# Does volume tell you which way a breakout will go? A pre-registered test of common trading-volume beliefs on Nasdaq-100 (NQ) and S&P 500 (ES) futures, the SPY ETF, Bitcoin and large US stocks, 2018-2026, with every entry, stop and exit checked on 1-minute bars after costs. Published by 345 Strategies (https://345strategies.com), October 25, 2026. Research only: not a strategy, not a forecast, not financial advice. Interactive page: https://345strategies.com/research/volume Machine-readable results (JSON): https://345strategies.com/research/volume.json Code to reproduce it: https://github.com/345-strategies/strat-enrichment/tree/main/volume ## Short answer: no - High volume did not make breakouts more likely to work: not on the setup bar, the breakout minute, the day, or with every timeframe agreeing. Busy and quiet breakouts finished within a few hundredths of a risk unit (R) of each other on NQ, ES and SPY. - Low volume did not make breakouts more likely to fail. Quiet breakouts reached +1R before the stop as often as busy ones, about half the time either way. A quiet pullback (inside bar on drying-up volume) did not predict the trend continuing. - Volume does say something about how far price may travel: it made forecasts of the day's range about 2-4% more accurate on futures, mostly from overnight volume. That did not change any sizing, exit or premium-selling decision tested. - On Bitcoin, high-volume breaks lost less money after a 0.10% fee, but only because a percentage fee costs narrow, quiet setup bars more in R. Before fees the gaps disappeared, and every group lost money. - On five large stocks (AAPL, MSFT, NVDA, AMZN, TSLA, May 2018 to October 2026), 0 of 9 direction tests passed. The near miss, high-volume setup bars (+0.05R in 2023-2026), halved and fell inside the luck range before the 0.02% cost: the same arithmetic as Bitcoin. One pattern did hold on five more stocks: breaks of yesterday's high or low held into the close more often on busy days (about 63% against 55%). That is about how far a move carries, not which way, and it is not a tested trade. ## How it was tested - Trade: on 15, 30 and 60-minute charts, the first 1-minute close through the prior bar's high (long) or low (short), the TheStrat "in force" break. Entry at the next 1-minute open, stop at the other side of the prior bar, exit at the stop or the session close. Costs: 0.5 point round trip on NQ and ES, $0.02 a share on SPY, 0.10% of price on Bitcoin, 0.02% of price on stocks. - Volume: relative volume, meaning volume against the median for the same minutes over the prior 20 sessions, for the setup bar, the breakout minute, the breakout bar so far and the session so far. The busiest third of trades is compared with the quietest third. - Periods: 2018-2022 (where a pattern could be noticed) and 2023-2026 (where it has to hold). NQ 2018-2026, ES 2020-2026, SPY 2018-2026, Bitcoin on Binance 2020-2026 and Coinbase 2018-2026, stocks May 2018 to October 2026 (Nasdaq-exchange volume). Regular hours (9:30 AM to 4:00 PM ET, 6:30 AM to 1:00 PM PT) for the index markets, and the UTC day for Bitcoin. - Luck range: plus or minus 1.96 standard errors of the gap, with trades on the same day counted together. A real effect should fall outside it in both periods. - Pre-registration: each round's questions, pass rule and data were written down and committed before its results were run. Pass rule for a direction test: same sign in both periods on the primary market, |t| >= 2 in 2023-2026, Holm-corrected permutation p < 0.05 across the nine tests, and the same sign on the other markets. ## The questions, answered ### Does a high-volume setup bar make the break more likely to work? No. On NQ, ES and SPY, breaks from a high-volume setup bar finished within 0.05R per trade of breaks from a low-volume one, inside the range luck alone produces. That held for 2-1-2 and 3-1-2 setups too. On five large stocks, high-volume setup bars did 0.05R better in 2023-2026 but only 0.01R better in 2018-2022, and before trading costs the later gap halves and falls inside luck: the same cost arithmetic as Bitcoin. ### Does heavy volume on the breakout confirm the move? No. Heavy volume on the breakout minute, or on the breakout bar so far, did not separate breaks that ran from breaks that failed. The later-years gaps run from −0.02R to +0.03R on NQ, ES, SPY and five large stocks. ### Do breakouts work better on high-volume days? No. Breaks taken on busy sessions and quiet sessions finished about the same on every index market and on five large stocks, in both periods. ### Does volume leaning your way help (buying volume on a long, selling volume on a short)? No. It