The ‘volume precedes price 73 percent of the time’ claim has no peer-reviewed academic source. Four major academic papers from 1987 through 2009 contain zero quantification of the specific hit-rate. Lee and Swaminathan’s data shows the high-volume winners retail traders chase reverse fastest over the following years. Acting on the unverified signal costs $1,241,031 over 35 years on a typical retail portfolio.
- FOUNDATIONAL Karpoff (1987), Journal of Financial and Quantitative Analysis: 72-study survey establishing that volume correlates with price-change magnitude, not direction.
- FOUNDATIONAL Jones, Kaul & Lipson (1994), Review of Financial Studies: NASDAQ-NMS daily volatility decomposes into transaction count, not trade size.
- FOUNDATIONAL Lee & Swaminathan (2000), Journal of Finance: NYSE-AMEX 1965-1995, the winner-minus-loser momentum spread is larger for high-volume firms (1.46% monthly) than low-volume (0.54%); high-volume winners reverse faster.
✓ Fact-checked against Lee & Swaminathan (2000) Journal of Finance 55(5)
✓ Verified with Karpoff (1987) JFQA 22(1):109-126
✓ Cross-referenced with Jones-Kaul-Lipson (1994) RFS 7(4):631-651
Quick Answer
The ‘volume precedes price 73% of the time’ claim has no peer-reviewed source. A 2026 audit covered Karpoff 1987, Jones-Kaul-Lipson 1994, Lee-Swaminathan 2000, and Tsang-Chong 2009. Zero of these four papers quantify the specific 73% hit-rate claim. Lee-Swaminathan complicates the folk-claim for the most common retail case.
Among momentum portfolios the winner-minus-loser spread was larger for high-volume firms (1.46% monthly) than low-volume (0.54%) across 1965-1995 NYSE-AMEX data, yet those high-volume winners reversed faster over the following years. One peer-reviewed exception applies. Tsang and Chong (2009) documented OBV outperformance in six of nine Greater China indices, applicable to a small minority of US retail accounts. The compound cost on $120,000 plus $300 monthly over 35 years reaches $1,241,031.
The ‘volume precedes price 73% of the time’ claim has zero peer-reviewed academic source.
Lee-Swaminathan 2000 found the winner-minus-loser momentum spread was actually larger for high-volume firms (1.46% monthly) than low-volume (0.54%) across 1965-1995 data — but those high-volume winners reverse fastest over the following years. Most retail traders never see this asymmetry. They watch a green candle, watch the volume bar widen, click Buy on the next bar, and the price reverses soon after.
TheFinSense’s quant analysis of four peer-reviewed papers across 1987-2009 finds zero academic quantification of the ‘73%’ folk-claim. This $1,241,031 spread extends the same arithmetic this site has been tracking across the broader Pillar C cluster. Our MACD analysis pegged the same compound arithmetic at $359,104 over twenty years; the Bollinger Band squeeze audit at $26,328 over fifteen. Each indicator carries a different mechanism, while the executor-layer cost compounds along the same arc.
This applies to US large-cap retail trading on commission-free brokerages; intraday algorithmic strategies are out of scope.
What Is the ‘Volume Precedes Price 73%’ Claim, and Who Repeats It?
Priya — a composite stand-in for the median early-career retail trader, not a real person — opens TradingView this morning and the OBV row sits one click below MACD in the Volume submenu.
The strongest case for watching volume is that some prominent traders insist they have used it profitably for decades and platforms institutionalize it. Three reinforcers stack the case: Granville’s original 1963 OBV theory and the visibility of OBV alongside MACD on every retail platform. The third reinforcer is the steady drumbeat of pattern-confirming Twitter case studies. If you adjust your account today, you can still keep volume in view as a secondary timing filter rather than a primary direction signal.
In 2026, screen-tap trading apps surface OBV alongside MACD as one-click overlays, presenting it as peer-tier without academic backing.
Priya opens TradingView and sees OBV listed in the Volume submenu without academic warning label.
📚 Source: Foundational citation — Karpoff (1987), JFQA 22(1):109-126 · doi.org/10.2307/2330874. Full four-paper audit methodology at /methodology/.
