Does WTI have days, months or hours with a bias you can actually measure? We aggregated 4,555,284 M1 candles of the WTI crude CFD from January 2013 to October 12, 2026 by day of week, calendar month, session and UTC hour, and ran every result through a significance test. Bottom line: 1 of the 17 calendar buckets comes out at p < 0.05, but it does not survive a multiple-comparison correction. A hint, not a rule.
📊 The data behind this article, as tables and CSV: /stats/WTI publishes WTI seasonality, sessions and spread as a downloadable reference, and /data has the OHLCV candles (D1 down to M1) so you can replicate every table. This article is the analysis, with its tests; those pages are the raw numbers.
1. Seasonality by day of the week
Each day closes at 19:00 UTC, a fixed hour present in every year of the feed. Test: one-sample t-test against zero; bold marks what survives a Bonferroni correction.
| Day | n | Mean % | Median % | Std % | Win rate | t-stat | p-value |
|---|---|---|---|---|---|---|---|
| Monday | 697 | -0.006% | +0.103% | 2.903% | 52.1% | -0.05 | 0.959 |
| Tuesday | 707 | -0.082% | +0.016% | 2.951% | 50.2% | -0.73 | 0.463 |
| Wednesday | 699 | +0.055% | +0.173% | 2.787% | 54.1% | 0.52 | 0.604 |
| Thursday | 706 | +0.210% | +0.141% | 2.709% | 54.4% | 2.06 | 0.040 |
| Friday | 692 | +0.017% | +0.122% | 2.381% | 53.6% | 0.19 | 0.853 |
Reading: days with p < 0.05:
- Thursday: +0.210% on average, 54.4% win rate, p=0.040, but it does not clear the Bonferroni threshold for 5 tests (0.0100). Across the two halves of the sample (2013–2019: +0.042%; 2019–2026: +0.379%) the sign holds.
2. Seasonality by month of the year
164 complete months (the current month is left out). Each month's return = last close of the month against the last close of the previous month.
| Month | n | Mean % | Median % | Std % | Win rate | t-stat | p-value |
|---|---|---|---|---|---|---|---|
| January | 13 | +2.19% | +1.76% | 10.75% | 53.8% | 0.73 | 0.477 |
| February | 14 | +1.91% | +2.67% | 7.37% | 64.3% | 0.97 | 0.350 |
| March | 14 | +1.61% | +2.99% | 21.59% | 57.1% | 0.28 | 0.784 |
| April | 14 | +2.51% | +2.35% | 10.74% | 57.1% | 0.88 | 0.397 |
| May | 14 | +4.37% | -0.09% | 24.97% | 50.0% | 0.65 | 0.524 |
| June | 14 | +2.14% | +4.43% | 8.68% | 64.3% | 0.92 | 0.373 |
| July | 14 | +0.27% | +0.30% | 11.47% | 57.1% | 0.09 | 0.930 |
| August | 14 | -1.40% | -0.54% | 5.58% | 50.0% | -0.94 | 0.363 |
| September | 14 | +0.06% | -2.00% | 7.30% | 42.9% | 0.03 | 0.977 |
| October | 13 | -1.87% | -2.68% | 7.81% | 38.5% | -0.86 | 0.405 |
| November | 13 | -4.10% | -3.65% | 12.84% | 30.8% | -1.15 | 0.272 |
| December | 13 | +0.48% | +4.94% | 9.98% | 53.8% | 0.17 | 0.865 |
Reading: no month reaches p < 0.05. The best on average is May (+4.37%) and the worst November (-4.10%), but with 14 observations per month a single extreme year moves those averages a lot.
Each thin line is one year; the red one is the mean. When the thin lines fan out, a month's average says little about next year.
3. Seasonality by trading session (UTC)
Fixed UTC windows, the same across the whole series. Drift = session close against its open; range = high minus low, as % of the open.
| Session | n | Mean drift % | Win rate | p-value | Mean range % |
|---|---|---|---|---|---|
| Asia (00–07 UTC) | 2,264 | +0.0004% | 51.4% | 0.985 | 1.254% |
| London (07–12 UTC) | 2,265 | -0.0160% | 50.6% | 0.558 | 1.701% |
| New York (12–17 UTC) | 2,266 | +0.0284% | 52.4% | 0.472 | 2.472% |
| NY late (17–22 UTC) | 2,192 | +0.0139% | 52.1% | 0.575 | 1.525% |
- New York (12–17 UTC) is where it moves: mean range 2.472%, 2.0× that of Asia (00–07 UTC).
- No session has a significant drift: it moves, but with no preferred direction.