looked slightly helpful on futures in the early years, then not in the later ones, and SPY went the opposite way. On stocks it did nothing in either period. Once you know whether the setup bar's minutes went up or down, weighting them by volume adds nothing. ### With full timeframe continuity, does high volume add anything? No. With the 60-minute, daily and weekly bars all on the trade's side, high-volume setups did slightly worse than low-volume ones in all six index market-and-period cells, though none by more than luck explains. On stocks high volume leaned slightly better (+0.03R), also within luck, and the lean disappears before costs. ### Is a 2-2 reversal off a volume spike more reliable? Not reliably. In the later years spike reversals did a little better on all three index markets and on stocks, but every gap sits inside the luck range, a fresh check on 2026 data did not confirm it, and on stocks spike reversals did clearly worse in 2018-2022. ### Does a high-volume break of yesterday's high or low hold? No, if anything the opposite. High-volume breaks held into the close 4 to 9 points less often than low-volume ones, then went the other way on five symbols the pattern wasn't found on. On five large stocks there was no difference either way. ### Are low-volume breakouts more likely to be fakeouts? No. Breakouts on the quietest third of breakout minutes reached +1R before the stop 49.6% to 51.3% of the time on NQ, ES and SPY, the same as the busiest third (49.1% to 51.5%). On five large stocks it was 50% to 51% either way. Their average result per trade was the same too. ### Does a quiet setup bar mean nobody is behind the move? No. Breaks from the quietest setup bars averaged −0.03R to +0.03R per trade on NQ, ES and SPY, within luck of breaks from the busiest ones, and reached +1R first just as often. The one exception: on five large stocks in 2023-2026, the quietest setup bars reached +1R first slightly less often (49.6% against 51.0%), a gap that was absent in 2018-2022. ### Should you skip breakouts on low-volume days? Not on this evidence. Breaks taken on the quietest sessions finished within about 0.05R per trade of breaks on the busiest sessions, on every index market and on five large stocks, in both periods. ### Does low volume on a pullback (inside bar) mean the trend will continue? No. Inside bars where volume dried up (quietest quarter) broke no better than other inside bars: −0.03R on NQ, +0.09R on ES and −0.07R on SPY and −0.03R on five large stocks in the later years, all inside the luck range. ### Does a low-volume break of yesterday's high or low fail? No, the reverse. Low-volume breaks of yesterday's high or low held into the close more often than high-volume ones on NQ, ES and SPY (59% to 69% of days against 55% to 60%). The later-years gaps were within luck, the pattern flipped on new symbols, and on five large stocks there was no difference. ## Single stocks Five of the most traded US stocks, AAPL, MSFT, NVDA, AMZN and TSLA, May 2018 to October 2026, pooled (each date counted once) and one at a time, about 334,000 breakout trades. Pass rule as above, plus: the pooled gap had to hold before costs too (|t| >= 2), and at least 4 of 5 stocks had to agree. Result: 0 of 9 passed. High-volume setup bars did +0.05R better in 2023-2026 after costs (all five stocks agreed) but +0.01R in 2018-2022, and +0.02R before costs, inside the luck range. 3 of 45 single-stock results were beyond luck in 2023-2026, about what chance gives. Lead found by looking (not a pre-set question): when a stock broke yesterday's high or low on a busy session, the break held into the close about 64% of the time, against about 56% on quiet sessions, in both periods and on all five stocks. It is about how far a move carries on a busy day, not which way, and it was not clear on NQ, ES or SPY. Confirmation on META, AMD, GOOGL, AVGO and COST: held (pre-registered: pooled busy minus quiet > 0 with one-sided t >= 1.65, and at least 4 of 5 stocks positive). Busy-day breaks closed beyond the level 63.5% vs 55.5%, +8.0 points (t 5.9), all five positive: AMD +9.9, GOOGL +9.1, AVGO +10.5, COST +6.0, META +1.9 (META from July 2022 only, since it traded as FB before). Volume used is known before the break. Not tested as a trade after costs. Stock volume here is Nasdaq-exchange trades only. Dark-pool and other off-exchange volume is not included. ## What volume is good for Forecasting the day's range at the open improved by 4.3% on NQ and 3.8% on ES when prior-day and overnight volume were added to a price-only forecast (2023-2026), and not at all on SPY (-0.3%) or Bitcoin (-0.2%, -0.4%). Rest-of-session range forecasts