Capital Flows and Price Formation in Financial Markets

Prices are not determined by fundamentals: they are determined by capital flows that interpret fundamentals. Tracing those flows means getting to the source of price movements beyond the narratives that attempt to explain them after the fact.

In 2023, the Magnificent 7 (Apple, Microsoft, Nvidia, Amazon, Meta, Alphabet, Tesla) accounted for more than 60% of the S&P 500’s gain (Goldman Sachs). The cap-weighted S&P 500 clearly outperformed the equal-weight S&P 500 (S&P Global). The “market” did not rise: seven companies rose while the rest of the market went sideways. This was not the result of a synchronized fundamental re-rating across all 500 index components: it was the result of a concentration of flows into a narrow segment of the market, amplified by index ETF mechanics and short-dated options. Understanding prices without understanding flows is like confusing the thermometer with the disease.

Who buys, who sells: and why it determines everything

The durability of a price move depends on the nature of the buyers or sellers behind it. A structural illustration is provided by our analysis of the structural strong-dollar regime and its market transmission, where flow composition shapes the persistence of the price regime. Three participant categories structure the ecosystem, and their behavior differs radically depending on the regime.

Institutional flows (pension funds, insurers, sovereign wealth funds) operate on long horizons with allocation constraints driven by liabilities. The Government Pension Fund Global (Norway, $1.6T, NBIM) rebalances quarterly toward its 70/30 equity/bond target, when equities fall, it mechanically buys (forced contrarian). U.S. pension funds ($4.1T in DB plans, Federal Reserve Flow of Funds) reduced their equity allocation from 65% to 45–50% between 2000 and 2023 (Milliman): a long-term structural selling flow masked by valuations. These moves are slow, predictable, and create durable underlying trends.

Retail flows are historically pro-cyclical: buyers at market tops, sellers at bottoms. Net flows into U.S. retail equity funds reached a peak in January 2021 (ICI): the exact market peak before the 2022 correction. Net flows were negative in October 2022 (ICI): exactly at the trough. Retail buys enthusiasm and sells fear. The rise of zero-commission platforms (Robinhood: 23M accounts, SEC filing 2023; Trade Republic, eToro) amplified the volume of these flows and their sensitivity to media narratives, the GameStop episode (Jan 2021: +1,600% in two weeks, NYSE) demonstrated retail’s ability to temporarily dominate price formation in narrow segments.

Speculative flows (hedge funds, CTAs or Commodity Trading Advisors, market makers) amplify short-term moves. CTAs (~$350B AUM, BarclayHedge) are structurally trend-followers: they buy when prices rise and sell when prices fall, amplifying both directions. Their aggregate positioning, as revealed by CFTC Commitments of Traders data, is a sentiment indicator, and a reversal signal when positioning reaches extremes (record net long = maximum complacency, record net short = maximum pessimism).

The passive shift: how ETFs changed price formation

The rise of ETFs is the most significant structural transformation in financial markets over the past two decades. Global ETF assets surpassed $11T in 2023 (ETFGI). In the U.S., ETFs and index funds now hold more than 50% of total equity market capitalization (Morningstar, 2024). Passive management captures most inflows: in 2023, U.S. passive funds saw net inflows while active funds experienced outflows (Morningstar). See also, on price formation in a physical commodity chain, where the cocoa price actually forms.

This shift profoundly alters price formation. When an S&P 500 ETF receives inflows, it mechanically buys all 500 index constituents, regardless of individual valuations. The 7 largest caps (Magnificent 7: ~30% of S&P 500 weight, S&P Global 2024) mechanically capture 30% of every dollar invested in an S&P 500 ETF, independent of price, fundamentals, or valuation. This “blind” flow levitates the most heavily weighted stocks as long as the index gathers inflows, and triggers indiscriminate selling that creates valuation anomalies in fundamentally solid names when the index sees outflows.

The effect is amplified by concentration: the Magnificent 7 at 30% of the S&P 500 means a “diversified” S&P 500 ETF concentrates 30 cents of every dollar into seven companies. An investor buying an MSCI World ETF for diversification actually holds 20% of their portfolio in the same seven U.S. tech firms (MSCI). The analysis of AI thematic ETF risks shows how this flow concentration creates major valuation distortions: a phenomenon further developed in the Equities and ETFs pillar.