If you have seen the ‘73%’ line on Twitter and felt unsure where it came from, this article answers the question every skeptic has half-asked. If you already trade on volume confirmation, the compound math on a $120,000 portfolio across 35 years will reframe the cost you have been treating as zero. If you research indicator literature as method, the Karpoff-to-Lee-Swaminathan asymmetric arc carries the lineage no platform explainer assembles. Before evaluating any technical signal, retail traders benefit from grounding in what a stock actually represents at the corporate-claim level.
| Where the claim appears | Cited source | Verifiable academic basis |
|---|---|---|
| StockCharts ChartSchool OBV page | Granville 1963 (popular) | 0 of 4 peer-reviewed papers |
| NinjaTrader indicator blog | Unsourced “studies show” | 0 of 4 |
| Twitter trading-tip threads | Anecdotal case studies | 0 of 4 |
| YouTube/Udemy course modules | Generic technical-analysis lineage | 0 of 4 |
Three platforms repeat the statistic. Four peer-reviewed papers studied volume and price — none quantified it.
The pattern underneath is straightforward. The folk-claim asks whether volume predicts price direction. The four peer-reviewed papers ask different questions: Karpoff measures magnitude versus direction, Lee-Swaminathan measures cross-sectional returns conditioned on turnover, Jones-Kaul-Lipson decomposes volatility into trade count and trade size.
None of these is the question the folk-claim is asking. The “73 percent” has no peer-reviewed source because no peer-reviewed paper measured that specific question. Acting on the misframed signal compounds to roughly $1.2 million over a retail lifetime.
Read this guide if: you trade US large-cap equities on a commission-free retail brokerage and have seen “volume precedes price” cited as a directional signal.
Does not apply to:
- active intraday traders with sub-1-minute volume bars
- quant funds running explicit Lee-Swaminathan V1/V3 turnover-conditional momentum
- Greater China index ETF rotators
- market-making desks where volume IS the model
If three top journals did not measure it, who decided 73 percent?
How Many Peer-Reviewed Papers Quantify the 73% Claim?
The literature audit covers four peer-reviewed papers spanning 1987 to 2009. Karpoff (1987) surveyed 72 prior empirical studies and established that volume correlates with price-change magnitude in equity markets, not with price-change direction. Jones, Kaul, and Lipson (1994) decomposed daily NASDAQ-NMS volatility into transaction count and average trade size, finding the size component carried zero incremental information beyond count. Lee and Swaminathan (2000) examined NYSE and AMEX data from 1965 through 1995 and found the winner-minus-loser momentum spread was larger for high-volume firms (1.46 percent monthly) than for low-volume firms (0.54 percent), with high-volume winners reversing faster over the next several years. Zero of these four papers quantify the specific 73 percent hit-rate claim that circulates on retail platforms.
The audit answer arrives across four decades of peer-reviewed silence.
| Paper | Year | Journal | Sample | Finding on ‘73%’ |
|---|---|---|---|---|
| Karpoff | 1987 | JFQA | 72-study survey (1959-1986) | No quantification |
| Jones, Kaul & Lipson | 1994 | RFS | NASDAQ-NMS 1986-1990 | No quantification |
| Lee & Swaminathan | 2000 | JF | NYSE+AMEX 1965-1995 | No quantification (asymmetric reversal) |
| Tsang & Chong | 2009 | Economics Bulletin | Greater China + global indices | No quantification (6/9 OBV outperformance) |
What did each of the four papers actually measure?
Karpoff (1987) surveyed 72 prior empirical studies on volume-price correlation magnitude. Jones-Kaul-Lipson (1994) decomposed daily NASDAQ-NMS volatility into transaction count and trade size. Lee-Swaminathan (2000) ran NYSE-AMEX turnover-conditioned momentum sorts 1965-1995. Tsang-Chong (2009) tested OBV on nine indices, finding it profitable in six Greater China markets.
0 of 4 peer-reviewed studies surveyed by TheFinSense audit quantify the ‘73%’ folk-claim hit-rate.
While this article focuses on the volume citation gap, the same compound-drag arithmetic appears across the TheFinSense MACD and Bollinger Band squeeze win rate audits.
