Relative Strength Leaders/Laggards

Methodology

Returns. The universe is the S&P 500 plus SPY. Yahoo Finance daily adjusted closes for the trailing ~100 trading days provide split- and dividend-adjusted closes. Trailing return runs from the close 21 trading days ago (1-month) or 63 trading days ago (3-month) to the latest close. Relative (excess) return is (1 + stock return) / (1 + SPY return) − 1, not a simple subtraction, so it compounds properly at larger magnitudes. Small Cap vs. Large Cap Spread and International vs. US use the same construction.

A stock is tagged "Leader" if its 3-month relative return is in the top fifth of the universe that week, "Laggard" if it's in the bottom fifth, and "Neutral" otherwise. Ranking uses the 3-month lookback, not 1-month. The 1-month figures are shown for context but are noisier and more prone to single-event whipsaws (an earnings beat, a guidance cut).

Persistence. The chart above tests whether this week's relative-strength rank says anything about what happens next. Comparing one week's 3-month rank with next week's 3-month rank would be misleading. Two 63-trading-day windows one week apart share 58 of those days, so their ranks would correlate almost mechanically, regardless of any momentum effect (the same overlapping-window problem noted on Share Count Trends). Instead, each point pairs a stock's rank at one weekly snapshot with its realized relative return over the following snapshot interval, a forward window built from entirely new trading days. The data is pooled across every stock and every consecutive pair of snapshots (up to 20 weekly snapshots, roughly five months). A positive, significant Pearson/Spearman relationship would mean today's leaders tend to keep outperforming. A null or negative one would mean the opposite: that relative strength mean-reverts, or says nothing about the future at this horizon. Both a linear (Pearson) and a rank-based (Spearman) fit are shown, since one can mislead when the relationship is nonlinear or driven by outliers.

Refreshes weekly (Saturday, after that morning's other data jobs). A 21/63-trading-day momentum reading barely moves day to day, and the persistence test needs weekly snapshots spaced apart. Sector and company name come from Sector Beeswarm. The persistence test is built from accumulated weekly snapshots, so it shows "not enough history yet" for the first week or two.