Crypto Divergences is a signal desk, not a new oscillator. The useful context is the work it sits next to: Wilder and Appel’s indicators, closed-candle scanners, TradingView studies, and research-style mark-to-market books.
The oscillators
- RSI (Wilder, 1978) — relative strength as an average-gain / average-loss ratio, 14 periods by default. Divergences are read on the RSI line vs swing highs/lows in price.
- MACD (Appel) — 12/26 EMA difference, 9-period signal. Histogram divergences are the scanner’s second pass.
- Stochastic, CCI, OBV — also valid disagreement tools. They show up on this desk when a print names them; they are not in the mechanical scanner yet.
TradingView’s built-in RSI and MACD studies are the closest interactive cousin: same math family, live pane, no frozen public tape. The pair pages here embed that widget so you can confirm a row without leaving the site.
Closed candles vs repainting studies
Many “live divergence” scripts fire on the forming bar and disappear when the bar closes the other way. That is fine for a discretionary overlay. It is a poor source of a public accuracy book. This scanner waits for pivot confirmation on both sides of the extreme. Related project in spirit: any non-repainting pivot indicator that only emits after right bars exist.
Mark-to-market vs TP/SL engines
VectorBT, Backtrader, Jesse, and Freqtrade all want an entry rule plus optional stops. Our score is simpler on purpose: one snapshot in, one snapshot out, signed by side. That is how you compare a heterogeneous tape (desk + scanner, many TFs) without inventing a shared stop model nobody agreed to.
If you want the other kind of number — “would a 1.5R take-profit have hit before a 1R stop” — export the CSV and run it in one of those engines. The desk will not pretend it already did.
What to read next here
- Understanding market divergences — the short primer.
- Regular vs hidden — geometry and live hit rates by kind.
- The scanner — how the mechanical pass is wired.
- Accuracy desk — the live book.