Share buybacks: the market’s largest buyer, until it disappears

S&P 500 buybacks reached very large amounts in 2022 and 2023 (S&P Global). Net buybacks by nonfinancial corporations were the largest source of net demand for U.S. equities in 7 of the 11 years from 2014 to 2024, and the second largest over the period as a whole, behind households (Federal Reserve, Z.1, table F.5.1.1). Apple alone repurchased more than $600B of its own shares between 2013 and 2023 (SEC filings): equivalent to the market cap of the vast majority of S&P 500 companies.

This flow creates a structural support floor under indices, particularly visible during shallow corrections. Companies with board-authorized programs intensify repurchases when their stocks fall: a mechanically counter-cyclical behavior that dampens corrections as long as firms retain sufficient free cash flow and funding access.

The fragility lies in conditionality: buybacks vanish precisely when they are most needed. In Q1–Q2 2020, S&P 500 buybacks fell sharply (S&P Global) as companies suspended programs to preserve cash. In recessions or liquidity crises, the support floor disappears. Counter-cyclical flow becomes pro-cyclical by absence, amplifying corrections instead of cushioning them. Buybacks are funded by free cash flow and increasingly by corporate debt: when spreads widen and refinancing becomes costly, repurchase programs are the first to be cut.

Sector rotations: the cycle read through flows

Within equity markets, capital does not distribute evenly, rotations across sectors are among the most informative signals, a regime-shift dimension developed in our analysis of the delayed effects of restrictive monetary policy of institutional cycle expectations.

Early cycle (recovery phase): flows move into cyclicals, materials, industrials, consumer discretionary, small caps. In Q2–Q3 2020, U.S. small-cap ETFs (IWM) gathered inflows (Bloomberg) after years of outflows. The Russell 2000 clearly outperformed the S&P 500 between Nov 2020 and Mar 2021 (Russell Investments). Late cycle: flows migrate toward defensives (healthcare, utilities, consumer staples) and quality stocks (strong balance sheets, predictable cash flows). In Q3–Q4 2022, “quality factor” ETFs (QUAL) gathered inflows while growth ETFs (QQQ) saw outflows (Bloomberg).

The 2022 growth → value rotation was particularly strong: the Russell 1000 Value outperformed the Russell 1000 Growth by 22% that year (Russell Investments). This reversal is directly tied to the rate regime shift: growth stocks, whose valuations depend on distant cash flows discounted at low rates, are most sensitive to rate increases, when discount rates move from 0% to 5%, the present value of distant cash flows collapses. The sector rotation study decodes these regime shifts and their predictive power.

Idle liquidity: $6 trillion on the sidelines

U.S. money market fund assets reached a record $6T at the end of 2023 (ICI). This shift fits within our piece on interest rates, financial markets and asset allocation, where positive real rates have rehabilitated cash as a competing asset class. This massive shift to cash is not irrational risk aversion; it is the arithmetic consequence of the rate regime change: when money market funds yield 5.0–5.4% (Crane Data, 2023), the opportunity cost of staying uninvested is low, and cash compensation is the most attractive since 2007.

This liquidity reserve has a dual interpretation. Negative signal: risk aversion is elevated, investors prefer cash over risk. Positive signal: $6T of dry powder can be redeployed into risky assets if sentiment improves, enough to turn technical stabilization into a powerful rally. Historically, peaks in money market assets (2009 and 2020, ICI) preceded major equity rallies, not because cash “automatically returns to stocks,” but because asset peaks coincide with peaks in pessimism, which coincide with market bottoms.

Conversely, unusually low money market balances (2000 and 2007, ICI) signal near-full capital deployment, the question becomes: who is left to buy? This configuration preceded the last two major bear markets.

Algorithmic flows: the invisible volatility layer

Alongside slow flows (institutional, retail, buybacks), ultra-fast flows now operate at unprecedented speed and scale. Algorithmic trading accounts for roughly 60–70% of daily U.S. equity volume (SEC, 2023). Zero-day-to-expiration (0DTE) options now represent over 40% of S&P 500 options volume (CBOE, 2024), up from less than 5% in 2019. Trend-following CTAs (~$350B AUM, BarclayHedge) reposition portfolios within minutes when technical thresholds break.

