0.94, 29 episodes) · illiquid half: +0.535% (t=0.34, 28 episodes) · period 1/4 (2017-11-09 a 2022-06-14): +3.595% (t=1.94, 15 episodes) · period 2/4 (2022-06-15 a 2023-11-22): -0.933% (6 episodes — too small, does not count) · period 3/4 (2023-11-23 a 2025-05-03): +1.300% (6 episodes — too small, does not count) · period 4/4 (2025-05-04 a 2026-07-27): -0.982% (5 episodes — too small, does not count)
multiple testing1 variation(s) tested · t=0.73 across 29 episodes (equivalent to t=0.70) · p≈0.4833 · false positives expected by chance ≈ 0.48

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: ×3.15
worst drawdown from the peak: −67%
days below the previous peak: 1,501
signals refused for lack of capital: 47%
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: 1
gross per trade: +0.94%
net per trade: +0.74%
exit rule: the technique itself (held until the opposite signal)
median duration bars: 3
mean duration bars: 6.1
max duration bars: 85
fee per leg: 0.001
episode days: 112
seed: 20260728
equity curve average: 10
slope period: 80

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

The central claim of `Equity Trends` (p. 1086), in the fast variant. The rule, given in full: «by applying a simple moving average to the cumulative profits and losses, we could exit positions when the equity trend is down»; and the implementation, «when the equity trend was up on day t compared to t−1, we used the returns of day t+1; otherwise, we used a return of zero». An 80-day base, a 10-day average over cumulative profit, one unit per signal — «only 1 contract was traded on each signal, rather than a risk-adjusted position size». THE CLAIM IS NOT THE ONE THE RUNBOOK RECORDED. The runbook said «stop trading after N losses, come back when equity crosses its own average»; THERE IS NO COUNT OF LOSSES anywhere in the section. The rule is the daily gate above, and that is what will be audited. THE SOURCE ALREADY REPORTS THAT IT DOES NOT WORK: «none of the three cases show consistency, although there were improvements for the S&P and gold using both the 10- and 60-day trends. Bonds, the most trending market, showed significant deterioration with this method». And it lists three defects of its own test: «the equity trend method also did not include any switching costs, which would be more significant for the 10-day trend»; «we can infer that this method would perform poorly on a diversified portfolio»; and «applying this to hypothetical equity patterns is not likely to work, because hypothetical results are always better than actual trading results». THE PRECONDITION IT DECLARES NECESSARY, THIS CORPUS HAS ALREADY MEASURED AND NOT FOUND: «you first need to find out if returns are positively autocorrelated, which means that there is persistence in positive or negative returns» — and the entry in `confirmacoes.py` publishes, over 742,175 crypto days, that daily direction is a coin flip with no memory. Family-level prediction, recorded before measuring: (1) NONE survives the corpus-wide Benjamini-Hochberg; (2) the FRACTIONAL ENTRY and the two PYRAMIDS ON PROFIT will move the mean by LESS THAN THE ERROR BAR and will NARROW the deviation — it is the mechanism the protocol has already measured twice, at breakeven (median from −2.75% to 0.00%, mean unchanged) and at partial profit taking (−0.023% ± 0.518%, deviation from 19.9% to 12.4%): position management moves the SHAPE and is read as if it had moved the CENTRE; (3) the EQUITY CURVE GATE fails, and for lack of autocorrelation — this is the family's most falsifiable prediction, because the number already exists in the corpus: the source states that «first one must find out whether returns are positively autocorrelated», and `confirmacoes.py` publishes, over 742,175 crypto days, that daily direction is a coin flip WITH NO MEMORY; (4) AVERAGING DOWN will be the worst in the family and will die at CONTROL 1, concentration, because adding to losses fattens the tail — and the source itself already rejects it in 3 of 4 markets; (5) the gravedigger in crypto will be CONTROL 1 and not the benchmark, which is the correction of the prediction refuted in part 76 and what part 74 had already established: in crypto, what pays a lot pays it concentrated; (6) the 80-day BASE will do better than any layer built on top of it.

