Price of the chosen index against the delivered value of a set of stocks, and how that delivery splits between the stocks' up days and down days. Statistics below show whether the two move together, and whether delivery has said anything about the index's next weeks.
With one index the top area shows its actual price; with comparisons all prices are rebased to 100 at the start of the period so they share one axis (moving averages are rebased the same way). NSE switched index P/E to consolidated earnings from April 2021 (dotted line), so readings before and after that date are not strictly comparable. The dotted horizontal line is the first index's median valuation over the period shown, read on the right axis.
Each Friday the portfolio builder's top 15 (as shown on that day, from time travel) are held to the next Friday, equal weight, before costs and tax. This is a gauge of how momentum is behaving, not an achievable return: weekly rebuilding would lose a large share to costs and short-term tax (the tested rule, every 4th Friday after costs and tax, is far lower). Both lines are rebased to 100 at the start of the period; the bars show the portfolio's own return in each week (right axis). The table tests whether the portfolio's recent behaviour said anything about the index over the following weeks. Week-by-week holdings: weekly picks grid; every universe stock: universe grid.
Up day / down day = the stock's own close vs its previous close (adjusted for splits and bonuses). Universe membership is point-in-time (the week's list), from Oct 2020; constituents are today's lists. Statistics use data up to day t to describe what happened from t+1. With overlapping 20-day windows the effective sample is small, so treat |t| below about 2.5 as noise.