How the crypto short signals work, and the full backtest

Get Short Signals watches USDT perpetual futures on Binance, Bybit and Gate.io for coins that have pumped hard and started to reverse. This page gives the exact rules, the grades, the exit plan and the backtest behind them, including what the backtest does not support. Last updated 16 November 2026.

The strategy in one paragraph

A coin qualifies when its hourly close is at least 60% above its prior 24-hour low on heavy volume. A short signal appears only after the reversal is confirmed: price at least 10% off the high, two hourly closes below the 20-period EMA, and fading volume. The signal is then graded by the funding rate. Setups where funding is at or below the 0.01% baseline were profitable in the backtest; setups where funding is above it lost money and get no trade plan.

The rules

Universe
USDT crypto perpetuals on Binance, Bybit and Gate.io. If a coin has no perpetual, there is nothing to short. A pump listed on several exchanges is shown once, on Binance first.
Pump
Hourly close at least 60% above the prior 24-hour low, 24-hour turnover of $2 million or more, and 24-hour volume at least 3 times the prior 7-day daily average.
Reversal confirmation
Price at least 10% below the highest price of the pump, the last two hourly closes below the 20-period EMA, and recent volume under 70% of the volume into the peak. A new high cancels the signal and re-arms it.
Entry window
Within 3 hours of confirmation and not more than 5% below the confirmation price. After that the setup is marked Late and should not be chased.
Position sizing
Size the trade so that the stop costs a fixed fraction of capital, between 0.25% and 1%. With a 20% stop that is 1.25% to 5% of capital as notional. Leverage is not needed, and it brings liquidation before the stop.

Grades

GradeConditionsBacktest result
A Confirmed reversal, funding at or below 0.01%, Bitcoin not up more than 1% on the day, first signal on this pump 78 trades, 63% wins, +3.3% per trade, profit factor 1.52
A + B Confirmed reversal with funding at or below 0.01% (B: Bitcoin is rising, or an earlier signal on the same pump already failed) 118 trades, 62% wins, +3.2% per trade, profit factor 1.49
Skip Confirmed reversal with funding above 0.01%: leveraged longs are still paying to chase 47 trades, 38% wins, −2.4% per trade. No trade plan is shown.

Funding is the latest settled rate, per funding interval, exactly as the backtest measured it. Most pumped perpetuals settle every 1 or 4 hours; scaling the rate to an 8-hour equivalent was tested and performed worse, so it is not used.

Stop, profit targets and time stop

Thirty-six stop and target pairs were tested on the 118 funding-gated trades. A 20% stop gave the best return per unit of risk at every target: tighter stops were shaken out, wider ones lost too much when wrong. The best target depends on how violently the coin moves.

Coin profileStopTarget 1 (half)Target 2 (rest)Backtest result
Calm: average hourly range under 10% of price +20%−20%−30% 58 trades, 66% wins, +2.7% per trade, profit factor 1.46
Wild: average hourly range 10% or more +20%−20%−50% 60 trades, 58% wins, +3.6% per trade, profit factor 1.53

Whatever is still open after 48 hours is closed at market. The wild-coin result was positive in both halves of the data. The calm-coin result came mostly from the held-out month (+0.9% in development, +7.4% held out), so treat it as weaker.

The backtest

  • Data: 15-minute candles and funding history for 654 Binance crypto perpetuals, including delisted ones, from 17 August to 15 November 2026 (90 days). Bitcoin is excluded as the market benchmark.
  • Costs: 0.30% per round trip plus the actual funding paid or received.
  • Split: the first 60 days were used to choose the rules and the last 30 days were held out. Grade A made +2.4% per trade in development and +5.3% held out.
  • Fills: entry at the next bar's open after confirmation. If a stop and a target fall inside the same 15-minute bar, the stop is assumed to be hit first.
  • Frequency: about 7 grade A signals a week across the whole market.

What the backtest does not support

  • The confirmation rules alone are not an edge. Without the funding gate, the same rules made +1.6% per trade over 165 trades (55% wins, profit factor 1.21), which is not statistically different from zero.
  • Smaller pumps do not work. At a 30% to 40% pump threshold the same rules lose money.
  • Shorting without confirmation is dangerous. After a 60% pump the median coin is 12% lower in 24 hours and 21% lower in 48 hours, but the average short makes far less, because a minority keep running 100% or more against the position.
  • Stops slip. The worst single trade in every group lost about 25%: in a fast market a stop fills beyond its 20% level.
  • The sample is small. One 90-day window, 78 grade A trades, t about 1.7 to 1.9. The edge can shrink or reverse.
  • Only Binance was tested. Grades on coins listed only on Bybit or Gate.io use the same rules, untested on those exchanges.

Questions and answers

What is a pump-exhaustion short signal?

A signal to short a coin that has risen at least 60% from its prior 24-hour low and has then started to reverse: price at least 10% below the pump's high, the last two hourly closes below the 20-period EMA, and recent volume under 70% of the volume that drove the peak.

Why does the funding rate decide the grade?

In the backtest, confirmed reversals with settled funding at or below the 0.01% baseline made +3.2% per trade (118 trades, 62% wins, profit factor 1.49). The same setups with funding above the baseline lost 2.4% per trade (47 trades, 38% wins). High funding means leveraged longs are still paying to chase the move, which is continuation rather than exhaustion.

How accurate are the signals?

Grade A signals won 63% of the time in the backtest: 78 trades over 90 days, +3.3% per trade after costs, profit factor 1.52. The edge is modest, comes from a single 90-day window and is not statistically conclusive. It can shrink or reverse.

What stop loss and profit target do the signals use?

The stop is 20% above entry. Half the position is closed at 20% profit. The rest is closed at 30% profit for calmer coins or 50% for wild coins, and anything still open is closed after 48 hours.

Which exchanges are covered?

USDT perpetual futures on Binance, Bybit and Gate.io. The backtest used Binance data only, so grades on coins listed only on Bybit or Gate.io are indicative.

Is this financial advice?

No. It is a research tool. Shorting pumped coins has unlimited loss potential. Never trade it without a hard stop and risk-based position sizing.

This is a research tool, not financial advice. The edge above is modest, comes from one 90-day window and is not statistically conclusive; it can vanish or reverse.

← Back to live signals