XLE forward 1‑week return after steepening vs. flattening of 2s10s spread
The yield curve's slope is the market's shorthand for the macro outlook: when the 2s10s spread steepens, the read is stronger growth ahead, which should favor cyclical sectors like energy. This study put that logic to work on XLE, testing whether a daily steepening in the Treasury curve predicts above-baseline forward one-week returns over the past three years. The answer, in short: no.
Steepening days did see a slightly higher average five-day return than flattening days — roughly 0.36% versus 0.27%, against a 0.31% all-day baseline. But the gap is minuscule relative to XLE's typical weekly volatility, and the test statistics place it firmly in noise territory, with a p-value of 0.71. A result this weak is what you'd expect from a coin flip, not a growth-signal trade.
The full evidence, including methodology and distribution charts, is in the analysis below. The thesis fails on the merits: the steeper-curve-lifts-energy narrative finds no meaningful support in this sample.
For XLE over the past ~3 years, does a steepening of the 2s10s Treasury spread predict above-baseline forward one-week returns? Thesis: A steepening curve signals improving economic growth expectations, benefiting energy stocks, so XLE rallies in the following week.
How this was measured
Resampled XLE minute bars to daily close and computed forward 5‑trading‑day close‑to‑close return (ret_fwd_5d). Reindexed treasury_2y_df and treasury_10y_df onto the XLE trading calendar with forward‑fill; spread = 10y − 2y (in percentage points). Daily change of the spread (Δspread = spread[t] − spread[t−1]) classifies each day as steepening (Δspread > 0) or flattening (≤0). Welch’s two‑sample t‑test compares the mean forward 5‑day returns in the two regimes. Pearson correlation between Δspread and forward 5‑day return provides a continuous measure. The sample reflects the available ~3‑year data range of the `trader` tier.
The key numbers
Reading the numbers
Steepening days averaged a 0.36% forward weekly return versus 0.27% on flattening days and a 0.31% baseline, an extra 0.09 percentage points. But the p-value of 0.71 means this gap is easily within random noise — the data do not support the thesis.
The charts
The left bar for steepening days is slightly taller than the flattening bar and the all-days bar: about 0.36% versus 0.27% and 0.31%. The difference looks like a small edge, but it is far too small relative to the day-to-day swings in XLE returns to be meaningful. The bar chart alone flatters the thesis; the statistics show the gap is not distinguishable from chance.
This scatter of 738 points shows daily changes in the 2s10s spread from about -0.13 to +0.14 percentage points against the next week's XLE return, which ranged from roughly -19% to +8%. There is no visible upward tilt: the cloud is flat, and the correlation is essentially zero at 0.016. In plain terms, a larger steepening on Monday does not predict a stronger XLE week.
The histogram of the 328 steepening days shows a wide, roughly bell-shaped spread of forward weekly returns, from about -11.7% up to +6.9%, with a mean of 0.36%. The key detail is the wide left tail: even on steepening days, XLE can lose far more than the average gain. So the average return is a weak summary; individual steepening days are all over the map.
For the 410 flattening days, the histogram has a slightly lower mean of 0.27% and a wider downside tail, reaching about -19.4% versus -11.7% on steepening days. The two histograms overlap heavily, meaning most flattening-day outcomes look like steepening-day outcomes. This overlap is why the regime split does not produce a reliable one-week forecast for XLE.
Forward 5‑day return summary by 2s10s steepening regime
| regime | N | mean | std |
|---|---|---|---|
| steepening (Δspread>0) | 328 | 0.0036 | 0.029 |
| flattening (Δspread≤0) | 410 | 0.0027 | 0.0334 |
| all days | 738 | 0.0031 | 0.0315 |
The takeaway
Over the past ~3 years, a daily steepening of the 2s10s Treasury spread did not predict above-baseline one-week returns for XLE. Steepening days were followed by an average +0.36% over the next five trading days, versus +0.27% after flattening days — an extra 0.09% relative to the all-day baseline of +0.31%. That edge is tiny next to the typical weekly swing of roughly 3%, and the statistics say it's indistinguishable from noise: the Welch test gives p=0.71, and the correlation between the spread change and forward return is +0.02 with p=0.67. In plain terms, there's about a 71-in-100 chance you'd see a gap this large by luck even if the true effect were zero. This is basically a coin flip, not a tradable signal. The thesis that a steeper curve lifts energy stocks the following week finds no meaningful support in this sample.
The fine print
- Steepening vs flattening is defined at the zero threshold; a wider filter (e.g., moves >1bp) would cut noise but shrink the sample.
- Forward 5-day returns overlap, so successive days aren't independent — the t-test likely overstates significance.
- Treasury yields are forward-filled and may arrive with a one-day lag; same-day alignment may not hold in real time.
- Single-ticker, single-threshold test with no controls for equity-market moves or oil shocks; don't generalize too far.