Money

Fading morning gap-ups paid in liquid stocks and lost in illiquid ones

By Strota Newsroom · 2026-08-11 · How Strota reports

researchbacktestintradayliquiditygap fade

A decade-long test of intraday short-selling on Indian equities found the largest raw edge in the hardest-to-trade names, which is exactly where costs swallowed all of it and more.

Say a stock ends one session quietly and opens the next morning sharply higher on overnight news. Does the jump hold through the day, or does the market spend the session handing it back? Strota put that question to a decade of Indian equity data. The jump tends to fade, the fade is largest in thinly traded stocks, and in exactly those stocks the cost of trading it is larger still. Net of realistic costs, the illiquid version of the trade lost money. Only the liquid version survived, and it survived by a margin thin enough to make this an execution question rather than a market one.

Before costs, the pattern is unusually clean. Sorting a panel of 481 Indian stocks into five groups by how much a given amount of trading moves the price, the average drift from the opening print to the closing print was negative in every group. It also worsened steadily as liquidity thinned, from about 11.1 basis points a day in the most liquid fifth to 17.4 in the least liquid. A basis point is one hundredth of a percentage point, so these are small daily numbers that compound into large annual ones.

Concentrating the short into the stocks with the biggest overnight jumps turned that mild tendency into something that looks, on a spreadsheet, absurd. Selling the twenty largest gap-ups at the open and covering at the close returned 38 basis points a day in the liquid group and 53 in the mid-liquidity group, with Sharpe ratios of 4.6 and 7.0. A Sharpe ratio measures return against the wobble in that return; anything above about two is rare in public markets. Every one of the eleven years tested was positive. Figures that large are a warning that something is missing, and what was missing here is the cost of getting in.

Three costs were then layered on. Statutory charges - transaction tax, exchange fees, stamp duty and GST - come to roughly four basis points a round trip. Slippage, the gap between the price on screen and the price you actually get, was scaled to how liquid the band was. The third is the one most backtests quietly skip: an entry haircut, meaning you do not get to sell at the printed opening price. Part of the spike has already faded by the time an order fills, and the biggest gap-ups are precisely the names where the opening print is least available to a real order.

With charges set at levels the study called realistic, the liquid group held up. Restricting the trade to shares turning over more than fifty crore rupees a day left roughly 8 to 12 basis points a day, a Sharpe ratio near 1.0 to 1.2, a compound annual rate of 21 to 31 percent, nine of the eleven years positive, and a worst peak-to-trough fall of about 30 percent. Push the cost assumptions from realistic to harsh and the same trade goes negative.

Here is the part that makes this a finding rather than a table. In the middle band, where daily turnover runs between ten and fifty crore rupees and where the illiquidity idea points you, the raw edge was bigger and the net result was minus 7 basis points a day. Losing. Whatever extra return you earn for accepting illiquidity is smaller than the extra cost illiquidity charges you. Cost scales faster than edge, and the only place the trade paid was the liquid end, where the raw signal is weakest.

That inversion is the entire contribution of the illiquidity factor. Ranking stocks explicitly by the measure inside a band added nothing beyond the choice of band; the factor reduced to fade harder where trading is thinner, which is the same information as picking the band in the first place. Its value proved diagnostic rather than predictive. It showed where the money appeared to be and, in the same breath, why it was out of reach.

Within the liquid band, the variant that held up best sold the ten to fifteen largest overnight gap-ups above one percent in equal rupee amounts and covered at the close. Fifteen names gave a Sharpe ratio of 1.22 and a compound rate of 31 percent over the window. Concentration mattered: ten names scored 1.41 against 0.71 for thirty. That describes what historical data did under one set of assumptions, and nothing more.

Those assumptions carry nearly all the weight. Between the gross fade and the modelled cost sits a window of roughly 30 basis points a day, and everything the strategy earned lives inside it. Whether a live account can genuinely sell near the open, on the most excitable names of the morning, cannot be settled from historical bars. Only real fills settle it.

Decay is the second problem, and it shows in the recent numbers. Yearly net results ran to plus 88 in 2024 and plus 52 in 2025, then turned to minus 12 in 2026. Gap fading is a crowded trade, and Strota own paper engine is part of the crowd. A pattern many participants are watching for is a pattern whose payoff shrinks.

Two further limits deserve naming. Shorting a riser is not always possible - some hit the upper circuit band, and same-day borrow is not always available on every name. The universe was also assembled from present-day index membership, so companies that collapsed or shrank out of the list never appear at all, a survivorship bias that flatters any long-horizon result built on such a sample. A multi-week momentum overlay was tested alongside and dropped, since a signal measured over weeks says nothing useful about a single session.

What remains is narrow. A same-day short of the morning biggest risers, confined to liquid names, cleared realistic costs in this test at a Sharpe ratio near 1.2, while being fragile, fading, and dependent on a fill nobody has verified with live money. The sturdier lesson came out of the illiquidity work by accident: the measured edge was biggest exactly where the measured cost was biggest, and a study that skips the second half of that sentence will show you a strategy that was never there.

Fading morning gap-ups paid in liquid stocks and lost in illiquid ones

More money stories