filed on 2026-08-04, before the number existed

The original, as it was filed

A alegação central de `Equity Trends` (p. 1086), na variante rápida. A regra, dada por inteiro: «aplicando uma média móvel simples aos lucros e prejuízos acumulados, poderíamos sair de posições quando a tendência do patrimônio está de baixa»; e a implementação, «quando a tendência do patrimônio estava de alta no dia t comparado a t−1, usamos os retornos do dia t+1; caso contrário, usamos um retorno de zero». Base de 80 dias, média de 10 dias sobre o lucro acumulado, uma unidade por sinal — «apenas 1 contrato era operado a cada sinal, em vez de um tamanho de posição ajustado ao risco». ⚠️ A ALEGAÇÃO NÃO É A QUE O RUNBOOK REGISTRAVA. O runbook dizia «pare de operar depois de N perdas, volte quando o capital cruzar a própria média»; NÃO HÁ CONTAGEM DE PERDAS em lugar nenhum da seção. A regra é o gate diário acima, e é ela que será auditada. ⚠️ A FONTE JÁ REPORTA QUE NÃO FUNCIONA: «nenhum dos três casos mostra consistência, ainda que tenha havido melhoras para o S&P e o ouro usando tanto a tendência de 10 quanto a de 60 dias. Os títulos, o mercado com mais tendência, mostraram deterioração significativa com este método». ⚠️ E ela lista três defeitos do próprio teste: «o método da tendência do patrimônio também não incluiu nenhum custo de troca, que seria mais significativo para a tendência de 10 dias»; «podemos inferir que este método teria desempenho ruim numa carteira diversificada»; e «aplicar isto a padrões hipotéticos de patrimônio provavelmente não vai funcionar, porque resultados hipotéticos são sempre melhores que resultados de operação real». ⚠️ A PRÉ-CONDIÇÃO QUE ELA DECLARA NECESSÁRIA, ESTE CORPUS JÁ MEDIU E NÃO ENCONTROU: «primeiro é preciso descobrir se os retornos são positivamente autocorrelacionados, o que significa que há persistência em retornos positivos ou negativos» — e o verbete de `confirmacoes.py` publica, em 742.175 dias de cripto, que a direção diária é cara-ou-coroa sem memória. Previsão da família, registrada antes de medir: (1) NENHUMA sobrevive ao Benjamini-Hochberg do corpus; (2) a ENTRADA FRACIONADA e as duas PIRÂMIDES SOBRE LUCRO moverão a média por MENOS QUE A BARRA DE ERRO e ESTREITARÃO o desvio — é o mecanismo que o protocolo já mediu duas vezes, no breakeven (mediana de −2,75% para 0,00%, média parada) e na realização parcial (−0,023% ± 0,518%, desvio de 19,9% para 12,4%): gestão de posição mexe no FORMATO e é lida como se tivesse mexido no CENTRO; (3) o GATE PELA CURVA DE PATRIMÔNIO reprova, e por falta de autocorrelação — esta é a previsão mais falsificável da família, porque o número já existe no corpus: a fonte declara que «primeiro é preciso descobrir se os retornos são positivamente autocorrelacionados», e `confirmacoes.py` publica, em 742.175 dias de cripto, que a direção diária é cara-ou-coroa SEM MEMÓRIA; (4) a MÉDIA PARA BAIXO será a pior da família e morrerá no CONTROLE 1, concentração, porque adicionar nas perdas engorda a cauda — e a própria fonte já a reprova em 3 de 4 mercados; (5) o coveiro em cripto será o CONTROLE 1 e não o benchmark, que é a correção da previsão refutada na parte 76 e o que a parte 74 já tinha estabelecido: em cripto, o que rende muito rende concentrado; (6) a BASE de 80 dias irá melhor que qualquer camada sobre ela.

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.

cryptoFAILED← this one
forexINCONCLUSIVEopen that card

This claim has been audited once — there is no history to compare against.