These ultra-fast flows generate volatility episodes disconnected from fundamentals. On August 5, 2024, the Nikkei fell 12.4% in a single session: largely due to the mass unwind of yen/dollar carry trades (borrowing in yen at 0% → investing in higher-yield assets) amplified by automated hedging algorithms. The next day, the Nikkei rebounded sharply. No economic fundamental changed in 48 hours: only flows reversed. The AI, liquidity, and fast-shock analysis explores how these new flow sources reshape market microstructure.

The coexistence of flows with radically different time horizons, from quarterly pension rebalancing to millisecond algorithmic repositioning, complicates market interpretation. The same price move may reflect a fundamental perception shift or simply the unwind of a gamma hedge. Distinguishing the two requires understanding the flow structure, not just direction.

Real-rate gaps: what sorts cross-border flows

Across borders, capital responds to real rates more than to posted policy rates, and real rates diverge persistently between economies. Brazil held its Selic at 15.00% in January 2026 against twelve-month IPCA inflation of 4.44% (Banco Central do Brasil, IBGE), a real rate of 10.6 points, among the highest of major economies. The easing cycle has since taken the Selic to 13.75% on 17 September 2026. Turkey long combined one of the highest nominal rates in the world with a deeply negative real rate. Disinflation flipped the balance: in December 2025 its 36.5% policy rate stood above inflation of 30.9% (TurkStat), a real rate of +5.6 points, where a year earlier inflation was still running at 44.4%.

Configurations this far apart generate cross-border capital flows that restructure financing conditions globally. Portfolio managers adjust allocations to real returns by region, and multinational corporations factor real-rate gaps into investment-location decisions. A side-by-side reading of policy rates misses the point. A 5% rate with 2% inflation and a 5% rate with 6% inflation correspond to opposite monetary conditions, and only real rates adjusted for local inflation and risk premia make the comparison meaningful.

Crypto assets: the laboratory of pure flows

The crypto-asset market offers a privileged observation field for capital flow analysis: their influence is direct and not mediated by fundamentals. The absence of cash flows, dividends, or traditional valuation metrics makes prices almost entirely determined by instantaneous supply-demand balance. The SEC’s approval of 11 spot Bitcoin ETFs (Jan 2024) generated $12B in net inflows within three months (Bloomberg): directly converted into buying pressure on an asset with fixed supply (maximum 21M BTC). IBIT (BlackRock) set the fastest ETF asset-gathering pace in history over its first weeks (Bloomberg). Speed of that order says less about the asset than about the intermediation layer around it, itself a product of financialization and its long expansion.

The adoption and price surge analysis formalizes the mechanism: expanding investor base → inflows → fixed supply → mechanical price rise → attracts new investors → reflexive loop. The same mechanism works in reverse: outflows → fixed supply but urgent sellers → decline → cascading liquidations (>$1B in 24h during the Aug 2024 flash crash, CoinGlass) → bearish reflexive loop. The Crypto Assets pillar examines this flow dependence and its link to the global liquidity cycle.

Tracing flows: sources and limitations

Flow tracking requires aggregating multiple sources, each offering partial insight. ETF flow data (Bloomberg, ETFGI: daily) reveal retail and advisor allocation preferences. The Bank of America Global Fund Manager Survey (monthly) captures institutional positioning. CFTC Commitments of Traders reports (weekly) detail hedge fund futures positioning. ICI fund/ETF aggregate flow data (weekly) provide a broad view. The Federal Reserve Flow of Funds (quarterly) reconstructs flows by investor category: the most comprehensive view but published with a three-month lag.

These data have limitations: family office flows (~$6T global AUM, Campden Wealth) are absent from public statistics. Pension fund flows (outside the largest plans) remain opaque. Algorithmic flows are largely invisible in aggregated data. The picture is necessarily incomplete, but vastly more informative than the opacity that characterized markets twenty years ago.

The sector rotation analysis explains how sector flows reveal cycle expectations. The AI ETF risk study documents valuation distortions created by passive flow concentration. The AI and fast-shock analysis explores how algorithmic flows reshape market microstructure. The crypto adoption study formalizes the mechanical link between inflows and valuation for fixed-supply assets.

← Back to pillar page Financial Markets

Last updated — 20 October 2026