← 返回 Quant Digests · 站点首页

别把高 VPIN 当 15m 方向确认:它更像 breakout-short / Fib / EMA-PSAR 的 two-speed jump-risk overlay

更新时间:2026-03-23 08:59 UTC 研究时间:2026-03-23 05:10 UTC 类型:近 5 年论文 + 开源实现仓库 + Binance 公共数据最小快检 主题标签:breakout-short/final-verdict/fibonacci/retest-hold/ema/psar/vpin/order-flow-toxicity/jump-risk/regime-gate/position-sizing/risk-overlay/crypto/5m/15m 证据类型:论文证据 + 工程实现 + 公共数据快检

源文件:research/quant_digests/2026-03-23_0510_vpin-jump-toxicity-twospeed-overlay.md

1. 这次看了什么

这次主读的是 Kitvanitphasu, Kyaw, Likitapiwat, Treepongkaruna (2025/2026):*Bitcoin wild moves: Evidence from order flow toxicity and price jumps*。它最值得我们 desk 借的旁支,不是“再造一个入场 alpha”,而是把 order-flow toxicity (VPIN) 定位成 jump-risk 的 regime 层。实现参考用的是开源仓库 hanxixuana/flowrisk(VPIN 递归估计器)。

2. 核心结论

3. 为什么和当前项目有关

3.5 策略拆解(必填)

4. 可复刻的最小实验

  1. Binance USDⓈ-M fapi/v1/klines 1m(含 taker buy volume,公开 REST);
  2. Binance USDⓈ-M fapi/v1/klines 15m(事件与绩效评估);
  3. 更新频率:1m/15m 持续更新,免费可拉取。
  1. 用 1m taker_buy_volumesell_volume 做 volume-bucket VPIN(bucket≈日均成交量/50,VPIN窗=50 buckets);
  2. 映射到 15m 后按滚动分位分成 low(<q40) / mid(q40~q80) / high(>q80)
  3. 把三条策略现有触发分别跑三组:base vs base+binary(high veto) vs base+two-speed sizing
  1. post-cost short/long expectancy(按方向拆开,不合池);
  2. N-bar failure rate(breakout 回收、Fib 失守、EMA/PSAR 假翻转)。

5. 风险与保留意见

6. 来源

  1. Kitvanitphasu, A., Kyaw, K., Likitapiwat, T., & Treepongkaruna, S. (2025). *Bitcoin wild moves: Evidence from order flow toxicity and price jumps*. Research in International Business and Finance (2026 print, Vol.81, 103163).
  2. DOI: 10.1016/j.ribaf.2025.103163 Readable URL: https://doi.org/10.1016/j.ribaf.2025.103163 Repo URL: N/A(论文未给官方代码仓库)

  1. Han, X. (hanxixuana). (2018, updated). *flowrisk: A Python Implementation of Measures for Order Flow Risk, e.g. VPIN*. GitHub Repository.
  2. Readable URL: https://github.com/hanxixuana/flowrisk Repo URL: https://github.com/hanxixuana/flowrisk

  1. Easley, D., López de Prado, M., & O’Hara, M. (2012). *Flow toxicity and liquidity in a high-frequency world*. Review of Financial Studies, 25(5), 1457–1493.
  2. DOI: 10.1093/rfs/hhr100 Readable URL: https://doi.org/10.1093/rfs/hhr100

  1. Binance Developers. (n.d.). *USDⓈ-M Futures API – Kline/Candlestick Data*.
  2. Readable URL: https://developers.binance.com/docs/derivatives/usds-margined-futures/market-data/rest-api/Kline-Candlestick-Data