4. Day × session heatmap
| Day | Asia (00–07) | London (07–12) | New York (12–17) | NY late (17–22) |
|---|---|---|---|---|
| Monday | -0.100% | -0.041% | +0.011% | +0.032% |
| Tuesday | +0.057% | -0.084% | -0.102% | -0.019% |
| Wednesday | -0.031% | +0.035% | +0.083% | -0.014% |
| Thursday | +0.053% | +0.016% | +0.178% | -0.028% |
| Friday | +0.023% | -0.006% | -0.029% | +0.102% |
- Most bullish cell: Thursday × NY (+0.178%, p=0.039). Most bearish: Monday × Asia (-0.100%, p=0.068).
- 2 of 20 cells come out at p < 0.05; chance alone would give about 1.0. That is what chance produces: do not read the heatmap as a signal map.
5. Most volatile hours (UTC)
| UTC hour | Mean range % | Mean range (cents) | Std % | Session |
|---|---|---|---|---|
| 14:00 | 1.246% | 83.7 | 1.007 | NY |
| 13:00 | 1.161% | 78.6 | 0.995 | NY |
| 15:00 | 1.146% | 77.5 | 0.811 | NY |
| 16:00 | 0.957% | 63.9 | 0.800 | NY |
| 18:00 | 0.905% | 57.9 | 1.477 | NY late |
| 17:00 | 0.871% | 57.8 | 1.344 | NY late |
The busiest hour (14:00 UTC) moves 4.2× more than the quietest (04:00 UTC, 0.297%). Unlike the calendar bias, this is stable year to year: it is the most useful part of this analysis for deciding when to be at the screen or where to place a stop.
6. What it costs to trade: the measured spread
The feed carries real bid and ask, so the spread is measured, not a broker's marketing figure. Median since 2013: 5.00 cents. But a spread in cents does not say how heavy it is: what matters is how much of the hour's typical move it eats.
| UTC hour | Median spread (cents) | Mean range (cents) | Cost |
|---|---|---|---|
| 14:00 | 5.00 | 75.5 | 6.6% |
| 15:00 | 5.00 | 69.8 | 7.2% |
| 13:00 | 5.00 | 68.8 | 7.3% |
| 21:00 | 5.90 | 20.1 | 29.3% |
| 04:00 | 5.00 | 17.1 | 29.2% |
| 03:00 | 5.00 | 20.6 | 24.3% |
Trading at 14:00 UTC costs 6.6% of the hour's range; at 21:00 UTC, 29.3%. The cross-market comparison is in the real spread by market and hour.
What this data does not tell you
- Multiple comparisons: this article runs 41 tests (5 days, 12 months, 4 sessions and 20 cells). At a 5% threshold about 2 would come out "significant" by chance. That is why what survives Bonferroni is flagged separately.
- Few years per month: each month has about 14 observations. One extreme year can create or erase a "bullish month".
- An average bias is not a strategy: after spread, commission and swap, a drift of hundredths of a percent per session rarely survives. Before trading it, backtest with real costs and walk-forward.
- This is a Dukascopy CFD (bid price), not the exchange future or cash index: it can differ in rollovers, gaps and trading hours.
Methodology
- Source: Dukascopy ticks aggregated to M1 candles by EV Trading Labs (WTI, 2013-01-01 → 2026-10-12).
- Total: 4,555,284 M1 candles with at least one tick; 3,501 daily and 164 monthly returns.
- Price: open and close at mid price (bid+ask)/2; high and low on the bid. Where the feed has no ask, bid.
- Daily return: (C[t] − C[t−1]) / C[t−1] × 100, where C is the close at 19:00 UTC. Days with fewer than 60 candles, or whose last candle is more than 3 h before the cut-off, are dropped.
- Monthly return: last daily close of the month against the previous month's; the current month is excluded.
- Sessions: drift = (close − open) / open × 100; range = (high − low) / open × 100; sessions with less than 40% of their minutes are dropped.
- Test: one-sample t-test against zero, p < 0.05; Bonferroni per block (days, months, sessions, cells). Stability: the same mean computed on each half of the sample.
- Software: Python, pandas, scipy, matplotlib.
The full seasonality series
The same method applied to 19 markets. The contrast between them is the most useful part: a pattern that shows up in a single asset is more likely to be noise.
- Forex: EURUSD · GBPUSD · USDJPY · GBPJPY · AUDUSD · USDCHF · EURGBP
- Metals: XAUUSD (gold) · XAGUSD (silver)
- Indices: Dow Jones · NASDAQ-100 · S&P 500 · DAX 40 · FTSE 100 · Nikkei 225
- Oil: Brent
- Crypto: Bitcoin · Ethereum