Weekly Outlook

What this week’s tape is actually saying.

Six indices. 5 years of weekly bars. We classify the current regime, find the K-most-similar historical weeks, and show what happened next. Every number on this page is reproducible from the source candles. No opinions.

Math QA passed Probability bands, not price targets 5-year history per index

How the probability cone is built

For every weekly bar in the last 5 years we compute a feature vector: RSI(14), distance from 50-week and 200-week SMA, ATR as a percent of close, and 4-week momentum. We z-normalise the features so each one weights equally, then find the K nearest historical weeks to today’s setup using Euclidean distance. The +1-week and +4-week returns of those K neighbours form the distribution you see in the cone.

Why this is honest forecasting

We never claim a single price. The cone shows the 10th, 25th, 50th, 75th and 90th percentile of what happened in similar historical setups, plus the hit rate (fraction that closed up). If the sample size is small, you see it. If the band is wide, you see it. Markets are conditional probability, not arithmetic.

Indicator definitions

  • RSI(14) — Wilder’s relative strength index over the last 14 weekly bars. >70 overbought, <30 oversold.
  • ATR(14) — Wilder-smoothed weekly Average True Range. Shows expected weekly range.
  • %B — position within the 20-week Bollinger Band (2σ). 0 = lower band, 1 = upper.
  • Pivots — classical floor-trader pivots from the prior weekly bar’s H/L/C.