Lee’s framing matters because it specifies magnitude and persistence, never directional precedence. The folk-claim invented the third leg.
| Paper | Papers quantifying 73% hit-rate |
|---|---|
| Karpoff 1987 | 0 |
| JKL 1994 | 0 |
| Lee-Sw 2000 | 0 |
| Tsang-Chong 2009 | 0 |
The cleanest refutation in the audit comes from Karpoff. Volume correlates with how much price moves, not which way it will move next. That single distinction, magnitude versus direction, is what the “73 percent” claim flattens. It is also what no peer-reviewed paper has ever supported.
📚 Source: Karpoff, J.M. (1987). Journal of Financial and Quantitative Analysis 22(1):109-126 · doi.org/10.2307/2330874
📚 Source: Lee, C.M.C. & Swaminathan, B. (2000). Journal of Finance 55(5):2017-2069 · doi.org/10.1111/0022-1082.00280
📚 Source: Jones, C.M., Kaul, G., & Lipson, M.L. (1994). Review of Financial Studies 7(4):631-651 · doi.org/10.1093/rfs/7.4.631
Sullivan, Timmermann, and White’s 1999 study applied White’s Reality Check bootstrap to roughly 7,846 trading rules on a century of Dow data. The best rule held up in-sample through 1986 even after correcting for data-snooping, but its edge vanished out-of-sample over 1987-1996.
Four peer-reviewed papers. Zero measured the 73 percent. One viral statistic.
Each segment reaches the same number from a different door, and the door does not change what waits on the other side.
What does Lee-Swaminathan’s data actually show?
Why Does Lee-Swaminathan’s Data Complicate the Folk-Claim?
Lee and Swaminathan’s 2000 finding complicates the folk-claim through a specific portfolio-sort mechanism. Their NYSE-AMEX 1965-1995 data shows that the winner-minus-loser (R10−R1) momentum spread was larger for high-volume firms (1.46 percent monthly) than for low-volume firms (0.54 percent) — a 92-basis-point gap. Momentum is therefore stronger among high-volume names in the intermediate term, but high-volume winners reverse fastest over the following three to five years, so the recent high-volume winners retail traders chase are the ones most prone to multi-year reversal. Karpoff’s 1987 survey already separated volume-magnitude correlation from volume-direction precedence — and Jones-Kaul-Lipson (1994) showed the signal is fundamentally a transaction-count proxy with zero information in trade size. Three mechanisms hide inside one folk-claim.
Lee and Swaminathan answered with a turnover sort that cuts against the naive reading of the folk-claim.
One clarification on what “complicates” means here. Lee-Swaminathan is a cross-sectional finding on portfolio sorts, not a temporal-precedence test. The direct refutation of temporal precedence, the question of whether volume now predicts price next, stays with Karpoff’s magnitude-versus-direction distinction. Lee-Swaminathan is the additional finding: even among the recent winners that retail traders pile into, the high-volume subset reverses fastest over the next three to five years.
Why does the asymmetric volume momentum effect reverse for retail traders?
Lee and Swaminathan showed the winner-minus-loser momentum spread was larger for high-volume firms (1.46 percent monthly) than low-volume (0.54 percent), but high-volume winners reverse fastest over the next three to five years. Retail traders typically buy high-volume rallies, placing them in the high-volume tercile (V3) — exactly the cohort most exposed to that multi-year reversal.
She realizes the platform never required her to verify the 73 percent source before enabling the indicator.
How does the Lee-Swaminathan turnover sort work mechanically?
Lee-Swaminathan independently double-sorts NYSE-AMEX stocks 1965-1995 into 10 past-return deciles and 3 turnover terciles. The R10-V3 cell (high-momentum high-volume) holds for 6 months and rebalances monthly. Hansen-Hodrick autocorrelation correction handles overlapping holding periods across the 30-year sample window.
The RSI overbought signal cluster sits inside the same regime-conditional architecture as Lee-Swaminathan’s asymmetric volume sort. Thirty-five years of saving toward a balance. One click, and a reversal soon after.
Section 3 of Hwang (2026), Q1 2026 Balance Sheet Stress Report extends Lee-Swaminathan-style volume conditioning to balance-sheet stress quintiles in S&P 500 constituents. Methodology builds on Lee-Swaminathan rather than confirming it.
