Event Hours: Treasury Auctions and the London Fix, Measured
Updated October 12, 2026
Flow events with a fixed time#
Not everything with a time on the calendar is an economic release. Some are flow events: fixed moments when many participants have to buy or sell at once. Two of the most cited are U.S. Treasury auctions, which close at 13:00 New York time, and the London currency fix, at 16:00 London time.
Both come with directional rules: 'bonds drop before the auction and recover after', 'at month-end the dollar moves this way into the fix'. We measured them with one-minute bars since 2010, and against control days, which is the part you almost never see.
Treasury auctions: a pattern that existed and can no longer be told apart#
The idea has been described for years: in the hours before an auction the bond's price gives way to make room for the new issue, and after the close it recovers. The table measures the futures move from 10:00 to 13:00 and from 13:05 to 16:00, New York time, in basis points of price, net of what similar control days with no auction did.
For the 10-year auction, through 2015 the picture is there: -1.6 basis points before and +5.8 after, over 62 auctions. Since 2016 it is +0.2 and +0.9, over 131. Since 2022, +1.5 and +0.2, over 58: no drop before and no recovery after.
Over the whole sample, 193 ten-year auctions, it comes to -0.4 before and +2.5 after, and the difference between the two stretches cannot be clearly told apart from zero. The other maturities do not help: at the 7-year the futures rose before (+1.8), and at the 30-year they fell before (-4.2) and also after (-1.1).
And the test that matters: trading it. Selling from 10:00 to 13:00 and buying from 13:05 to 16:00 at every 10-year auction left, after commissions and one tick per side, -2.8 basis points per auction, over 194. At the other six maturities the net result was negative too.
US Treasury auctions: the future before and after the 13:00 ET deadline
ZT, ZF, ZN, ZB
Swipe to see the sample →
| Auction (future) and period | Before the auction, 10:00-13:00 ET (bp; negative = fall) | After the auction, 13:05-16:00 ET (bp; positive = recovery) | Auctions |
|---|---|---|---|
| 10-year note (ZN), to 2015 | -1.6 | 5.8 | 62 |
| 10-year note (ZN), 2016 onward | 0.2 | 0.9 | 131 |
| 10-year note (ZN), 2022 onward | 1.5 | 0.2 | 58 |
| 10-year note (ZN), 2010-2026 | -0.4 | 2.5 | 193 |
| 2-year note (ZT), 2010-2026 | 0.1 | 0.9 | 157 |
| 3-year note (ZT), 2010-2026 | -0.6 | 0.1 | 177 |
| 5-year note (ZF), 2010-2026 | -1.7 | 0.6 | 179 |
| 7-year note (ZN), 2010-2026 | 1.8 | 0.6 | 190 |
| 20-year bond (ZB), 2020-2026 | 7.9 | 1.7 | 78 |
| 30-year bond (ZB), 2010-2026 | -4.2 | -1.1 | 194 |
n = 1,168 · 2010-06-08 to 2026-10-08 · exchange data
US Treasury note and bond auctions whose bidding deadline is 13:00 ET, from June 2010 to October 2026 (the 20-year ones since 2020), without Fed decision days; the n in the footer adds the seven tenors. Each tenor is measured on the future of its sector: 2 and 3 years on ZT, 5 on ZF, 7 and 10 on ZN, 20 and 30 on ZB. 'Before' = price move of the future from 10:00 to 13:00 ET; 'After' = from 13:05 to 16:00 ET; in basis points of price (1 bp = 0.01%). Each cell is the mean of the auction days minus the mean of 6 control days per auction: same weekday, no more than 63 sessions away, with no auction of any tenor and no Fed decision, and with the closest 20-day volatility. The rows by period are for the 10-year auction; '2022 onward' is contained in '2016 onward'. It has been described in the literature that the price falls in the hours before the auction and recovers in the hours after it. In the 10-year future that shape appears to 2015 (after minus before: +7.5 bp, distinguishable from zero at 95%; so is the recovery alone, the prior fall alone is not) and is no longer distinguishable from zero: +0.9 bp since 2016 and −1.0 bp since 2022. Over the whole sample, with 95% confidence intervals: before −0.4 bp [−2.5 to 1.7], after +2.5 bp [0.4 to 4.6], after minus before +3.0 bp [−0.2 to 6.2]. Trading it (selling the future from 10:00 to 13:00 ET and buying it from 13:05 to 16:00 ET: two round trips, each with USD 4.50 of commission plus 1 tick per side) left, on ZN with the 10-year auction, +3.1 bp before costs and −2.8 bp [−5.5 to −0.01] after costs (194 auctions); since 2022, −3.1 bp [−8.5 to 2.3] (58 auctions). In the other six tenors the result after costs is negative too (from −1.1 to −11.0 bp per auction). Front contract, CME 1-minute bars; the price at a given time is the last trade before it, carried forward at most 15 minutes. It is what happened, not a signal.
