VanEck Semiconductor ETF vs Nvidia (SMH vs NVDA) 2026: Returns, Risk & Volatility

Q: Which is a better investment: SMH or NVDA?

Over the past year, SMH looks like the better investment: it had higher total return (SMH +127.0% vs NVDA +63.2%) and a higher Sharpe ratio (SMH 2.79 vs NVDA 1.53). On risk, SMH was less volatile (SMH 29.9% vs NVDA 33.3%), SMH had a smaller max drawdown (SMH -14.9% vs NVDA -20.2%), and SMH had less severe losses on the worst 5% of days (Expected Shortfall of SMH -3.99% vs NVDA -4.24%).

Written by Gale Finance Team

Reviewed by Sid Kalla CFA Charterholder

Quick answer

Which is a better investment: SMH or NVDA?

Over the past year, SMH outperformed NVDA. SMH returned +127.0% compared with NVDA’s +63.2%. SMH had the better risk-adjusted return, with a Sharpe ratio of 2.79 versus NVDA’s 1.53. SMH was less volatile than NVDA, and SMH had a smaller max drawdown than NVDA.

Total Return

SMH +127.0%

NVDA +63.2%

Sharpe Ratio

SMH 2.79

NVDA 1.53

Annualized Volatility

SMH 29.9%

NVDA 33.3%

Max Drawdown

SMH -14.9%

NVDA -20.2%

Metric winners: Total Return: SMH; Sharpe Ratio: SMH; Annualized Volatility: SMH (less volatile); Max Drawdown: SMH (smaller drawdown).

SMH Total Return

↑ +127.0%

NVDA Total Return

↑ +63.2%

Relative Performance of SMH vs NVDA (Normalized to 100)

SMH NVDA

Normalized to 100 at start date for comparison

Trade SMH or NVDA

Access these assets on trusted platforms.

Hyperliquid

Trade perpetual futures exposure where supportedSMHNVDATrade on Hyperliquid

Binance

Trade futures exposure on a global exchange where supportedSMHNVDATrade on Binance

Affiliate disclosure

Key Takeaways

Total Return: SMH delivered a +127.0% total return, while NVDA returned +63.2% over the same period. SMH outperformed on total returns.
Risk-Adjusted Return (Sharpe Ratio): SMH had a higher Sharpe (2.79 vs 1.53), indicating better risk-adjusted performance.
Volatility (Annualized): NVDA was more volatile, with 33.3% annualized volatility, versus 29.9% for SMH.
Maximum Drawdown: SMH's maximum drawdown was -14.9%, while NVDA experienced a deeper drawdown of -20.2%.
Tail Risk (VaR & Expected Shortfall): At the 5% level (daily log returns), SMH's VaR was -3.18% and its Expected Shortfall (CVaR) was -3.99%; NVDA's were -3.47% and -4.24%. VaR is the cutoff; Expected Shortfall is the average move on the worst days.
Skew & Kurtosis: Skew: SMH -0.20 vs NVDA -0.01. Excess kurtosis: SMH 0.81 vs NVDA 0.49. Negative skew leans downside; higher excess kurtosis means fatter tails.
Tail Days & Extremes: 2σ tail days (down/up): SMH 9/7, NVDA 8/5. Worst day: SMH -5.76% (2025-10-10) vs NVDA -5.46% (2026-02-26). Best day: SMH +5.76% (2026-03-31) vs NVDA +7.87% (2026-02-06).
Risk ratios: Sortino - SMH: 4.38 vs. NVDA: 2.33 , Calmar - SMH: 8.58 vs. NVDA: 3.15 , Sterling - SMH: 9.28 vs. NVDA: 2.94 , Treynor - SMH: 0.42 vs. NVDA: 0.29 , Ulcer Index - SMH: 3.88% vs. NVDA: 8.07%

Investment Comparison

If you invested $10,000 in each asset on May 14, 2025:

SMH $22,701.47 +127.0%

NVDA $16,315.94 +63.2%

Difference: $6,385.53 (SMH ahead)

VanEck Semiconductor ETF vs Nvidia Performance Over Time

Metric	SMH	NVDA
30 Days	28.5%	17%
90 Days	35.3%	16.2%
180 Days	63.3%	18.2%
1 Year	127%	63.2%

Shorter time frames can show different leaders as market conditions change. Consider your investment horizon when comparing performance.