📚 Source: Hwang, D. (2026). TheFinSense Q1 2026 Balance Sheet Stress Report (internal working paper, not peer-reviewed) · ssrn.com/abstract=6614679
One sort. Two volume tiers. A hidden asymmetry.
Formula:
FV_passive = P×(1+r/12)^(12t) + PMT×((1+r/12)^(12t)−1)/(r/12)FV_follower = same formula with r=6% (drag-applied)GAP = FV_passive − FV_follower
Model: LUMP_PLUS_CONTRIBUTION two-path comparison (Buy-and-Hold passive vs Volume-Following 200bps composite drag).
Assumptions:
- 8% gross annual return on the passive baseline
- 6% net annual return for the volume-follower path (200bps composite drag)
- $120,000 initial balance plus $300 monthly contribution
- 35-year horizon, monthly compounding, end-of-period contributions
- Pre-tax (tax-advantaged shell assumed for base case)
- Federal Reserve SCF cohort-median anchored persona
Does not apply to:
- Greater China index ETF holdings (Tsang-Chong 2009 exception)
- Lee-Swaminathan-style asymmetric quant strategies with explicit V1 vs V3 turnover conditioning
- Intraday market-microstructure contexts where volume IS the model
Regulatory catalyst: N/A (TYPE II audit, no FASB/IRS/SEC change).
Tax-drag note (AP-4): Part of the gap accrues in taxable brokerage accounts, where high-turnover OBV-signal trading adds a short-term capital-gains layer on top of the 200bps composite drag. A precise STCG figure requires a turnover-specific tax model with assumed realization rates and bracket, and is not estimated here; the headline $1,241,031 is the pre-tax compound gap only.
*Priya is a hypothetical composite drawn from common early-career retail-trader patterns, not a real individual.*
Karpoff (1987) established the magnitude-not-direction distinction. Jones-Kaul-Lipson (1994) showed the volume signal is mechanically a transaction-count proxy. Lee-Swaminathan (2000) documented the asymmetric volume-momentum reversal; Tsang-Chong (2009) found OBV outperformance in 6 of 9 Greater China indices as the empirical exception.
Lee-Sw (NYSE+AMEX 1965-1995), JKL (NASDAQ-NMS 1986-1990), Karpoff (cross-market 1959-1986). Modern post-2000 HFT/decimalization era applicability requires temporal hedge.
The common thread across Karpoff (1987) and Lee-Swaminathan (2000) is that ‘volume precedes price’ was not measured directionally. Both papers measured magnitude or asymmetric conditional sort instead.
Lee-Swaminathan’s data shows the recent high-volume winners retail traders buy are the ones that reverse fastest over the following years.
Lee-Swaminathan’s asymmetry shows the mechanism on academic data; the next section drops the same arithmetic onto Priya’s $120,000 sleeve across thirty-five years.
Translating Lee-Swaminathan’s monthly basis-point spread into 35 years of compounded dollars sets the case study scale.
Same data. A hidden asymmetry. Zero retail awareness.
How much does this cost over thirty-five years?
The $1,241,031 Cost of Acting on an Unverified Signal
Priya’s $1,241,031 spread arrives at year 35 from one missing citation.
Lee-Swaminathan’s asymmetry shows the mechanism on academic data, and the next section drops the same arithmetic onto Priya’s $120,000 sleeve across thirty-five years.
A 31-year-old retail trader with $120,000 invested over 35 years at 8 percent versus 6 percent faces a $1,241,031 spread. The 200-basis-point composite drag captures execution friction, behavior-gap losses from acting on the unverified signal, and slippage during high-turnover exits. Most retail traders anchor on per-trade visible cost of $5 to $15 in spread, missing the compounding lifetime effect by a factor of roughly 230 to 700, depending on the per-trade cost assumed. The $1,241,031 figure represents roughly fifteen years of after-tax income for a typical early-career professional (about $84K net, on a first-year base in the $105K–$115K range).
Each segment reaches the same number from a different door, and the door does not change what waits on the other side.
