BATS COMPONENT · 5-DAY RATE OF CHANGE
How fast did the market just move?
The simplest possible momentum measure: today’s close on the benchmark index divided by the close 5 trading days ago, expressed as a percent. A one-week snapshot of speed and direction. When the market crashes hard for a week, ROC-5 goes deeply negative. When it rips for a week, ROC-5 spikes positive. Most days it hovers within a fraction of a percent of zero.
The recipe
For each trading day, take the index close and the close from five trading days earlier:
- ROC-5:
(closetoday / close5 days ago − 1) × 100
Why 5 days? It’s one trading week — short enough to react to real-time panic and euphoria but long enough that a single outsized day doesn’t dominate the reading. It’s also genuinely independent from our RSI-14 component roughly 40% of the time at the extremes (see the backtest below), so it adds signal rather than noise.
How it maps to a BATS score
We clamp ROC-5 at ±6% (roughly the 1st and 99th percentiles of daily readings since 1990) and linearly map that band to a 0–100 BATS score:
- ROC-5 ≤ −6% → score 0 (extreme bearish — a week’s worth of crash)
- ROC-5 = 0% → score 50 (neutral)
- ROC-5 ≥ +6% → score 100 (extreme bullish — a week’s worth of melt-up)
What the backtest showed (1990–2026)
Both tails deliver roughly 2× the baseline 12-month S&P forward return — the classic contrarian U-shape:
- ROC-5 < −6% (crash tail): +21.2% average 12M return
- Baseline (all days): ~+10.5% average 12M return
- ROC-5 > +6% (rally tail): +20.9% average 12M return
The interesting part is the disagreement with RSI. ROC-5 has a Pearson correlation of r = +0.596 with our SPY 14-day RSI — related, but far from redundant. On days where ROC-5 flagged an extreme crash (< −6%), RSI-14 was still above 30 in 59% of those cases. That’s the edge: a fast one-week collapse can happen from a starting RSI in the 40s or 50s, and ROC-5 catches it before the slower RSI does.