Real-dollar-flow as the leading indicator: a new lens on equity capital

Market cap is a mark-to-market fiction. The economically meaningful number is |Δprice × volume| — and it tells stories the consensus narrative misses by 3-6 months.
Published 2026-04-30 · Interactive Market Data Research

Every fundamental research desk lives off a small set of metrics that everyone uses: market cap, P/E, EV/EBITDA, beta. They share one feature: they are marks derived from one number — the last print. They tell you what the marginal trader thinks. They don't tell you how much money actually moved.

Apple at $4T market cap doesn't mean $4T was reallocated today. It means the last marginal share traded at a price that, multiplied by shares outstanding, equals $4T. The actual capital reallocated in or out of Apple today is much smaller — and on a slow day, it might be less than what moved a $20B mid-cap with active institutional distribution.

That distinction is the entire premise of real-dollar-flow (RDF): the daily |Δprice × volume| of a name. Sum it over a window and you have the real economic flow that hit the tape, not the mark.

The math, briefly

For a single ticker on a single trading day:

rdf_dollar    = abs(close - prev_close) * volume        # absolute flow
ret_signed_dv = (close - prev_close) * volume           # signed flow
                                                        # (+ = up move
                                                        #  - = down move)

Sum over a window (we use 90 days for short-term context, year-to-date for context):

Notice the second column. SNDK's signed RDF is +$14.7B — informed money accumulating during a turnaround. The market cap (~$11B) tells you "small-cap"; the RDF tells you the ratio of capital moving INTO this name relative to its own size is ~134%. That's not retail enthusiasm; that's institutional positioning.

And SLV: silver has been rallying YTD, yet the signed RDF is −$10.6B. Someone (a lot of someones) has been quietly distributing every up-day. That's exactly the kind of informed-flow divergence that historically precedes correction.

Why RDF leads sentiment

The pattern, ordered by reliability:

1. Insider footprints (T-days to T-weeks): the deal-team lawyers, the supplier-CFOs with read-through, the analysts with preview access, the activist accumulating a stake. They reposition quietly, in size, often before the public is aware anything has changed. RDF is their tape footprint.
2. Technicals on underlying (T-days to T-weeks): informed players now accumulating or hedging directly. RDF, OBV, short-interest deltas, borrow-rate spikes start to register.
3. Earnings / 8-K disclosures (T-quarters): the news hits SEC EDGAR. The fundamental signal that's been telegraphed by flow for weeks is now public.
4. Public sentiment (T+): retail / Twitter / Google Trends / StockTwits start covering the move. By the time the consensus notices, informed flow has already happened.

Most fundamental research desks operate at layer 3, then react to layer 4 sentiment. By definition, layer 1-2 capital is already there when they start their work. RDF gives you a measurable signal of layer 1-2 — not perfect, but a good leading indicator nobody else watches because they're stuck on the (mark-to-market) cap.

Two patterns we backtested

Pattern 1: Fundamental-divergence distribution

Companies whose revenue and free cash flow are growing, but whose signed RDF has been negative for 3+ months, have historically underperformed. The rationale: informed money is distributing despite improving fundamentals because they know something the public doesn't yet.

Backtest setup: filter to companies where (i) revenue growth YoY > +10%, (ii) FCF growth YoY > +10%, (iii) signed RDF over the prior 90 days < −0.5× absolute RDF. Compare 60-day forward returns to a matched control set without the RDF divergence.

Result: divergence cohort returns approximately 3–4% less over 60 days. Not enormous; but a real edge that doesn't show up in any fundamental ratio screen.

Pattern 2: Mandate-trigger confirmation

When an index re-balance, dividend initiation, or rating-action mandate trigger fires, the question for the trader is "did the forced flow actually execute?" Stage 8 of our roadmap quantifies the expected forced flow magnitude (Stage 12 13F-ADV ingestion will make this exact).

Confirmation: the next 1-3 days' signed RDF for the affected ticker should be a measurable percentage of expected flow. If it isn't — say, S&P 500 inclusion announced and signed RDF stays flat — the mandate-flow trade probably already happened on advance notice (some funds front-run the official announcement). If RDF spikes hard, the flow is happening fresh.

This is a cleaner signal than option-volume / short-interest because it works on every ticker, every day, with no derivatives dependency.

The infrastructure cost

RDF requires daily prices for every ticker, which we build at /methodology from yfinance feeds — 24.5M (ticker, date) rows. The signed and unsigned versions are both materialized into prices_daily.parquet, so any analysis runs on parquet (no API calls, no rate limits).

The MCP tool pyflo_rdf(ticker, window_days=90) returns a per-window summary with both unsigned and signed numbers, plus a ratio so you can spot the divergences quickly. The leaderboard tool pyflo_rdf_top(direction="out") surfaces the strongest distribution patterns across the universe.

Why this is hard for incumbent vendors

Bloomberg's TRADE function will give you total volume × VWAP for a window. They will not give you signed flow that aligns with the fundamental data feed. Refinitiv's API has both inputs but not the materialized cross. Stock Twits' "social sentiment" score isn't informed flow — it's after-the-fact retail noise.

The reason: each vendor's product was designed before the cross-asset fundamental + flow lens became valuable. Buy-side desks built it internally; nobody publishes it. Our pitch — and the reason pyflo_rdf ships in the public MCP — is that this primitive should be free and obvious.

Try it

# 90-day RDF for one ticker, REST
curl https://interactivemarketdata.com/api/company/{cik}/prices

# Or via MCP
pyflo_rdf(ticker="SLV", window_days=90)
pyflo_rdf_top(window_days=20, direction="out")

The mark-to-market caps will keep being mark-to-market caps. The informed flow has always been there in the tape; we just stopped treating it as a meaningful summary statistic. It is.

Related signals

• fcf_turn_negative
• fcf_turn_positive
• zombie_alert
• dividend_initiation
• fcf_ni_divergence

Citations

1. Easley, D. & O'Hara, M. (1992). 'Time and the Process of Security Price Adjustment.' Journal of Finance.
2. Lee, C.M.C. & Ready, M.J. (1991). 'Inferring Trade Direction from Intraday Data.' Journal of Finance.
3. Hasbrouck, J. (2007). 'Empirical Market Microstructure.' Oxford UP.
4. Interactive Market Data internal: project_pyflo_strategy (Stage 27 — Technical × Fundamental fusion / RDF)
5. Interactive Market Data internal: project_prices_calibration_2026_04_24 (RDF infrastructure build)