Priya has just opened TradingView this morning, finger hovering over Add Indicator. The OBV row sits below MACD in the menu, the same one-click distance. Three Twitter accounts told her last night that volume precedes price seventy-three percent of the time — a figure that circulates on social media with no academic citation behind it. She has been saving for thirty-five years toward a balance she has never seen yet.
She has clicked OBV before, watched a volume spike on a green candle, entered long the next bar, and seen the price reverse soon after.
Our does MACD work audit pegged the 20-year cost at $359,104, applying the same compound-drag mechanism over a shorter horizon. The SMA vs EMA crossover wealth gap mirrors the same trade-frequency arithmetic that compounds Priya’s drag across 35 years.
| Parameter | Value |
|---|---|
| Initial balance | $120,000 |
| Monthly contribution | $300 |
| Time horizon | 35 years |
| Passive return rate (buy-and-hold) | 8.0% annual |
| Follower return rate (volume signal) | 6.0% annual |
| Composite drag (TC + behavior gap + slippage) | 200 basis points |
Most readers will guess the lifetime cost of acting on the ‘73%’ claim falls somewhere between $20,000 and $80,000.
Readers anchor on per-trade cost, perhaps $5 to $15 spread times 10 trades per year times 35 years, roughly $1,750 to $5,250 lifetime. They underestimate the compound execution drag by two to three orders of magnitude — the lifetime gap is 230 to 700 times that crude per-trade estimate.
| Year | Buy-and-Hold ($) | Volume Follower ($) | Gap ($) | What That Gap Buys |
|---|---|---|---|---|
| Year 5 | $200,825 | $182,793 | $18,032 | Family vacation |
| Year 10 | $321,241 | $267,491 | $53,749 | New mid-tier crossover, paid in cash |
| Year 15 | $500,642 | $381,737 | $118,905 | Home down payment, roughly 24% on a $500K property |
| Year 20 | $767,922 | $535,837 | $232,086 | Private 4-year college tuition plus room and board |
| Year 25 | $1,166,129 | $743,695 | $422,434 | Mortgage paid off on a mid-size home |
| Year 30 | $1,759,395 | $1,024,064 | $735,332 | Healthcare endowment, retirement-long premium coverage |
| Year 35 | $2,643,271 | $1,402,239 | $1,241,031 | About 15 years of $84K net annual income |
| Year | Buy-and-Hold ($) | Volume Follower ($) |
|---|---|---|
| Year 5 | 200,825 | 182,793 |
| Year 10 | 321,241 | 267,491 |
| Year 15 | 500,642 | 381,737 |
| Year 20 | 767,922 | 535,837 |
| Year 25 | 1,166,129 | 743,695 |
| Year 30 | 1,759,395 | 1,024,064 |
| Year 35 | 2,643,271 | 1,402,239 |
Priya stops. Reads $1,241,031 twice. Year 10 already shows $53,749 gone.
Priya taps Confirm on the OBV indicator.
Eleven scenarios stress-test Priya’s $1,241,031 result. The most impactful variable is the drag magnitude itself, with secondary sensitivity to time horizon.
Sensitivity table, 11 rows by 5 fields
| Scenario | Variable changed | With Buy-and-Hold | With Volume Follower | Gap |
|---|---|---|---|---|
| BASE | $120K + $300/mo / 35yr / 8% vs 6% (drag 200bps) | $2,643,271 | $1,402,239 | $1,241,031 |
| Row 1 | Drag = 1.0pp (other params at BASE) | $2,643,271 | $1,921,055 | $722,216 |
| Row 2 | Drag = 1.5pp (other params at BASE) | $2,643,271 | $1,640,300 | $1,002,970 |
| Row 3 | Drag = 2.5pp, high-volatility stress (other params at BASE) | $2,643,271 | $1,200,285 | $1,442,986 |
| Row 4 | Drag = 3.0pp, severe drag (other params at BASE) | $2,643,271 | $1,028,874 | $1,614,397 |
| Row 5 | Horizon = 20yr (other params at BASE) | $767,922 | $535,837 | $232,086 |
| Row 6 | Horizon = 30yr (other params at BASE) | $1,759,395 | $1,024,064 | $735,332 |
| Row 7 | Horizon = 40yr (other params at BASE) | $3,960,109 | $1,912,342 | $2,047,767 |
| Row 8 | Initial = $50K (other params at BASE) | $1,502,792 | $833,591 | $669,202 |
| Row 9 | Initial = $200K (other params at BASE) | $3,946,675 | $2,052,123 | $1,894,551 |
| Row 10 | Monthly = $500 (other params at BASE) | $3,102,047 | $1,687,181 | $1,414,866 |
| Row 11 | Monthly = $100 (other params at BASE) | $2,184,494 | $1,117,297 | $1,067,197 |
⚡ Jump to the finding: $1,241,031 equals roughly 15 years of Priya’s working-life income →
$1,241,031. From one missing citation.
