What this shows: an approximate Cumulative Volume Delta (CVD), normalized by market cap and expressed as a blended score, reset each trading day. This is an approximation, not true tick-by-tick CVD — it uses only each day's own open, close, and total volume, not a trade-by-trade breakdown of buys versus sells against the bid/ask. Treat it as a same-day sentiment proxy, not an exact order-flow measurement. Rows from before this open-vs-close definition was introduced were computed against the prior day's close instead, and were not retroactively recomputed — so very early history is not directly comparable to newer rows. See the CVD5 tab for a trailing 5-day view.

What this shows: a weighted average of a ticker's trailing 5 trading days of CVD score, rescaled to sit in a similar range to a single day's score × 5, shown for the last 5 trading days — the most recent day counts for more than older ones. Shown as "—" if any of the 5 needed days is missing data, rather than averaging an incomplete window and presenting it as a full 5-day result. See the CVD tab for the single-day figure and full formula.

What this shows: today's snapshot for each ticker — CVD5, closing price, the 50-day simple moving average (SMA50), and a signal that flags divergence between money flow and price trend. SMA50 is the average of the 50 most recent trading-day closes (today included). A ticker shows "—" for SMA50, CVD5, or the signal if any of the three isn't available yet, rather than guessing from an incomplete picture. This is a simple rule you defined, not investment advice, and it describes what already happened, not what will happen next.

Price/SMA50 = (closing price ÷ SMA50) × 100, one decimal place — e.g. 103.7 means price is currently 3.7% above its 50-day average; 96.2 means 3.8% below it.

Sorting: every column is independently sortable. Sorting by CVD5 breaks ties using Signal (BUY, then WATCH, then SELL); sorting by Signal breaks ties using CVD5 (highest first) — so within a CVD5 sort, BUY-flagged names surface first, and within a Signal sort, the strongest CVD5 names surface first inside each group. The "CVD5 > 0 only" checkbox filters the table before sorting, to the same effect either way.

What this shows: each tracked ticker ranked by its trailing 6-month price return — a momentum/relative-strength ranking, named for Jegadeesh & Titman (1993), "Returns to Buying Winners and Selling Losers," the academic paper documenting that stocks with strong trailing 3-12 month returns tend to keep outperforming over the following 3-12 months. CVD5 and SMA50 are the same figures shown on the Signals tab — repeated here so momentum and short-term money flow/trend can be read side by side for the same ticker, without switching tabs. Left as "—" if this ticker doesn't yet have enough history on file for that particular column (a brand-new listing, or the first day after this tab was added — history isn't backfilled).

The box above the table shows the benchmark ticker's own price and 6-month return, fixed in place regardless of how the table below is sorted, searched, or how long it is — so you always have it in view rather than needing to scroll and find it as a row.

The "vs OEF" column is each ticker's 6-month return minus the benchmark's — positive means that ticker beat the benchmark over the trailing 6 months, negative means it lagged. Sort by this column to rank the whole roster by relative strength against the benchmark directly, rather than comparing two separate numbers by eye.

The benchmark ticker is OEF by default (an S&P 100 proxy ETF) — add it from "Manage tickers" under the CVD tab if you haven't already; until it's in the roster and has data, the box above stays hidden and "vs OEF" shows "—" everywhere.

This ticker list is this app's own roster (Manage tickers, under the CVD tab), not the official S&P 100 index constituent list. It's a reasonable large-cap starting set, not a verified, live membership feed — add or remove tickers there any time.

What this is not: Jegadeesh & Titman's original strategy ranks a large universe (hundreds+ of stocks), holds long the top decile and short the bottom decile, and rebalances periodically — applying it to a roster this size, long-only, is a much smaller-sample version of the idea, not a reproduction of the paper's tested strategy. This tab is a ranking, not a buy/sell signal.

How this works: every morning at 7:00 AM IST (Tuesday to Saturday), a Netlify function pulls the previous US session's closing prices, asks Claude (with web search) for an updated allocation, and saves it here. Every change from the previous view is listed with its reason; a move of 5 percentage points or more on any position is flagged as a major change. Paper portfolio: starts at $10,000 on the starting allocation and follows every view at that day's closing price. It assumes no trading costs, no taxes and perfect fills, so it will look better than real trading. Compared with $10,000 put into OEF on the same day.

Your actual portfolio: built only from trades entered below, starting from $10,000 of capital (uninvested money counts as cash). It is compared with putting the same dollars into OEF on the same dates. Values use the latest stored closing prices, not live prices.

Sources: only links the agent's web search actually returned are shown. Reasons starting with "Guess:" are the agent's own judgment. The agent's view is not a tested trading signal.