High-Volume Return Premium

The rules

Within small, medium and large size groups by market value, buy stocks whose volume on the last day of a 50-day interval is in the top 10% of their own 50 daily volumes, and sell short those in the bottom 10%: either $1 in total on each side, equally weighted (zero-investment portfolios), or $1 in each stock, offset by $1 the other way in an equal-weighted portfolio of its size group (reference-return portfolios). Hold 1 to 20 days (up to 100) without rebalancing (Gervais, Kaniel and Mingelgrin 2001, NYSE stocks at $5 or more, 1963-1996). Stocks with a merger, delisting, partial liquidation or new share offering on the interval's last day or in the year before it are excluded.

Universe

US common stocks via the Market Screener: the 30 with the highest volume against their 20-day average

Bars

daily

Position size

3.2% of equity per stock, at most 30 held. Published: $1 bought and $1 sold short per size group, spread equally (zero-investment), or $1 per stock, offset by its size group's equal-weighted portfolio (reference-return).

Rebalance

Daily check, buying only when a slot of the 30 is free; each stock held 20 trading days. Published: formed every 50 trading days, held 1 to 20 days.

What the research found

Gervais, Kaniel and Mingelgrin (2001) studied NYSE stocks from August 1963 to December 1996. Buying stocks with unusually high volume and selling short those with unusually low volume, $1 on each side, earned 0.41% (1 day) to 0.96% (20 days) per $1 bought among small stocks and 0.14% to 0.55% among large ones, before costs, with t-statistics all above 3 (a t-statistic of about 2 or more is the usual bar for a result unlikely to be chance). The high-volume side alone, which this build buys, beat its size group's equal-weighted portfolio by 0.43-0.50% over 20 days. At 50 and 100 days, gains from daily volume shocks levelled off, fell for large firms and grew less significant, so the authors call the premium's permanence ambiguous; after weekly shocks they persisted. The tendency of returns to continue or reverse, firm announcements, market risk and liquidity did not seem to explain the premium. It was about as large, and in many tests larger, when the volume shock came with little or no price change, and was mainly found in stocks whose prior 49-trading-day return was below the median stock's. The authors' cost test, on 1993-1994 data, counted only the bid-ask spread, the gap between what sellers ask and buyers bid. Buying at the ask and selling at the bid lost money in every size group, while posting buys at the bid and sells at the ask (limit orders) was mildly profitable for small and medium firms, and more so for those with no unusual price move on the volume day. Because orders move prices, they add, none of the strategies would be profitable on a large scale. Kaniel, Ozoguz and Starks (2012), with one author in common, studied 41 developed and emerging markets and found the premium in almost all the developed ones and many of the emerging ones. In their US sample (NYSE and Nasdaq, 1963 to June 2001), with a broader top and bottom 20% cut-off, the 20-day premium was still significant in 1997-2001 (1.21% zero-investment, 0.57% reference-return). Across the G-7 markets, the US among them, the premium stayed significant after retail investors' estimated explicit costs, such as commissions, but not after adding large investors' implicit costs, such as the price moves their big orders cause. In China, Wang, Wen and Singh (2017) found no premium, and the reverse among speculative stocks (small ones, and those with low institutional ownership or analyst coverage).

Results

Hypothetical: a backtest with modeled costs, not traded with real money.

Hypothetical growth of $10,000

Rebuild it in Orbital

Paste this into Orbital Chat. The Strategy Developer builds the workflow and asks before it runs any backtest.

On daily bars, use a Market Screener universe of US stocks with the type set to Common stock only, priced at $5 or more, with 20-day average volume of at least 500,000 shares, Rel volume (the day's volume over its 20-day average) of at least 2, Change % of at least -5% and at most +5%, and Volatility 1M (average daily range) of at least 0.75%, sorted by Rel volume (highest first), top 30. Buy each stock in the screen with 3.2% of equity at the next open, skipping any already held. Sell each position at the open 20 trading days after entry (a time stop of 20 daily bars), with no other exit, even if it leaves the screen. Hold at most 30 positions, with no daily limit on new ones: fill free slots in screen order, skip the rest, and never sell a holding early to make room. Long only; uninvested capital stays in cash. Backtest from March 1, 2017 to October 30, 2026 with $100,000.

Is it real?

Strategies tested together

7

Beta to SPY

0.93

Alpha vs SPY

−11.5% a year

Costs paid

$13,955, 91% of the gain before costs

Data

Daily bars, 2,432 sessions

Fills

Next open at market, sized at the fill

Read this before you trust it

Long only. The paper buys high-volume NYSE stocks and sells short low-volume ones within each size group, in the zero-investment and reference-return portfolios the rules describe. The paper's test is that the day's volume is in the top 10% of the stock's own last 50 days. Here it is a top-30 ranking of US common stocks by the day's volume against their own 20-day average (that day included), with a floor of twice the average. The screen is checked daily, but with at most 30 stocks held, each for 20 trading days, new stocks are bought only into freed slots: stocks bought together are sold together, and the open that sells them can buy replacements from the previous day's screen. The paper forms its portfolios once every 50 trading days. Stocks are bought at the next open, while the paper measures returns from the close of the volume day. The paper drops stocks with mergers, delistings or share issues; nothing here checks for those events. The plus or minus 5% limit on the volume day's price change follows the paper's finding that the premium is about as large, and in many tests larger, without an extreme return. The 0.75% floor on the average daily range (high to low) of the last 21 trading days is an addition: it keeps out shares pinned near a fixed price for most of a month, though a takeover target can still pass in the first weeks after its deal is announced. A $5 price floor and a floor of 500,000 shares of 20-day average volume keep out the thinly traded small stocks, where the paper found the premium largest. Each stock gets a fixed 3.2% slot. Orders fill at the next open with 0.05% slippage (the assumed trading cost) on each buy and sale.

Sources