If the cost is this large, why does the platform still default-list OBV next to MACD?
When Does Volume Actually Carry Tradable Information?
The platform defaults because volume does carry information in markets where the asymmetric test holds.
Volume signals carry exploitable information in specific niches that the folk-claim ignores. Tsang and Chong’s 2009 study found On-Balance Volume rules outperformed buy-and-hold in six of nine indices, concentrated in Greater China markets including Hang Seng and Shanghai. Lee-Swaminathan’s asymmetric framework also remains tradable when implemented with explicit V1 versus V3 turnover conditioning, though that requires programmatic execution beyond retail dashboards. A small minority of US retail traders hold Greater China exposure. For everyone else, acting on the signal is largely execution drag against the academic data.
The strongest case for watching volume is that some prominent traders insist they have used it profitably for decades and platforms institutionalize it. Three reinforcers stack the case: Granville’s original 1963 OBV theory and the visibility of OBV alongside MACD on every retail platform. The third reinforcer is the steady drumbeat of pattern-confirming Twitter case studies. If you adjust your account today, you can still keep volume in view as a secondary timing filter rather than a primary direction signal.
The 73% claim might one day acquire its missing source, somebody could run the test and confirm the folk-wisdom retroactively.
Who Should Use a Different Approach?
A small minority of US retail traders hold Greater China index exposure, where Tsang-Chong’s 2009 evidence on OBV applies.
If your portfolio holds Greater China index funds or you trade Lee-Swaminathan-style asymmetric momentum, keep volume as a secondary filter, not a primary signal.
The Greater China evidence uses index-level outperformance counts rather than the monthly returns the rest of the article tracks.
Step 1: Check the cited source for any volume signal claim
Verify whether the volume-signal claim links to a peer-reviewed paper. Of the four academic papers TheFinSense audited from Karpoff 1987 through Tsang-Chong 2009, none quantify the specific ’73 percent volume precedes price’ hit-rate. Treat any unsourced viral statistic as a default-fail until proven otherwise.
Step 2: Filter by mechanism, magnitude versus direction
Karpoff (1987) survey established that volume correlates with price-change magnitude in equity markets, not with direction. When a claim says volume predicts which way price moves next, it conflates the magnitude relation with a directional one. Filter out any signal that fails this distinction immediately.
Step 3: Apply the Lee-Swaminathan asymmetric test
If a signal proposes that high-volume rallies confirm direction, check Lee-Swaminathan’s V1 versus V3 turnover sort. The winner-minus-loser momentum spread was larger for high-volume firms (1.46 percent monthly) than low-volume (0.54 percent) across 1965-1995 NYSE-AMEX data — but high-volume winners reverse fastest over the next several years, exactly the cohort the buy-high-volume-rallies rule loads up on.
Step 4: Log drag and reassess every 90 days
Compute personal composite drag against a no-signal buy-and-hold baseline every quarter. A 200-basis-point annual gap compounds to $1,241,031 over 35 years on a $120,000 portfolio with $300 monthly contributions. Cancel any signal that fails the four-paper audit and the drag log together.
📚 Source: Tsang, W.W.H. & Chong, T.T.L. (2009). Economics Bulletin 29(3):2424-2431 · ideas.repec.org
If no source link, treat as default-fail.
If yes, the signal is a category error.
If yes, the asymmetric-reversal risk applies.
If yes, you have evidence on your own positions.
All four PASS: Retain volume as secondary timing filter only. Any FAIL: Remove OBV from default chart layout. Never use OBV as a primary direction signal.
So here is where Priya can stop the bleed before the next thirty years compound it further. The calculator below runs her exact parameters and any sensitivity she wants to test against the academic finding.
