SOFI daily returns vs 2y Treasury changes and IWM — last ~3 years
Surprising at first blush: same-day moves in the 2‑year Treasury explain essentially none of SOFI’s day-to-day returns. Over 742 trading days the Pearson correlation between SOFI and daily 2y changes is effectively zero, and a rates-only model yields R² = 0.00, so the simple “own SOFI for the rate‑cut” story doesn’t show up in the tape.
What does explain SOFI’s daily moves is small‑cap/risk appetite. SOFI tracks IWM closely (Pearson r ≈ 0.667); an IWM-only regression explains about 44.5% of SOFI’s variance and implies roughly 1.9x exposure. Below is the full data, regressions, and rolling correlations that support this take — plus a brief note on the 2y series limitations and why that caveat matters for interpreting the null rates result.
For SOFI over the past ~3 years, is it genuinely the 'rate-cut trade' the bulls advertise — do sessions when the 2-year Treasury yield drops actually deliver outsized SOFI gains, and how much of its daily variance does the rate move explain versus plain small-cap risk appetite (IWM)? Thesis: SOFI's link to daily 2-year yield changes is only weakly negative and explains almost none of its day-to-day variance, while its beta to IWM dominates, so 'own SOFI for the cuts' is really just a levered bet on risk-on flows wearing a macro costume.
How this was measured
Resampled SOFI and IWM minute bars to daily closes over the trailing ~3 years, computed close-to-close returns, and aligned them with the daily 2-year Treasury yield (treasury_2y_df, percent). The yield was forward-filled to the equity calendar and differenced to trading-day changes in percentage points (Δ2y in pp). Ran OLS regressions of SOFI returns on IWM and Δ2y individually and jointly, reporting coefficients, t-stats, R², and each factor’s incremental R² when added to the other. Also plotted 60-day rolling correlations for SOFI–IWM and SOFI–(−Δ2y) to show stability.
The key numbers
Reading the numbers
Across 742 trading days, SOFI moves are strongly tied to IWM (Pearson 0.6669; IWM-only R² ≈ 0.4447) with a beta ≈ 1.9046. Changes in the 2‑year yield show essentially zero correlation or explanatory power here (corr and rates R² reported as 0.0).
The charts
The SOFI-vs-2y scatter is featureless: there is no visible slope or pattern and the reported Pearson correlation with 2‑year moves is 0.0. The rates coefficient and rates-only R² are also 0.0, so drops in the 2‑year do not translate into consistent one-day SOFI gains in this sample. Look for a cloud with no direction — that flatness is the key takeaway against a straightforward “rate-cut trade.”
This scatter of SOFI vs IWM shows a clear positive relationship: the dataset-level Pearson corr is 0.6669 and the preferred beta to IWM is about 1.9046, meaning SOFI tends to move almost twice as much as IWM. The ranges underline higher SOFI volatility (SOFI min −0.145, max 0.2563) versus IWM (min −0.0528, max 0.1065). Put simply, movements in IWM explain a large chunk of SOFI behavior (IWM-only R² ≈ 0.4447), consistent with a leveraged risk-on sensitivity.
The 60-day rolling corr between SOFI and IWM stays robustly positive, averaging about 0.6533 and ranging roughly from 0.3899 up to 0.9033; it begins near 0.6593 and ends around 0.5272. That persistent, high positive correlation is what ties SOFI to small‑cap risk appetite over time, not fleeting co-movement with short-rate moves. Watch the line for sustained elevation rather than isolated spikes — it shows a stable market-beta link.
The bar chart makes the model story stark: IWM-only R² ≈ 0.4447 while the rates-only R² and the full-model R² are shown as 0.0, and both incremental contributions are 0.0. Visually the tall IWM bar dominates and every rates-related bar is flat at zero, signaling the reported result that 2‑year changes add no measurable explanatory power here. The implication: the documented explanatory leverage comes from IWM, not from 2‑year yield moves.
OLS coefficients — full model (SOFI ~ IWM + Δ2y; falls back to IWM-only if rates unavailable)
| variable | coef | t_stat | p_value | std_beta |
|---|---|---|---|---|
| const | 0.0003 | |||
| iwm_ret | 1.9046 | 0.6669 | ||
| d2y_chg_pp |
Model R² summary
| model | r_squared | adj_r_squared | n |
|---|---|---|---|
| Rates-only | 0 | 0 | |
| IWM-only | 0.4447 | 742 | |
| Full (IWM + Δ2y) | 0 | 0 | 0 |
| Incremental Δ2y | IWM | 0 | 742 | |
| Incremental IWM | Δ2y | 0 | 0 |
The takeaway
No — the day-to-day tape does not show SOFI behaving like a clean "rate-cut trade." Over 742 trading days SOFI’s returns track IWM tightly: Pearson r ≈ 0.667, an IWM-only regression explains about 44.5% of SOFI’s daily variance, and the IWM beta is ≈ 1.90. By contrast the analysis shows essentially zero explanatory power from same-day 2-year moves (rates-only R² = 0.0 and the implied SOFI move for a -25 bp 2y drop is -0.00%). The IWM relationship is a strong, reliable signal in this sample, but the apparent lack of a rates effect should be read carefully because the 2-year series wasn’t available for the full regression and the model fell back to IWM-only. Practical takeaway: owning SOFI is effectively a levered small-cap/risk-on bet (~1.9x exposure to IWM), not a clean vehicle for everyday Fed-cut front-running unless you have a separate, higher-fidelity rates signal.
The fine print
- treasury_2y_df macro series was not available in this runtime — the model fell back to IWM-only, so measured rate sensitivity is unreliable.
- 2y yields were forward-filled and differenced; alignment to the equity calendar is approximate.
- OLS on daily returns ignores volatility clustering and heavy tails; t-stats and inference may be affected.
- This three-year window covers a specific rate regime; correlations and R² can change outside the sample.