Are Volatile Days Better for Trading? What the Efficiency Ratio Shows

Scope: measured on the ETFs listed (SPY, QQQ, IWM, DIA, GLD, SLV). Findings apply to the instruments measured and do not automatically transfer to other markets.

In all six ETFs measured, the widest quarter of sessions is also the straightest: the share of a day's range that survives as net movement from open to close, called efficiency here, runs 0.52–0.63 on the widest quarter against 0.33–0.39 on the calmest. Wide days deliver more net movement than their range alone implies. This matters if you ever size up because the morning is already moving: that instinct is what is tested here. The quartile is assigned from the closing range, so this describes what wide days are like rather than spotting one at 10:00; though once a session's range has crossed the top-quartile cutoff it cannot leave that quartile, and the instinct to size up on a merely busy morning is exactly the part the data does not support.

What is the intraday efficiency ratio?#

Efficiency = |16:00 close − 09:30 open| ÷ (session high − session low). It asks how much of the day's total travel survived as net displacement. A reading of 1 is a session that opened at one extreme and closed at the other; a reading near 0 is a session that covered ground and finished where it started.

Range on its own cannot separate those two days. A session that runs 1% up, gives it all back, and closes flat has the same range as one that walks 1% in a straight line and stays there. "Big range day" and "good trending day" are different statements, and efficiency is the number that keeps them apart. That is why the question "do volatile days trend more" is answerable at all — without a displacement term there is nothing to compare.

Do volatile days trend more, or just travel further?#

Both, and the second half is the finding. Sorting SPY's sessions into quartiles by that day's range, the calmest quarter travelled a median 0.510% and kept 0.157% of it as net move — efficiency 0.34. The widest quarter travelled 1.756% and kept 1.013% — efficiency 0.54. Wider and straighter.

SPY, 2025-02-10 to 2026-08-28, regular session 09:30–16:00 ET, one-minute bars, n = 97 / 97 / 97 / 96 sessions per quartile (387 in total). Session range = (high − low) ÷ 09:30 open. Net move = |16:00 close − 09:30 open| ÷ 09:30 open. Efficiency = |close − open| ÷ (high − low), computed per session, then the median taken within each quartile. Quartiles are cut within each ETF separately. Sessions with fewer than 300 minute bars are excluded, which removes early closes.

Do the arithmetic and the disproportion is the whole point. From Q1 to Q4 the median range widens 1.756 ÷ 0.510 = 3.4×, but the median net move grows 1.013 ÷ 0.157 = 6.5× — the net move scales 1.9× faster than the range that carried it. If wide days were simply quiet days magnified, those two ratios would match: three times the range with three times the chop nets out to the same efficiency. They do not match. The extra travel on a wide day is disproportionately directional.

Read the calm quartile from the other side and it looks worse than its small range suggests. A median quiet SPY session covers 0.510% and retains 0.157% of it — roughly two thirds of an already-thin range consumed going back and forth. The quiet day is not a smaller version of the busy day; it is a smaller version that also wastes more of what little it has.

One honest wrinkle: the ladder is not smooth. SPY's Q2 efficiency of 0.55 sits above Q3's 0.48, and QQQ (0.52 against 0.45) and DIA (0.45 against 0.41) invert in the same place. The middle of the distribution does not rank cleanly. The statement the data supports is calmest versus widest, not "more range, more efficiency" as a monotonic rule.

Does this hold outside SPY?#

It holds in all six ETFs measured, with no exceptions at the two ends. Every Q1 median efficiency lands between 0.33 and 0.39; every Q4 median lands between 0.52 and 0.63. The gap is narrowest in IWM (0.39 to 0.58) and widest in DIA (0.34 to 0.63), but the direction never reverses.

Same window and construction as the SPY table above: 2025-02-10 to 2026-08-28, 09:30–16:00 ET, one-minute bars, quartiles cut inside each ETF. n = 386 sessions each for IWM, DIA, GLD and SLV; 387 each for SPY and QQQ. The underlying medians, calmest to widest quartile — range, then net move: QQQ 0.739% → 2.285% and 0.270% → 1.470%; IWM 0.895% → 2.478% and 0.326% → 1.349%; DIA 0.590% → 1.621% and 0.197% → 1.023%; GLD 0.658% → 2.076% and 0.199% → 1.261%; SLV 1.049% → 4.075% and 0.317% → 2.234%.

The same disproportion appears everywhere. GLD's widest quartile ranges 3.2× its calmest but nets 6.3×; DIA ranges 2.7× and nets 5.2×; SLV, the noisiest of the six, ranges 3.9× and nets 7.0×. Even IWM, where the efficiency gap is smallest, moves 2.8× the range for 4.1× the net displacement.

What does this mean for trading high volatility days?#

Less than it first appears, and this is where the claim is usually misapplied. Every quartile here is assigned from the day's realised high and low. Range only grows during a session, so a day enters the widest quartile the moment its range crosses the cutoff, which for SPY in this window was 1.311% of the open; but its efficiency is unknown until 16:00, and at 09:30 nobody knows which quartile the session will land in. A market that has already covered ground by 10:00 may still hand it all back.

So this is a description of what wide days are like, not a way to find one while it is happening. The reasoning that runs "it is moving a lot this morning, so today will trend, so size up" quietly converts a retrospective label into forward-looking information it never contained — and a morning move that fully reverses by 16:00 is exactly the session efficiency is built to score near zero.

What it can legitimately do is calibrate expectations about a session whose range is already past the widest-quartile cutoff: on that kind of day, more of the range than usual has tended to end up as displacement rather than as chop, the back-and-forth that leaves price where it started. That is a shift in what to expect, not a signal. How often a session that is wide by 10:00 finishes in the widest quartile has not been measured here.

What this does not say#

• The quartiles are retrospective. Sorting is by realised range, known only after the close. Nothing here is an intraday classifier.
• The middle is not a ladder. Three of the six ETFs invert between Q2 and Q3. Only the calmest-versus-widest contrast is clean.
• Each column is its own median. Median efficiency times median range does not reproduce median net move — 0.34 × 0.510% is 0.173%, not the 0.157% listed — because each statistic is taken independently across the sessions in the quartile.
• Efficiency ignores path. It reads open, close, high and low only. A day that runs one way, fully reverses and closes near the open scores near 0 whether or not the reversal was tradeable, and a high-efficiency day may have reached its close by a route nobody could have held.
• Nothing here measures profit. No entries, no exits, no sizing, and no spread or slippage — costs that scale differently across calm and wide sessions.
• One window, one regime. About 386 sessions per ETF from 2025-02-10 to 2026-08-28. Different volatility conditions would redraw the quartile boundaries and could change the size of the gap.
• The overnight gap is excluded by construction. Everything is measured inside 09:30–16:00 ET.

Related reading#

• How much does SPY move in a day? — the range distribution these quartiles are cut from.
• The calendar effects are real and still useless — another pattern that is true about days and unusable on the day.
• What a volume spike tells you — and what it doesn't — the same retrospective-versus-live problem on volume.
• Before you believe a reversal story — what a full round trip inside a wide session actually looks like.
• Does price snap back to VWAP? — the mean-reversion side of the same session.