📅 Originally Published: · Last Updated:
Trading volume can add context, but the research reviewed here does not support a universal rule that it predicts the next price move. The popular hit-rate claim could not be traced to a peer-reviewed source. Treat volume or OBV as supporting evidence, not a reason to change an order on its own.
A wide volume bar feels persuasive. It shows that many shares changed hands, and the charting platform places that information next to price. The extra confidence matters when the bar changes a trade even though the directional case has not improved.
The “volume precedes price” shortcut hides several different questions. Is volume related to the size of a price move? Does it lead the move? Does it improve a specific trading rule? Does that rule still work after costs? Research can answer those questions, but they are not interchangeable.
What Does “Volume Precedes Price” Actually Mean?
The “volume precedes price” phrase is often used as though it describes one testable fact. In practice, it can refer to at least four different claims:
| Possible meaning | Question being tested | Evidence needed |
|---|---|---|
| Contemporaneous relation | Do high-volume periods coincide with larger price changes? | Volume and return data measured over the same interval |
| Lead-lag relation | Does volume at time t predict price direction at time t+1? | A defined horizon, directional rule, and out-of-sample test |
| Confirmation rule | Does volume improve a price-based signal? | A complete entry, exit, benchmark, and cost model |
| Profitable strategy | Does an implementable volume rule beat buy-and-hold? | Net returns, multiple samples, and robustness checks |
On-Balance Volume, or OBV, is one specific transformation of price and volume. It adds the day’s volume when the closing price rises and subtracts it when the closing price falls. Traders usually interpret the slope or a divergence from price, rather than the indicator’s absolute level.
That definition already uses price to assign the sign of volume. A claim that OBV “leads” price therefore needs a separate rule describing what pattern counts as a signal and how far ahead the forecast extends.
Source: Tsang and Chong define OBV and test moving-average crossovers in Profitability of the On-Balance Volume Indicator.
A failed viral statistic does not settle the broader issue. Volume may still improve a specific, testable strategy, so the remaining evidence has to be separated by hypothesis rather than dismissed in one sweep.
Test 1: Can the 73% Figure Be Traced?
The “volume precedes price” hit-rate figure could not be traced to a peer-reviewed paper in this review. Exact-phrase and variant searches were followed by a manual check of four frequently cited papers on price-volume relations, momentum conditioned on turnover, and an OBV trading rule. None reports a universal 73% directional hit rate.
This is a bounded conclusion. It does not prove that no paper, working paper, book, or private backtest has ever printed the number. It means the statistic should not be presented as established academic evidence without a citation that identifies the sample, rule, horizon, and calculation.
The figure also has no clear denominator. Seventy-three percent of what: next-day closes, breakouts, trend continuations, individual stocks, index observations, or profitable trades? Without that definition, even a genuine backtest result would not transfer safely to another strategy.
For readers evaluating similar claims, the more useful habit is to open the source before opening a position. The same discipline matters when reviewing a polished historical result, which is why TheFinSense’s guide to technical-analysis backtests focuses on out-of-sample tests, costs, and data-snooping risk.
Test 2: What Do Foundational Studies Measure?
The foundational papers support narrower statements than the folk claim. Karpoff’s 1987 survey summarized contemporaneous relationships between trading volume and price changes. Jones, Kaul, and Lipson’s 1994 study found that the familiar volume-volatility relation was largely associated with the number of transactions, not average trade size. Neither paper establishes a universal forward-direction hit rate.
| Study | What it measured | What it supports | What it does not establish |
|---|---|---|---|
| Karpoff (1987) | Survey of research on price changes and trading volume | Volume is related to the magnitude of price changes and, in equities, to signed price changes | A fixed probability that volume predicts the next move |
| Jones, Kaul, and Lipson (1994) | Transactions, share volume, and return volatility | The number of transactions carries the main volatility relation; trade size adds little beyond frequency | A stand-alone directional trading rule |
| Lee and Swaminathan (2000) | Future returns of momentum portfolios conditioned on past turnover | Turnover helps distinguish the magnitude and persistence of momentum | A time-series OBV signal or a universal next-period forecast |
| Tsang and Chong (2009) | OBV moving-average crossover rules in nine stock indices | Results can differ sharply by market and rule length | A result that generalizes to every market or security |
Karpoff’s abstract is especially useful because it states the distinction plainly: volume was positively related to the magnitude of price changes and, in equity markets, to the price change itself. That is evidence of association. It does not specify that today’s volume predicts tomorrow’s sign with a stable success rate.
