NVDA vs AMD daily-return correlation — last 3 months
Despite both selling GPUs into overlapping markets, NVDA and AMD showed only a modest positive relationship over the last three months. We measured close-to-close daily percent-change returns on 63 overlapping trading days (2026-03-02 to 2026-05-29) and found a Pearson correlation of 0.2707 and a Spearman rank correlation of 0.3452 — indicating a weak-to-moderate positive link rather than tight coupling.
The analysis below reproduces the resampling and return calculations, reports a 20-day rolling Pearson series, and documents the time variation in co-movement. The slightly higher rank correlation implies clearer monotonic co-movement than strict linear alignment; the rolling series shows the relationship is not stable, so read the full charts and numbers to see when the link strengthened or weakened.
How correlated are NVDA and AMD daily returns over the last 3 months?
How this was measured
Resampled minute OHLCV for NVDA and AMD to daily closes, computed close-to-close daily percent-change returns, then restricted to the most recent 3 calendar months of overlapping trading days. Reported full-sample Pearson and Spearman correlations over this window and a 20-trading-day rolling Pearson correlation series to show time variation within the window.
The key numbers
Reading the numbers
Over the 63 trading days the Pearson correlation is about 0.2707 (a weak-to-moderate positive linear relationship), while the Spearman rank correlation is about 0.3452 (a bit stronger when you look at ranks rather than raw returns).
The charts
The 20-day rolling Pearson line starts very high (first value 0.7243) and falls toward the end (last value 0.1327), with values across the series ranging from a low of -0.1261 to a high of 0.8084 and a mean of 0.3309 across 44 windows. The detail to watch is the large drop from those early strong correlations down into much weaker territory and the occasional negative dip, which shows the relationship is not stable month-to-month. In plain terms, there were periods of strong co-movement but by the end of the three months the short-term linear link had weakened, consistent with a modest average correlation.
The scatter of 63 daily pairs clusters tightly around zero for both stocks (NVDA returns range -0.0566 to 0.07 with mean 0.0031; AMD returns range -0.0659 to 0.2065 with mean 0.0162). Look for the dense cloud near the origin and a few AMD upside outliers up to 0.2065; that spread and concentration are why the cloud shows only a mild upward tilt. This visual matches the headline correlations (Pearson ~0.2707, Spearman ~0.3452): there is a small positive relationship, but most days the two names move independently rather than tightly together.
Correlation summary (last 3 months)
| window_start | window_end | n_days | pearson | spearman |
|---|---|---|---|---|
| 2026-03-02 | 2026-05-29 | 63 | 0.2707 | 0.3452 |
The takeaway
Short answer: NVDA and AMD showed a modest positive relationship over the last three months. The full-sample Pearson correlation is about 0.2707 and the Spearman rank correlation is about 0.3452 across 63 overlapping trading days (2026-03-02 to 2026-05-29). Those magnitudes point to a weak-to-moderate positive link — not zero, but far from a strong coupling. The rank-based measure being a bit higher (0.3452) suggests monotonic co-movement is clearer than strict linear alignment. The 20-day rolling correlation also moves around within the window, so the link is time-varying rather than stable. Practical takeaway: treat NVDA and AMD as mildly positively correlated over this period — useful as a directional signal, but not a dependable hedging substitute without ongoing monitoring.
The fine print
- Three-month window is short and regime-dependent; the signal can shift over longer horizons.
- Correlation is computed on daily close-to-close returns and ignores intraday co-movement.
- Values come from 63 overlapping trading days; that's a modest sample, so confidence is limited.
- Rolling correlation used a 20-day window — different window choices will change short-run behavior.