The weekly report, every Sunday
One email a week: what moved the macro picture, next week's calendar, rates and COT positioning, and the new study with its sample size. No signals; unsubscribe in one click.
The London fix: the hour is bigger, but it does not say which way#
The fix is the reference price calculated at 16:00 London time, used by funds and companies to value and to execute. It falls at 11:00 New York time most of the year and at 12:00 in the few weeks when the two countries have not changed their clocks together.
What does hold is size. In the euro (6E), the move in the hour centred on the fix was 1.50 times that of the same window two hours later, over 4,180 sessions. Across the six currency futures measured it ranges from 1.45 to 1.62 times.
The last day of the month is bigger still: in the euro, 1.44 times a normal day at that same hour, and between 1.32 and 1.59 in the others. It is the day when most portfolios are rebalanced. Both measures hold in the data since 2020.
Direction is another story. In the hour after the month-end fix the euro rose in only 37.1% of months and moved -3.0 basis points on average. It looks like a pattern, but since 2020 that average can no longer be told apart from zero, and selling that hour, after costs, left practically nothing.
The London fixing in currency futures: size, not direction
6E, 6J, 6B, 6A, 6C, 6S
Swipe to see the sample →
| Market | Hour of the fixing vs the same window two hours later | Last day of the month vs a normal day | The future rises in the following hour, last day of the month | Average move in that hour (bp) | Sessions |
|---|---|---|---|---|---|
| 6E | 1.50 × | 1.44 × | 37.1% | -3 | 4,180 |
| 6J | 1.45 × | 1.50 × | 37.6% | -2.4 | 4,179 |
| 6B | 1.62 × | 1.59 × | 38.6% | -2.7 | 4,180 |
| 6A | 1.51 × | 1.32 × | 39.6% | -3.4 | 4,180 |
| 6C | 1.55 × | 1.48 × | 46.2% | -1.4 | 4,180 |
| 6S | 1.57 × | 1.51 × | 35.5% | -3.8 | 4,180 |
n = 25,079 · 2010-06-07 to 2026-10-09 · exchange data
CME currency futures (euro, yen, pound, Australian dollar, Canadian dollar and Swiss franc), sessions from 7-Jun-2010 to 9-Oct-2026; the n in the footer adds the six markets. The London fixing is the reference price calculated at 16:00 London time: 11:00 ET in 93.4% of the sessions and 12:00 ET in the three or four weeks a year when the US and the UK have not changed their clocks at the same time (the time is computed date by date). Size = absolute value of the price move from 30 minutes before to 30 minutes after the fixing, on average. First column: that size over all sessions divided by the size of the same window two hours later. Second: the size on the last trading day of the month divided by that of the other days (197 month-ends). 95% confidence intervals (resampling) in 6E: 1.50 [1.44 to 1.58] and 1.44 [1.28 to 1.60]. Two hours later the last day of the month is not distinguishable from a normal day in five of the six markets (1.02 to 1.12; in 6C, 1.29). Since 2020 (82 month-ends) both measures hold: from 1.56 to 1.69 and from 1.41 to 1.76 depending on the market (1.60 and 1.76 in 6E). Direction: the last two columns look at the 60 minutes after the fixing on the last day of the month (share of months in which the future rose and average move in basis points of price, 1 bp = 0.01%). In 6E the mean was −3.0 bp [−5.0 to −1.1] over the whole sample and −2.5 bp [−5.1 to 0.1] since 2020; in the average of the six markets, −2.9 bp [−4.5 to −1.2] and, since 2020, −1.8 bp [−4.3 to 0.6], with a rise in 36.6% of the months: the fall after the fixing seen in the whole sample is not distinguishable from zero since 2020. Selling the future during that hour, with USD 4.50 of commission plus 1 tick per side, left since 2020 +0.1 bp [−2.4 to 2.5] per month in the average of the six and +1.2 bp [−1.4 to 3.9] in 6E: the direction does not survive costs. Front contract, CME 1-minute bars; the price at a given time is the last trade before it, carried forward at most 60 minutes. It is what happened, not a signal.
Crude oil and natural gas: the half hour of the weekly report#
The two energy reports with a fixed time are the weekly petroleum inventories, normally on Wednesday at 10:30 New York time, and the weekly natural gas storage, on Thursday at the same time. The table measures the range of the half hour from 10:30 to 11:00 against that same half hour on the other days of the week.
In crude oil (CL) that half hour was 1.64 times that of a day with no report, over 702 reports since 2010. In natural gas (NG), 2.57 times, over 725. It is the biggest jump in this whole guide.
Both figures are smaller over the last three years: 1.25 times in crude and 2.04 in gas. And, as with macro releases, the jump stays in the window: the range of the full session grows far less.
Direction does not show up. The half hour ended above where it started 48.9% of the time in crude and 46.5% in gas, and both intervals include 50%. The report makes the move bigger; it does not say which way.
