The Money Flow Indexcornerstone signal — the research case

QuantOrb (www.quantorb.pro) — research note published 27 August 2026 · written plainly, sources linked, nothing oversold

The short answer: every QuantOrb label leans on one invention from 1989 — the Money Flow Index, a volume-weighted cousin of the Relative Strength Index. We anchor on it because it holds up from three directions at once: it was designed by practitioners to fix a genuine defect (RSI ignores volume), it keeps resurfacing in independent streams of peer-reviewed quantitative research as the go-to buying/selling-pressure feature, and it earned its seat in our own 28-year back-tests, where money flow defines both published entry zones.

This note gathers the external evidence, shows how the indicator works, and states clearly how far the support reaches — and where it stops.

The MFI was introduced by Gene Quong and Avrum Soudack in Technical Analysis of Stocks & Commodities (March 1989) under the title Volume-Weighted RSI: Money Flow. Welles Wilder's 1978 RSI reads closing prices only — and classic market tops and bottoms come with violent swings in volume that a price-only oscillator cannot see. Quong and Soudack weighted each session's move by its traded volume and passed the result back through Wilder's formula. The output is the familiar 0–100 dial used ever since in most charting platforms.

How the dial is built

Five small steps (Fig. 1). Typical price collapses each session into one number; multiplying by volume turns it into dollars-at-work; splitting by up-day versus down-day sorts those dollars into opposing buckets; a 14-day ratio squeezes everything into the oscillator. Reading it is symmetric: below 50 means net selling pressure dominated the look-back window, above 50 means net buying pressure — the closer to the rails, the stronger the conviction.

Anatomy of the MFI. Every dollar of volume pushes its session toward either the buying or the selling bucket, so conviction gets priced in — not just direction. Source: Quong & Soudack (1989).

How we read it — and how chartists traditionally did

Standard treatments treat the dial as an extremes warning lamp: overbought near 80, oversold near 20, with divergences as secondary confirmation. QuantOrb uses the full sweep of the dial instead, because a rotation between Leader and Laggard registers long before any sector reaches the outermost bands:

The 0–100 dial as the orbs use it. Red = the distribution side — BUY territory begins at 40, marking sectors sold down hard. Green = the momentum side — CONTINUATION territory begins at 60. Orb size additionally weights money-flow distance from neutral at 50%. Traditional warning bands kept in outline for comparison. Note the strongest single-gauge conviction sits away from the centre — which is why orb size grows outward.

The colour rules follow directly: distribution (red) is money flow ≤ 40 with relative strength negative, accumulation (green) is money flow ≥ 60 with relative strength positive. The co-location requirement matters — money flow never acts alone, it confirms. Why 40/60 rather than something rounder: the thresholds were chosen against 28 years of outcomes, not aesthetics — the indicator-by-indicator breakdown lives in Exits and practical use.

Four independent research streams

We are not the first quantitative shop to lean on this oscillator. Recent peer-reviewed literature adopts it in four unrelated settings — and convergent adoption from uncoordinated researchers is exactly what earns a cornerstone role:

Study	Setting	Role MFI plays	Why it matters here
Jiang, Ji & Chang (2020) J. Risk Financ. Manag.	Portfolio rebalancing, S&P 500 (George Mason University)	One of 14 technical indicators fed to XGBoost models that steer a risk-aversion-adjusted, multi-period rebalance framework	Buying/selling pressure belongs inside allocation-grade pipelines — not only on chartists' screens
Malibari, Katib & Mehmood (2022) Applied Sciences (King Abdulaziz University)	Deep transformer reinforcement-learning trading agents, Tadawul	Momentum feature in the agent's state: money moving into/out of a ticker, overbought/oversold detection	Compact enough for machines — one bounded number summarises flow direction for an autonomous agent
Liu & Wei (2022) FinTech (Xi'an Jiaotong-Liverpool)	LSTM networks pricing Shanghai 50-ETF options	Among technical indicators selected into a regulated LSTM that beat both a plain LSTM and Black–Scholes on pricing-error metrics; cites earlier evidence of MFI predictability in Chinese stocks	When algorithms pick their own inputs, flow measures survive selection
Khan & Ahmad (2019) Sustainability (Quaid-i-Azam University)	Investor-sentiment econometrics, Pakistan equity market	Named among the literature's standard indirect sentiment proxies (alongside RSI, put-call ratios, closed-end discounts)	Volume-weighted pressure doubles as crowd psychology — sentiment and returns push back on each other