VanEck Semiconductor ETF vs Nvidia Correlation

Average Correlation

strongly correlated

0.75

Current (30-day) 0.50

30-day rolling range +0.50 to +0.88

VanEck Semiconductor ETF and Nvidia are strongly correlated over the past year. With a correlation of 0.75, these assets tend to move together, limiting diversification benefits.

For portfolio construction, this strong correlation means holding both SMH and NVDA provides limited risk reduction — they're likely to decline together in downturns.

Metric	Value
Current (30-day)	0.50
Average (full period)	0.75
Minimum (30-day rolling)	0.50
Maximum (30-day rolling)	0.88

Correlation measures how closely two assets move together. Values near +1 indicate strong co-movement, near 0 indicates independence, and negative values indicate inverse movement. Current, minimum, and maximum figures are 30-day rolling correlations on shared daily returns.

Drawdown

Maximum Drawdown

SMH

-14.9%

NVDA

-20.2%

VanEck Semiconductor ETF experienced its maximum drawdown of -14.9% from 2026-02-25 to 2026-03-30. It took 10 days to recover.

Nvidia experienced its maximum drawdown of -20.2% from 2025-10-29 to 2026-03-30. It took 25 days to recover.

Smaller drawdowns and faster recoveries indicate lower downside risk and greater resilience during market stress.

VanEck Semiconductor ETF vs Nvidia Volatility (SMH vs NVDA)

SMH Volatility

29.9%

±1.89% 1-day vol

NVDA Volatility

33.3%

±2.1% 1-day vol

1-day volatility (1σ)

SMH

±1.89%

NVDA

±2.1%

VanEck Semiconductor ETF's 29.9% annualized volatility translates to about ±1.89% one-standard-deviation daily volatility.

Nvidia's 33.3% annualized volatility translates to about ±2.1% one-standard-deviation daily volatility.

NVDA had the wider volatility profile over this window. That means its day-to-day return distribution was broader; SMH was calmer, but lower volatility does not by itself mean better returns.

Treat the ± daily figure as a one-standard-deviation estimate from historical returns, not a forecast or expected absolute daily move. For context, 15-18% annualized volatility is roughly ±1% one-standard-deviation daily volatility.

Risk-adjusted ratios

Sharpe Ratio of SMH and NVDA

Sharpe Ratio: SMH vs. NVDA

Return per total volatility

Sharpe gives us excess return per unit of risk. Upside and downside volatility both count as risk.

Higher is better

excess return / total volatility

Formula

\displaystyle \mathrm{Sharpe} = \frac{\mathbb{E}[R] - R_f}{\sigma_R}

Sharpe ratio measures return per unit of risk (volatility). A higher Sharpe indicates better risk-adjusted performance. SMH had a higher Sharpe (2.79 vs 1.53), indicating better risk-adjusted performance.

A Sharpe above 1.0 is generally considered good, above 2.0 is excellent. Negative Sharpe means the asset underperformed the risk-free rate. Calculated on each asset's full 365-day lookback of available prices and annualized using the asset calendar (365 for crypto, 252 trading days for equities/ETFs/metals).

Sortino Ratio of SMH and NVDA

Sortino Ratio: SMH vs. NVDA

Return per downside volatility

Sortino keeps the return-over-risk idea, but only returns below the target rate count as volatility.

Higher is better

excess return / downside volatility

Formula

\displaystyle \mathrm{Sortino} = \frac{\mathbb{E}[R] - R_f}{\sigma_{\mathrm{down}}}

Sortino ratio measures return per unit of downside risk. Unlike Sharpe, it only counts downside deviation (returns below the target return). SMH had better downside-adjusted returns.