Volume-Signal Compound Cost
Run your own portfolio against the 200-basis-point composite drag baseline to see the 35-year gap between buy-and-hold and acting on the unsourced 73% claim.
| Year | Buy-and-Hold | Volume-Following | Gap |
|---|
This calculator is presented as an embedded tool. Run the Volume Signal Cost Calculator with any portfolio size, drag assumption, and horizon to see your own compound exposure.
Future peer-reviewed retests will be tracked in the backtest catalog with full Python reproducibility.
Year 35 compound cost lingers as opportunity cost, never appearing on monthly broker statements as a line item.
| Platform | OBV Default | Academic Warning? | One-Click Add? |
|---|---|---|---|
| TradingView | YES (Volume submenu) | NO | YES |
| thinkorswim | YES (Studies Standard tab) | NO | YES |
| Interactive Brokers | YES (TWS Chart Indicators) | NO | YES |
| Charles Schwab | YES (Chart workspace Indicators) | NO | YES |
Removing OBV from primary direction signals saves Priya $1,241,031 across thirty-five years.
Volume signals work somewhere. The 73 percent claim still works nowhere academic.
The golden cross win rate breakdown rides under every execution debt this site tracks, including the volume citation gap.
Next time you see a percentage with no citation, ask: which paper, which year, which population.
When a new peer-reviewed paper tests the ‘73% volume precedes price’ claim on post-2020 US data, this article will update with the re-tested numbers.
Will the next folk-statistic survive the same four-paper audit?
Volume-Precedes-Price FAQ: Five Common Questions Answered
Five common questions about the volume-precedes-price claim share one academic backbone. The folk-claim has no peer-reviewed quantitative source for the 73 percent hit-rate. Volume and price action measure different properties: price tracks direction while volume tracks intensity, not future direction. On-Balance Volume reliability is conditional, succeeding in Greater China indices per Tsang-Chong 2009 but failing in US large-cap data per Karpoff 1987 and Lee-Swaminathan 2000. High volume statistically accompanies reversals without forecasting them, since Lee-Swaminathan showed the recent high-volume winners retail traders buy reverse fastest over the following years.
Does volume actually predict stock price movement?
Volume does not predict stock price direction in any peer-reviewed academic dataset on US large-cap equities. Karpoff (1987) surveyed 72 prior empirical studies and concluded that volume correlates with the magnitude of price changes, not the direction. Lee and Swaminathan (2000) found the volume-momentum relationship is asymmetric on NYSE and AMEX data from 1965 through 1995: the winner-minus-loser momentum spread was larger for high-volume firms (1.46 percent monthly) than low-volume (0.54 percent), and high-volume winners reverse fastest over the following years. The viral ’73 percent volume precedes price’ claim has no peer-reviewed quantification in this literature. Anyone treating volume as a directional predictor is conflating correlation magnitude with directional forecasting.
Volume vs price action: which matters more?
Volume and price action carry different information types, with price action measuring direction and volume measuring intensity. Karpoff (1987) established that volume correlates positively with the magnitude of price changes, meaning a large volume bar indicates a large move happened, not which way the next move will go. Price action, by contrast, encodes the directional signal itself. For retail traders deciding between the two, the academic literature points toward price action as the primary signal and volume as a secondary intensity filter at most. Treating volume as the primary direction signal is the category error the ’73 percent’ folk-claim institutionalizes.
Can high volume signal a price reversal?
High volume can accompany a reversal but does not predict one in US large-cap data per Karpoff 1987. The Lee-Swaminathan (2000) finding goes further: high-volume winners reverse faster than low-volume winners over the subsequent three to five years, so they tend to underperform over that longer window. The momentum spread itself is larger for high-volume firms (a 92-basis-point monthly gap), but the reversal is what catches the trader who bought the high-volume rally. So while a single high-volume bar near a top is statistically correlated with a coming reversal, the directional prediction does not follow from the volume signal itself. Acting on it as a forecasting tool produces the $1,241,031 compound cost the case study above quantifies.
Is OBV a reliable indicator?