Different countries (US, Saudi Arabia, China, Pakistan), different instruments (equity portfolios, single-stock indices, options), different machinery (gradient boosting, transformers, LSTMs, VAR econometrics) — the same ingredient keeps appearing. Nobody coordinated that.

How much weight this carries

These papers use MFI; they don't study us. None of them tests QuantOrb's sector-ETF rules or endorses anything of ours. The synthesis — that a 37-year-old practitioner dial remains the accepted pressure gauge across modern quant research — is our reading, offered openly.
Citation counts are modest (roughly 14–65 on Google Scholar at writing time): respectable applied finance, not landmark theory. We cite these works because they are real, verifiable via DOI, and representative — not to borrow reflected authority.
The decisive evidence for our published signals is our own: 28 years of daily data, no look-ahead, consecutive signals deduplicated, every figure shipped with sample size and ±SE. Start at the 28-year back-test report.
"Money flow" is a name, not audited plumbing. MFI multiplies typical price by volume; it gauges pressure rather than tracking actual settlement flows. We say "technical rotation signal" everywhere and show the plumbing on this page deliberately.

Where it plugs into the product

BUY — money flow ≤ 40, losing relative strength to the S&P 500, close below the lower band: a sector stretched downwards that mean-reverts. Historically higher 20 days later 62.7% ±1.0pp (n=2,403).
CONTINUATION — money flow ≥ 60, beating the S&P 500, close above the upper band: momentum permitted to run. Higher 20 days later 60.1% ±0.8pp (n=3,501).
Orb size — 50% of signal strength comes from money-flow distance from neutral; relative strength contributes 30%, volume surprise 20%.
Regime label — breadth over classification across all 11 sectors produces the RISK ON / MIXED / RISK OFF banner.

New research lands here before anywhere else. Industry baskets, multi-timeframe confirmations, alternative pressure measures — subscribers get one email when a note or a back-test goes live. Test results only; no marketing.

One field, one click, unsubscribing is a single reply. We store your address and which page you joined from — nothing else. No third-party trackers.

References

Quong, G., & Soudack, A. (1989). Volume-weighted RSI: Money flow. Technical Analysis of Stocks & Commodities, 7(3), 76–77. Archived original ↗
Wilder, J. W. (1978). New Concepts in Technical Trading Systems. Trend Research. — the RSI lineage MFI extends.
Jiang, Z., Ji, R., & Chang, K.-C. (2020). A machine learning integrated portfolio rebalance framework with risk-aversion adjustment. Journal of Risk and Financial Management, 13(7), 155. doi.org/10.3390/jrfm13070155 ↗
Malibari, N., Katib, I., & Mehmood, R. (2022). Smart robotic strategies and advice for stock trading using deep transformer reinforcement learning. Applied Sciences, 12(24), 12526. doi.org/10.3390/app122412526 ↗
Liu, D., & Wei, A. (2022). Regulated LSTM artificial neural networks for option risks. FinTech, 1(2), 180–190. doi.org/10.3390/fintech1020014 ↗
Khan, M. A., & Ahmad, E. (2019). Measurement of investor sentiment and its bi-directional contemporaneous and lead–lag relationship with returns: Evidence from Pakistan. Sustainability, 11(1), 94. doi.org/10.3390/su11010094 ↗

Keep reading

28-year back-test report — the summary: what was tested, what cleared the bar, and the no-SELL decision.
Full test tables — every candidate with n, rate and ±SE; baselines and protocol.
Exits & practical use — the two standard exit rules and the indicator-by-indicator breakdown.

This note is for informational purposes only and is not investment advice. Past performance does not guarantee future results. See our disclaimer.