A higher Sortino is better. It's useful when upside volatility is common (crypto is the obvious example). Downside deviation: SMH 19.0% vs NVDA 21.9%. Calculated on each asset's full 365-day lookback of available prices, using the daily risk-free rate as the target return, and annualized using the asset calendar (365 for crypto, 252 trading days for equities/ETFs/metals).

Calmar Ratio of SMH and NVDA

Calmar Ratio: SMH vs. NVDA

CAGR per worst drawdown

Calmar compares CAGR against the single deepest peak-to-trough loss over the period.

Higher is better

CAGR / max drawdown

Formula

\displaystyle \mathrm{Calmar} = \frac{\mathrm{CAGR}}{|\mathrm{MaxDD}|}

Calmar ratio compares CAGR to maximum drawdown. Higher Calmar means more return per unit of worst drawdown. SMH posted the higher Calmar ratio.

Calmar is computed on each asset's full 365-day lookback and uses the max drawdown over that same window.

Sterling Ratio of SMH and NVDA

Sterling Ratio: SMH vs. NVDA

Return per average drawdown

Sterling smooths the drawdown penalty by using average drawdown events instead of only the worst one.

Higher is better

excess annual return / average deep drawdown

Formula

\displaystyle \mathrm{Sterling} = \frac{\mathrm{CAGR} - R_f}{\overline{D}_{>10\%}}

Sterling ratio measures excess return per unit of average drawdown (typically drawdowns worse than 10%). SMH posted the higher Sterling ratio.

Sterling uses average drawdown events deeper than 10% and subtracts the risk-free rate to report excess return.

Treynor Ratio of SMH and NVDA

Treynor Ratio: SMH vs. NVDA

Excess return per market beta

Treynor divides excess annualized return by beta — the sensitivity of the asset to broad-market moves. The slope shown is each asset’s beta vs SPY.

Higher is better

excess return / market beta

Formula

\displaystyle \mathrm{Treynor} = \frac{\mathbb{E}[R] - R_f}{\beta}

Treynor ratio measures excess return per unit of market risk (beta) instead of total volatility. SMH posted the higher Treynor ratio.

Treynor uses beta vs the S&P 500 (SPY) on shared dates and the average 3-month Treasury rate as the risk-free rate.

Ulcer Index of SMH and NVDA

Ulcer Index: SMH vs. NVDA

Drawdown pain

Ulcer Index is a risk index, not a return-over-risk ratio. Lower means smaller and shorter drawdowns.

Lower is better

root-mean-square drawdown

Formula

\displaystyle \mathrm{UI} = \sqrt{\mathbb{E}[D_t^2]}

Ulcer Index captures drawdown depth and duration. Lower Ulcer Index means less drawdown pain. SMH had the lower Ulcer Index (less drawdown pain).

Ulcer Index is computed from each asset's drawdown series over the full lookback window.

Tail Risk & Distribution Shape (1-Year): VanEck Semiconductor ETF vs. Nvidia

This section looks at the shape of daily returns, not just the average. Tail stats are computed per asset on its own daily series (crypto includes weekends). We use daily log returns $\ln\left(\frac{P_t}{P_{t-1}}\right)$ so multi-day moves add cleanly.

Definitions: Value at Risk (VaR), Expected Shortfall, skew, kurtosis, and fat tails.

Tail Risk & Distribution Shape: SMH vs. NVDA (1-Year)

Actual daily return tails

The bars are real daily log-return observations from the article window. Darker bars are observations at or beyond each asset’s 5% VaR cutoff.

Observed returns

VaR marks the 5th percentile loss cutoff; Expected Shortfall averages the observations beyond that cutoff.

Formula

\displaystyle \mathrm{VaR}_{5\%}=Q_{0.05}(r_t),\quad \mathrm{ES}_{5\%}=\mathbb{E}[r_t\mid r_t\le \mathrm{VaR}_{5\%}]

Metric (1-Year)	SMH	NVDA
5% VaR (daily log return)	-3.18%	-3.47%
5% Expected Shortfall (CVaR)	-3.99% (worst 13 days)	-4.24% (worst 13 days)
Skew	-0.20	-0.01
Excess kurtosis	0.81	0.49
2σ tail days (down / up)	9 / 7	8 / 5
Worst day	-5.76% (2025-10-10)	-5.46% (2026-02-26)
Best day	+5.76% (2026-03-31)	+7.87% (2026-02-06)

Downside co-moves (2σ) — 1-Year

Computed on shared dates only (n=249). A “2σ downside move” means a shared-close log return more than 2 standard deviations below that asset’s own mean on this shared-date series. Dates below show simple returns (%) for readability.