Crude oil and natural gas in the half hour after their weekly report: size, not direction
CL, NG
Swipe to see the sample →
| Market, report and half hour | Range vs the same half hour on other days, whole history | The same, last 3 years | Price rose in that half hour | Reports |
|---|---|---|---|---|
| Crude oil (CL): weekly petroleum inventories, Wednesday 10:30-11:00 ET | 1.64 × | 1.25 × | 48.9% | 702 |
| Natural gas (NG): weekly gas storage, Thursday 10:30-11:00 ET | 2.57 × | 2.04 × | 46.5% | 725 |
n = 1,427 · 2010-06-07 to 2026-10-09 · exchange data
Crude oil (CL) and natural gas (NG) futures, 1-minute bars of the 9:30 to 16:00 ET session, from 7-Jun-2010 to 9-Oct-2026; the n in the footer adds the two markets. The weekly petroleum inventories report normally comes out on Wednesday at 10:30 ET and the gas storage report on Thursday at 10:30 ET. Range = high minus low of the half hour from 10:30 to 11:00 ET, as a percentage of the first price of that window. Each multiple is the mean of that range on the usual report day divided by the mean of the same half hour on the other weekdays: for crude, 1.263% against 0.768%; for gas, 2.234% against 0.870%. It is the same definition and the same figures the portal shows in its report-day context. With 95% confidence intervals (resampling of whole weeks): crude 1.64 [1.57 to 1.72] over the whole history (702 reports) and 1.25 [1.16 to 1.34] in the last 3 years (from 10-Oct-2023, 155 reports); gas 2.57 [2.48 to 2.67] (725) and 2.04 [1.89 to 2.20] (155). The multiple is smaller in the last 3 years in both markets. The range of the whole session grows far less: 1.15 times for crude and 1.34 for gas (1.05 and 1.20 in the last 3 years). Direction: the half hour ended above where it started 48.9% of the time for crude [45.2 to 52.6] and 46.5% for gas [42.9 to 50.1]; in the last 3 years, 48.4% [40.7 to 56.2] and 43.9% [36.3 to 51.7]. All those intervals include 50%: the report enlarges the move, it does not say which way. Holidays: in weeks with a US holiday the report comes out a day later and that was not corrected; those report days sit in the 'other days' group, which makes the multiple slightly conservative. It is what happened, not a signal.
What these tables do not say#
They do not say those flows are not there. They say that, in the futures and at these hours, they leave no direction that can be traded after costs. A pattern everybody knows tends to disappear, and the auction one did.
Averages hide very different days: a weak auction can move the bond far more than the average, either way. And basis points of price in a bond future are small; a few of them amount to a tick or two.
Fed decision days are excluded from the auction measurement, and the fix is measured in the futures, not in the spot market where it is calculated.
How to use it#
As a warning about size. If you trade currency futures, the hour of the fix is roughly one and a half times busier than any afternoon hour, and more so on the last day of the month. That matters for stop distance and for whether you want a position open at that moment.
And as a vaccine against two directional rules that are still taught. The auction one had support through 2015; today, measured against control days, it does not.
This content is informational material about historical data and is not personalized investment advice. Past results do not predict future ones, and trading futures carries a risk of loss.
Free PDF: 20 futures figures with their sample
The figures from our studies on one sheet, each with its n and date range (in Spanish for now). Subscribe to the weekly report and download it right away; the link also comes in the welcome email.
Frequently asked questions
- Do bonds drop before Treasury auctions and recover after?
- In 10-year futures that pattern shows up through 2015 (-1.6 basis points before and +5.8 after, against control days, over 62 auctions) and can no longer be told apart since 2016 (+0.2 and +0.9, over 131). Trading it over the whole sample left -2.8 basis points per auction after costs.
- What happens to currencies at the London fix?
- They move more, not in a reliable direction. In euro futures the hour of the fix was 1.50 times the same window two hours later, over 4,180 sessions, and the last day of the month 1.44 times a normal day. The average month-end drop afterwards cannot be told apart from zero since 2020.
- What time is the London fix in New York?
- At 11:00 most of the year, because it is calculated at 16:00 London time. In the three or four weeks when the United States and the United Kingdom have not changed their clocks together, it falls at 12:00 New York time.
These numbers, instrument by instrument
See all 30 instruments →These base rates, on your chart
The indicator is open source and free: it marks each session's Initial Balance and gap, and prints beside them the historical rate at which that bucket fills, with its sample size. Same data as the tables above.
How to cite this data
The tables on this page are our own work, built from exchange data. You may reproduce them, translate them or build on them, including commercially, as long as you credit the source and link back to the original page. If you use them, tell us: we like knowing what they get used for.
Licence: CC BY 4.0
Suggested citation:
Perfiltrade (October 12, 2026). Event Hours: Treasury Auctions and the London Fix, Measured. https://www.perfiltrade.com/en/learn/event-hours-treasury-auctions-london-fix
Try these concepts with real data
7 days free, no card. An analysis tool, not investment advice.
See plans →