after a break improved 2-4% on 2018-2026 but not on a fresh 2026 SPY window. Sizing an MES breakout system by the volume-aided forecast did worse than sizing by price alone (-0.78 vs -0.50 ticks per contract, 1,240 trades), and it did not improve when to sell SPY 0DTE straddles. ## Exits and stops Exiting on a volume spike (three times normal), with the trade or against it, cost 0.02-0.05R per trade against holding, and did no better than exiting on an unusually big price bar. Breakout volume did not tell you whether a wider or tighter stop would help. 0 of 7 exit and stop tests passed. ## What was not tested - Dark pools and off-exchange trades: a large share of US stock volume trades off-exchange. The stock volume (Nasdaq) and SPY volume before April 2026 (NYSE Arca) came from single exchanges, so off-exchange volume is missing, and block prints were not studied. Relative volume is a ratio against the symbol's own history, so a steady venue share cancels. - Order flow (delta, footprint, bid/ask), volume profile and volume at price, options volume. - Choosing which stocks to trade by relative volume from a broad list ("stocks in play"), small and mid-cap stocks, earnings days as a group, premarket volume. - Overnight futures sessions, and daily or weekly setups. ## Published claims checked - First half-hour return predicts the last half-hour, more on high-volume days (Gao, Han, Li and Zhou, Journal of Financial Economics 2018): does not hold on 2018-2026 SPY, ES, NQ or Bitcoin. - Index returns reverse more after high-volume days (Campbell, Grossman and Wang, Quarterly Journal of Economics 1993): partly holds on daily data, not on 30-minute bars. - High-volume breakouts throw back to the breakout level more often (Bulkowski): holds for prior-day high/low breaks on futures, reversed for intraday in-force breaks and on Bitcoin. - Volume tracks volatility (Karpoff 1987): holds. ## Questions a skeptic would ask **Does high volume confirm a breakout?** Not in this data. On NQ and ES futures and SPY from 2018 to 2026, breakouts with heavy volume on the setup bar, the breakout minute or the session finished about the same as breakouts on light volume, with gaps inside what luck alone produces. **Does low volume on a pullback mean the trend will continue?** Not in this data. Inside bars where volume dried up broke no better than other inside bars, and with timeframe continuity a quiet setup bar leaned slightly better in 2018-2026 but flipped on fresh 2026 data. **Is volume useful with TheStrat?** Not for direction. Across in-force breaks, 2-1-2, 3-1-2, 2-2 reversals and timeframe-continuity breaks, volume did not separate winners from losers. It does say a little about how far price will travel. **Does volume predict breakouts on Bitcoin?** No. On Binance and Coinbase, high-volume breaks lost less money after a 0.10% fee, but only because the fee costs a narrow, quiet setup bar more in R. Before fees the gaps vanished, and every group still lost money. **Does volume predict breakouts on individual stocks?** Not as a direction signal on large stocks. On AAPL, MSFT, NVDA, AMZN and TSLA from 2018 to 2026, 0 of 9 tests passed after costs. The near miss (high-volume setup bars) shrank inside the luck range before the 0.02% cost. One pattern held up on five other stocks: breaks of yesterday's high or low held into the close more often on busy days (about 63% against 55%), a sign of how far a move carries on a busy day rather than a direction signal. **Do dark pools change the volume result for stocks?** Not tested. The stock volume in this study is Nasdaq-exchange trades only, so dark-pool and other off-exchange volume is missing. Relative volume compares a stock with its own history, which cancels a steady venue share. **What is trading volume good for?** Forecasting how big the move or the day may be. Volume improved range forecasts by about 2-4% on futures, mostly from overnight volume, but that improvement did not change a sizing or premium-selling decision in these tests. ## Reproduce it Code: https://github.com/345-strategies/strat-enrichment/tree/main/volume (folder volume/, README has every command). Bring your own 1-minute bars: CME futures and NYSE Arca SPY from Databento for a few dollars, Bitcoin free from data.binance.vision, stocks from Databento XNAS.ITCH (about $5 for five stocks, 2018-2026). Fed the same bars, the code reproduces the NQ and ES tables exactly. An earlier version left the 60-minute open unset for SPY trades in the last half hour (12:30-1:00 PM PT, 3:30-4:00 PM ET), which randomized their timeframe-continuity flag. The published SPY numbers are the corrected ones and no conclusion changed.