Downside co-move map: SMH vs. NVDA (2σ)

Shared-close daily returns

Dots mark actual downside days: asset-colored dots are one-sided downside moves, and red dots are joint downside days. Grey dots add sampled shared-return context when available. The shaded lower-left zone shows where both SMH and NVDA crossed their own 2σ downside threshold.

2σ

Show downside tail dates

Dates below are shared-date observations. The “Date” is the period end (close). Tail thresholds are computed on log returns, but the table shows simple returns (%) for readability. Returns are computed from the previous shared close to this one (for example, Friday → Monday includes weekend moves).

Days when both SMH and NVDA had a big down day (2σ)

Date (interval)	SMH	NVDA
2025-10-10	-5.76%	-4.89%
2025-11-04	-3.64%	-3.96%
2026-03-26	-4.56%	-4.16%

Days when SMH had a big down day

Date (interval)	SMH	NVDA
2025-10-10	-5.76%	-4.89%
2025-11-04	-3.64%	-3.96%
2025-11-20	-4.22%	-3.15%
2025-12-12	-4.52%	-3.27%
2025-12-17	-3.61%	-3.81%
2026-02-04	-3.94%	-3.41%
2026-03-03	-3.77%	-1.33%
2026-03-06	-3.74%	-3.01%
2026-03-26	-4.56%	-4.16%

Days when NVDA had a big down day

Date (interval)	SMH	NVDA
2025-10-10	-5.76%	-4.89%
2025-10-14	-1.84%	-4.40%
2025-11-04	-3.64%	-3.96%
2026-01-16 → 2026-01-20	-2.50%	-4.38%
2026-02-26	-3.32%	-5.46%
2026-02-27	-1.37%	-4.16%
2026-03-26	-4.56%	-4.16%
2026-04-30	+1.43%	-4.63%

Read this as “how ugly the ugly days get”, not as a precise forecast. One-year samples are small, so tail estimates are inherently noisy.

Full Comparison of VanEck Semiconductor ETF vs. Nvidia (1-Year)

Metric	SMH	NVDA
Total Return	+127.0%	+63.2%
Annualized Volatility	29.9%	33.3%
Sharpe Ratio	2.79	1.53
Sortino Ratio	4.38	2.33
Calmar Ratio	8.58	3.15
Sterling Ratio	9.28	2.94
Treynor Ratio	0.42	0.29
Ulcer Index	3.88%	8.07%
Max Drawdown	-14.9%	-20.2%
Avg Correlation to S&P 500	0.79	0.63
5% VaR (daily log return)	-3.18%	-3.47%
5% Expected Shortfall (CVaR)	-3.99%	-4.24%
Skew	-0.20	-0.01
Excess kurtosis	0.81	0.49
2σ tail days (down / up)	9 / 7	8 / 5

Audit this calculation

Formulas, inputs, and conventions used to compute the metrics on this page.

Inputs & conventions

Shared window for pair metrics: 2025-05-14 → 2026-05-12 (last shared close).
Rolling correlation sample (shared closes): 220 rolling 30-day values (from 249 shared daily returns).
Annualization (days/year): SMH: 252 days/year; NVDA: 252 days/year.
Risk-free rate: Uses the 3-month U.S. Treasury yield (FRED: DGS3MO), averaged over each asset’s window:

SMH: 4.15% over 2025-05-14 → 2026-05-12.

NVDA: 4.15% over 2025-05-14 → 2026-05-12.
Volatility drag (rule of thumb): Estimated from annualized volatility (simple returns). For the log-return framing, see Log returns.