OBV reliability is conditional on market and use case. In Greater China indices, Tsang and Chong (2009) documented OBV outperformance in six of nine markets, including the Hang Seng and Shanghai composites. In US large-cap equities, Karpoff (1987) and Lee-Swaminathan (2000) NYSE-AMEX 1965-1995 data show the directional folk-claim has zero peer-reviewed support, and that high-volume winners reverse fastest over multi-year horizons. The practical answer involves using volume as a secondary timing-filter at most, never as the primary direction signal, and never on the strength of unsourced 73 percent claims. For US retail accounts, treat OBV as decorative on the chart, not predictive in the strategy.
Why does OBV fail in US markets?
OBV underperforms as a directional rule in US large-cap markets because the volume-momentum relationship is asymmetric per Lee-Swaminathan 2000. Their NYSE-AMEX 1965-1995 data shows the winner-minus-loser momentum spread was larger for high-turnover firms (1.46 percent monthly) than low-turnover firms (0.54 percent), but those high-turnover winners reverse fastest over the next three to five years. OBV, which accumulates volume on up days and subtracts it on down days, flags exactly the high-volume rallies that Lee-Swaminathan identified as the future underperformers. The Greater China exception (Tsang-Chong 2009) does not transfer because the institutional structure and liquidity profile differ from US retail markets. The 73 percent claim repackages a directional forecast onto a magnitude-only signal.
Bottom Line: The Volume Precedes Price Myth Costs $1,241,031
After the four papers, the calculator, and the $1,241,031 milestone, one structural pattern remains.
The mechanism Karpoff (1987) named, Jones-Kaul-Lipson (1994) decomposed, and Lee-Swaminathan (2000) mapped is one mechanism. The viral ‘73%’ claim packages it as a directional rule. Three academic mechanisms compressed into one folk-claim that no peer-reviewed paper from 1987 through 2009 has quantified. Our does MACD work audit traced the same arithmetic over twenty years. The cluster pattern is consistent across every signal this site has tested.
The drag is binary at the platform level: every retail trader pays the same surcharge in compound years.
Open your charting app today. Remove OBV from your default layout. If you keep it, label it timing only.
The latch holds quietly, and you only learn which side it opened toward thirty-five years later, when the room is already empty.
What the four-paper audit reveals beyond the citation gap is a structural pattern. Every viral retail-trading rule TheFinSense has audited shares the same compound-drag arithmetic. MACD at 20 years, Bollinger squeeze at 15 years, volume signal at 35 years. All three run at 1.5 to 2.0 percentage points of net annual return loss compounded over the working life. The 73 percent statistic was simply the most cite-able version of an indicator class that already lacked academic backbone.
We were never the trader who could read direction from volume alone.
The MACD audit pegged the 20-year cost at $359,104.
At sixty-six, Priya pulls up her balance.
The latch held quietly the whole time, and the lock arrived thirty-five years later. One million two hundred forty-one thousand dollars had passed by then.
YOUR TURN
Which other folk-statistic in your charting app deserves the same four-paper audit?
Update History
- 2026-05-21: Initial publish.
- 2026-05-22: Editorial accuracy pass (v29-h45 retro). Corrected ORCID identifier; restated Sullivan-Timmermann-White 1999 as a Reality Check bootstrap with in-sample/out-of-sample split (not FDR); relabeled the Lee-Swaminathan 0.54%/1.46% figures as within-volume-group winner-minus-loser momentum spreads and corrected the reversal interpretation; reframed Tsang-Chong 2009 as the pro-OBV Greater China exception rather than a refutation; replaced the composite-DOI source strip and pointed secondary citations to publisher DOIs; corrected the “factor of two hundred” to a 230-700x range; removed the unsupported “iceberg AP-4 / $744,619” figure; removed unsourced specifics (“forty minutes”, fixed 6%/94% retail figures, “rising fastest among first-time investors”) and de-anonymized community-signal asides; added grounding for the $84K income anchor and a hypothetical-persona label.
AI assistance disclosure: portions of the literature audit and compound-math validation were performed with AI tooling under editorial oversight. All academic citations, figures, and conclusions were human-verified by Danny Hwang.
Educational quantitative analysis based on published data. Not investment, tax, or legal advice. Consult a licensed professional before acting on any calculation. About TheFinSense.
