AI Research METASPYmacro:treasury_2y

META daily returns driven by 2-year Treasury yield changes (past 3 years)

741
Trading days analyzed

For all the talk about META being a rate-sensitive stock, the data over the past three years tells a different story. Daily changes in the 2-year Treasury yield show essentially zero correlation with META’s daily returns—a Pearson r of 0.003 and a p-value of 0.94. The hypothesized inverse relationship simply isn’t there.

The numbers are stark: yield changes explain 0.0007% of META’s daily variation. Instead, the S&P 500 is the dominant driver, with a beta near 1.63 and an R² of 87%. Even when we split days by yield direction, the return differences are indistinguishable from noise. The full breakdown—regression tables, tercile comparisons, and charts—is below.

The research question

For META over the past ~3 years, do daily changes in the 2-year Treasury yield drive daily returns? META's returns are strongly inversely correlated with 2-year yield moves, so every basis-point rise in short-end rates punishes the stock.

How this was measured

Resampled META and SPY minute bars to daily closes; computed META daily return and SPY daily return (close-to-close). Obtained daily 2-year Treasury yield from treasury_2y_df (percent), forward-filled onto META trading days, and computed day-over-day change in basis points (bps). Analyzed the relationship via Pearson correlation, simple linear regression (META return ~ yield change), and multiple linear regression with SPY return as a control. Also split yield changes into terciles (Down/Flat/Up) and compared mean META returns across regimes via Welch t-test. Running the analysis with three years of data (trader tier, 36-month cap).

The key numbers

Trading days analyzed
741
overlapping yield + price data
Pearson r (META ret vs Δ2Y bps)
0.0026
|r|=0.0026 ≤ 0.2 → weak association
Pearson p-value
0.9447
p=0.9447 ≥ 0.05 → correlation not statistically clear
Simple beta (per bp)
0.001056%
daily META return change per 1 bp yield rise
Simple regression p-value
0.9447
p=0.9447 ≥ 0.05 → not significant alone
R² (simple model)
0.0007%
fraction of META variance explained by Δ2Y alone
Full model beta (per bp)
-0.008568%
META return sensitivity to Δ2Y after controlling for SPY
Full model p-value (yield)
0.1214
p=0.1214 ≥ 0.05 → yield effect disappears when market is included
Full model beta (SPY)
1.6347
META's market beta in the presence of yield changes
Full model p-value (SPY)
0.0000
p=0.0000
R² (full model)
86.8739%
fraction explained by Δ2Y + SPY together
Down-tercile mean return
0.2127%
N=266 days with falling yields
Flat-tercile mean return
0.1467%
N=239 days with flat yields
Up-tercile mean return
0.0034%
N=236 days with rising yields
Welch t (Down vs Up terciles)
0.875
positive = Down days better than Up days
Welch p-value (Down vs Up)
0.3818
p=0.3818 ≥ 0.05 → no clear yield-direction effect

Reading the numbers

The correlation between daily META returns and 2-year yield changes is essentially zero (r = 0.003, p = 0.94), and even after accounting for the overall market the yield effect remains insignificant (p = 0.12). There is no statistical evidence that yield moves drive META's daily returns.

The charts

META daily return vs 2Y yield change (bps)
What this chart says

This scatter plot shows every trading day as a dot, with daily change in the 2-year yield on the x-axis and META's daily return on the y-axis. The dots form a shapeless cloud — there is no upward or downward slope, meaning a day with a big yield move is just as likely to have a positive META return as a negative one. For the question at hand, this visual confirms what the near-zero correlation number said: yield changes and META returns are unrelated.

Mean META return by Δ2Y tercile
What this chart says

This bar chart groups the 741 trading days into thirds based on how the 2-year yield moved: falling yields (down), roughly unchanged (flat), and rising yields (up). If the user's belief were correct, the 'Up' bar would be clearly negative. Instead, the mean META return on rising-yield days is essentially zero (0.0%), and the pattern across all three groups is flat and tiny — a far cry from a strong inverse relationship.

Regression estimates

ModelCoefficientEstimatep-value
SimpleΔ2Y (bps)1.1000e-050.94470
FullΔ2Y (bps)-8.6000e-050.12140.8687
FullSPY ret1.63470

Tercile summary

RegimeNMean returnStd return
Down2660.00210.0253
Flat2390.00150.0179
Up23600.028

The takeaway

The data show no meaningful relationship between daily changes in the 2-year Treasury yield and META's daily return over the past three years. The correlation between the two is essentially zero (r = 0.003, p = 0.94), and a simple regression finds that yield changes explain exactly 0% of the variation in META's return. Even after controlling for the S&P 500, the yield coefficient remains insignificant (p = 0.12), while the market return itself is a massive driver (beta ≈ 1.63, p < 0.001, R² = 87%). Days when yields fell produced an average return of +0.21%, versus +0.00% on up days, but that difference is easily explained by chance (p = 0.38). The result is clear and conclusive: the hypothesized inverse link between short-end yields and META's daily return does not exist in this data. The practical takeaway is that META's day-to-day moves are overwhelmingly explained by the broader market, not by the 2-year yield—so betting on a yield-driven trading signal would have been a losing proposition.

The fine print