我们的形态识别不是坊间传说。这里列出引擎识别的每种形态的实测胜率和期望值——完全按照产品实际采用的入场、止损和目标设置模拟,基于多家交易所约一年的真实市场数据。

产品以示例数据展示。
全部 20 种可交易形态,基于 5 家交易所的 48,703 个已确认信号测量。每种形态都按其实际使用的盈利目标计分——推动形态为 2R,反转形态为 1R。带星标的行同时满足我们的两项置信度门槛:至少 1,000 个信号并且期望值至少为 +0.50R。
| # | 形态 | 方向 | 目标 | 胜率 | 期望值 | 交易数 |
|---|---|---|---|---|---|---|
| 1 | 看跌动能★ | 看跌 | 2R | 62.9% | +0.81R | 1,145 |
| 2 | 看跌光头光脚线★ | 看跌 | 2R | 61.8% | +0.73R | 2,374 |
| 3 | 三只乌鸦★ | 看跌 | 2R | 60.0% | +0.70R | 1,672 |
| 4 | 看涨光头光脚线★ | 看涨 | 2R | 56.2% | +0.63R | 3,119 |
| 5 | 三外升 | 看涨 | 1R | 81.3% | +0.61R | 158 |
| 6 | 三外降 | 看跌 | 1R | 81.9% | +0.59R | 258 |
| 7 | 红三兵★ | 看涨 | 2R | 54.5% | +0.58R | 1,558 |
| 8 | 看跌吞没 | 看跌 | 1R | 77.8% | +0.54R | 523 |
| 9 | 放量突破 · 看跌★ | 看跌 | 2R | 53.1% | +0.51R | 1,079 |
| 10 | 放量突破 · 看涨★ | 看涨 | 2R | 51.3% | +0.50R | 1,374 |
| 11 | 看涨动能 | 看涨 | 2R | 50.7% | +0.49R | 1,742 |
| 12 | 看涨吞没 | 看涨 | 1R | 71.8% | +0.43R | 849 |
| 13 | 三内降 | 看跌 | 1R | 71.4% | +0.41R | 514 |
| 14 | 平头底 | 看涨 | 1R | 69.3% | +0.36R | 1,084 |
| 15 | 三内升 | 看涨 | 1R | 68.4% | +0.35R | 1,103 |
| 16 | 看跌孕线 | 看跌 | 1R | 66.9% | +0.33R | 3,844 |
| 17 | 射击之星 | 看跌 | 1R | 66.4% | +0.32R | 4,476 |
| 18 | 平头顶 | 看跌 | 1R | 65.6% | +0.30R | 1,163 |
| 19 | 看涨孕线 | 看涨 | 1R | 63.7% | +0.26R | 6,643 |
| 20 | 锤子线 | 看涨 | 1R | 60.9% | +0.21R | 5,689 |

一套固定且完全机械化的规则——没有事后诸葛亮,也没有刻意挑选数据——并且与信号触发时产品为您给出的入场、止损和目标设置完全相同。
R 是您每笔交易的风险——在这里是从入场到止损的 1 个 ATR。以 R 衡量,可以让所有形态在不同价格和周期下都具有可比性。
期望值不等于盈亏比。这些交易机会的目标是 1R 或 2R;期望值是把所有亏损交易都计入平均值之后剩下的部分。+0.63R 的期望值表示每笔交易平均盈利为其风险金额的 0.63 倍——这是一个会持续累积的数字,而不是每笔交易的固定收益。
正因如此,盈亏平衡点远低于 100%:2R 的交易方案只需 33.3% 的胜率,1R 的交易方案只需 50%。在全部样本中,这些形态有 62.5% 的时间以盈利了结,平均每笔 +0.41R。
风险为 1 个 ATR,因此会随周期增大,而 R 倍数保持不变。请根据您愿意给交易留出的波动空间选择合适的周期。
| 周期 | 典型止损距离 | 反转目标(1R) | 推动目标(2R) |
|---|---|---|---|
| 1h | 1.21% | 1.21% | 2.42% |
| 4h | 2.43% | 2.43% | 4.86% |
| 8h | 3.76% | 3.76% | 7.52% |
| 12h | 4.58% | 4.58% | 9.16% |
| 1d | 6.65% | 6.65% | 13.30% |
| 3d | 12.16% | 12.16% | 24.32% |
Past performance is not indicative of future results. The figures on this page are historical measurements of our pattern-detection engine over a fixed data set of past market data, calculated with the mechanical rules described above. They assume entry at the trigger price and include no fees, funding or slippage, so they are an upper bound on what the same rules would have returned in practice; the ranking between patterns is more meaningful than any single absolute figure. The sample covers roughly twelve months of crypto markets — one regime, not a full cycle. These numbers are not a prediction of future returns, not a guarantee, and not financial advice. Trading involves substantial risk of loss. Always do your own research and trade only what you can afford to lose.