Jones, Kaul, and Lipson narrow the mechanism further. Their result points to trading frequency as a major link between market activity and volatility. A busy market can therefore signal that information is being processed without telling a long-only investor whether the next return will be positive.
Sources: Karpoff (1987) and Jones, Kaul, and Lipson (1994).
Test 3: What Does Volume Add to Momentum?
Here volume becomes useful, but only inside a tightly defined momentum design. Lee and Swaminathan found that past turnover changed both the strength and persistence of momentum portfolios. For a six-month formation and six-month holding design, the winner-minus-loser spread was 1.46% per month among high-turnover stocks and 0.54% among low-turnover stocks. That finding concerns cross-sectional portfolios, not a chart-level rule that buys whenever volume rises.
The paper’s longer-horizon result matters just as much. High-volume winners and low-volume losers moved into reversal sooner than their counterparts. In the authors’ “late-stage” portfolio, the first-year gain was followed by negative returns in later years. That pattern makes volume informative, but in a more conditional way than the folk rule suggests.
There are three transfer limits:
- Turnover is not the same as a one-day volume spike. The study ranked firms using average daily turnover over the prior six months.
- Portfolio sorting is not an OBV crossover. The result compares groups of stocks rather than issuing a buy signal on one chart.
- The sample is historical. The main tests use NYSE and AMEX data from 1965 through 1995, so implementation today requires a fresh, cost-aware replication.
The right conclusion is that volume may refine a momentum model when the variable and horizon are defined. It does not grant a general license to treat every high-volume rally as confirmation.
Source: Lee and Swaminathan, Price Momentum and Trading Volume, Journal of Finance 55(5), 2017-2069.
Test 4: When Has OBV Worked?
Tsang and Chong provide the strongest exception to a broad dismissal of volume indicators. They tested four OBV moving-average crossover rules on nine stock indices with data ending in April 2009. The best OBV rule beat buy-and-hold on a gross basis in six of the nine indices. The result was concentrated in Greater China, while the tested Dow Jones, CAC 40, and FTSE 100 rules lagged buy-and-hold.
The paper defined an actual strategy: take a long position when OBV crosses above its 10-, 20-, 50-, or 100-day moving average, then exit when it crosses below. Short selling was prohibited. That makes the result interpretable in a way the generic phrase “volume precedes price” is not.
| Market group | Result in the study | Reader takeaway |
|---|---|---|
| United States and Europe | The tested rules generally trailed buy-and-hold; after the paper’s 0.25% transaction-cost assumption, all reported net relative returns were negative | No evidence for a broad developed-market OBV edge in this sample |
| Greater China | Several rule lengths produced positive net relative returns, with substantial variation by index | Market and parameter choice materially changed the result |
This paper is a counterweight to an overly broad dismissal of volume indicators. OBV was not useless everywhere. It also shows why a universal statistic is misleading: the same rule family produced very different outcomes across markets, moving-average lengths, and cost assumptions.
The study has limitations. It selects the best result from four parameter choices, uses index-level volume from DataStream, and ends before major changes in market structure over the following decade and a half. Its authors also note that the Chinese-market conclusion was influenced by the 2005-2007 boom. Those limits do not erase the finding, but they narrow how far it should travel.
Source: Tsang and Chong (2009), Profitability of the On-Balance Volume Indicator, Economics Bulletin 29(3), 2424-2431.
How Should Investors Use Volume Without Overreading It?
A credible volume precedes price claim begins as a rule another person could test. A “volume precedes price” rule should not answer a directional question until its signal and forecast horizon are defined.
1. Write the signal as a complete sentence
“High volume is bullish” is not a testable rule. A complete version names the asset universe, lookback window, threshold, forecast horizon, entry, exit, and benchmark. Once those pieces are visible, the claimed edge can be checked rather than admired.
2. Separate information from direction
Volume may show unusual activity, liquidity, participation, or the arrival of information. None of those automatically determines the sign of the next return. Use price, volume, and fundamentals for the roles they actually measure.
3. Demand a net, out-of-sample result
A strategy should survive transaction costs, bid-ask spreads, taxes where relevant, and a period that was not used to choose the settings. A high in-sample hit rate can disappear once the rule is moved to new data. The same issue appears in many indicator studies, including the evidence reviewed in TheFinSense’s MACD analysis.
4. Use volume as context unless the evidence earns more
For a long-term investor, volume can help explain liquidity around earnings, rebalances, index changes, or stressed trading. It can also be a secondary input in a documented system. Treating it as a stand-alone forecast requires much stronger evidence than a slogan or a chart screenshot.
Common questions
Does high volume mean a stock will rise?
No. High volume shows heavy trading activity. The same activity can accompany buying pressure, selling pressure, news processing, forced liquidation, or rebalancing. Direction requires a separate model or signal.
Is On-Balance Volume a leading indicator?
OBV is built from current and past closing-price direction plus volume. Traders may use divergences or moving-average crossovers as leading signals, but the forecast depends on the exact rule, market, and horizon. The label alone does not establish predictive power.
Did Lee and Swaminathan test OBV?
No. They sorted stocks by past returns and average daily turnover, then measured future portfolio returns. Their evidence shows that turnover can condition momentum, but it is not a direct test of OBV or a one-chart volume spike.
Can volume be useful without predicting direction?
Yes. Volume can help describe participation, liquidity, and how strongly a market is processing new information. Those uses do not require treating volume as a standalone forecast of the next price move.
What would make a volume strategy credible?
A credible test specifies the asset universe, data frequency, signal, lag, entry and exit rules, benchmark, costs, and sample split. Results should survive out-of-sample data and reasonable parameter changes.
Bottom Line: Treat Volume as Evidence, Not a Verdict
The phrase “volume precedes price” compresses several different ideas into one confident rule. The research reviewed here supports relationships between volume, trading activity, volatility, momentum, and future returns. Those relationships are conditional and measured in different ways.
The reader decision is not whether to delete every volume panel. It is whether the indicator has earned authority over a trade. Keep it as context when it helps you understand participation or liquidity. Before using it for direction, require a defined rule, a relevant sample, costs, and an out-of-sample result.
When a percentage arrives without those details, treat it as an unverified claim until you can inspect the underlying test.
YOUR TURN
Which volume rule in your charting setup can you describe precisely enough for someone else to reproduce?
- FOUNDATIONAL Karpoff (1987): survey of the relation between price changes and trading volume.
- FOUNDATIONAL Jones, Kaul, and Lipson (1994): transaction frequency, share volume, and volatility.
- SUPPORTING Lee and Swaminathan (2000): momentum portfolios conditioned on past turnover.
- SUPPORTING Tsang and Chong (2009): OBV moving-average rules across nine indices.
Question: Does the available evidence support a universal claim that volume predicts price direction?
Search scope: Exact-phrase and variant searches completed July 23, 2026, followed by full-text review of the four papers listed above.
Reasoning rule: Contemporaneous association, cross-sectional prediction, and a time-series trading rule were treated as separate hypotheses.
Backtest decision: Not applicable to the viral percentage as stated because it does not define a market, data frequency, signal, forecast horizon, or benchmark. Creating a backtest would require inventing the missing rule.
Limit: Failure to trace the statistic is not proof that it has never appeared anywhere. It is a reason to withhold academic authority until a reproducible source is produced.
AI assistance disclosure: AI tools assisted with source organization, consistency checks, and HTML validation. Danny Hwang reviewed the cited findings, calculations, and final conclusions.
Update History
- May 21, 2026: Initial publication.
- July 23, 2026: Rebuilt the article around defined research questions, removed the unsupported lifetime-cost model and calculator, corrected the scope of the cited studies, and moved detailed evidence into the bottom trust package.
Educational quantitative analysis based on published data. Not investment, tax, or legal advice. Consult a licensed professional before acting on any calculation. About TheFinSense.