SMH: ≈ -4.5%/yr

NVDA: ≈ -5.5%/yr
Data alignment: No forward fill. Correlation and tail co-moves are computed on shared closes only.
For cross-calendar pairs (e.g., crypto vs stocks), weekend/holiday moves roll into the next shared close.
Return conventions: Volatility/Sharpe/Sortino use simple daily returns. Tail-risk uses daily log returns for distribution stats (but tables show simple returns). Log returns.

Formulas

Daily simple return

r_t = \frac{P_t}{P_{t-1}} - 1

Volatility

\sigma_{ann} = \sigma(r_t)\sqrt{A}

Volatility drag (approx.)

\text{drag} \approx \tfrac{1}{2}\sigma_{ann}^2

Sharpe Ratio

S = \frac{A\,\bar{r} - r_f}{\sigma(r_t)\sqrt{A}}

Sortino Ratio

So = \frac{A\,\bar{r} - r_f}{\sqrt{\mathbb{E}[\min(0,\,r_t - r_f/A)^2]}\,\sqrt{A}}

Maximum Drawdown

MDD = \min_t\left(\frac{P_t}{\max_{s \le t} P_s} - 1\right)

Average Correlation

\rho = \frac{\operatorname{cov}(r^A,\,r^B)}{\sigma_A\,\sigma_B}

Log returns

\ell_t = \ln\left(\frac{P_t}{P_{t-1}}\right)

Notation

$P_t$: Price on day t.
$r_t$: Simple daily return.
$\ell_t$: Log daily return.
$\bar{r}$: Average daily return.
$\sigma(r_t)$: Standard deviation of daily returns.
$A$: Annualization factor (days/year).
$r_f$: Annual risk-free rate.

VanEck Semiconductor ETF vs Nvidia: Frequently Asked Questions

Which has higher volatility: SMH or NVDA?

NVDA showed higher volatility at 33.3% annualized, compared to 29.9% for SMH Over the past year. Higher volatility means larger price swings in both directions.

Does SMH provide diversification when held with NVDA?

SMH and NVDA are strongly correlated over the past year, with an average correlation of 0.75. This strong correlation limits diversification benefits.

How bad are the worst 5% days for SMH vs NVDA?

Over the past year, SMH's 5% VaR was -3.18% and its 5% Expected Shortfall was -3.99% (worst 13 days). NVDA's were -3.47% and -4.24% (worst 13 days).

Do SMH and NVDA crash together on bad days?

On shared dates (n=249), when NVDA has a 2σ down day, SMH also does 37.5% (3/8 days). In the other direction, when SMH has one, NVDA also does 33.3% (3/9 days).

Which has better risk-adjusted returns: SMH or NVDA?

SMH showed better risk-adjusted performance with a Sharpe ratio of 2.79 versus NVDA's 1.53 Over the past year.

Can SMH and NVDA be combined in a portfolio?

Yes, though allocation sizing matters. Their strong correlation provides limited risk reduction since they tend to move together. NVDA's higher volatility (33.3%) means even small allocations can materially impact overall portfolio risk.

Explore our financial glossary

Which is a better investment: SMH or NVDA?

Share This Comparison of SMH vs NVDA

Relative Performance of SMH vs NVDA (Normalized to 100)

Trade SMH or NVDA

Key Takeaways

Investment Comparison

VanEck Semiconductor ETF vs Nvidia Performance Over Time

VanEck Semiconductor ETF vs Nvidia Correlation

Drawdown

VanEck Semiconductor ETF vs Nvidia Volatility (SMH vs NVDA)

Risk-adjusted ratios

Sharpe Ratio of SMH and NVDA

Sharpe Ratio: SMH vs. NVDA

Sortino Ratio of SMH and NVDA

Sortino Ratio: SMH vs. NVDA

Calmar Ratio of SMH and NVDA

Calmar Ratio: SMH vs. NVDA

Sterling Ratio of SMH and NVDA

Sterling Ratio: SMH vs. NVDA

Treynor Ratio of SMH and NVDA

Treynor Ratio: SMH vs. NVDA

Ulcer Index of SMH and NVDA

Ulcer Index: SMH vs. NVDA

Tail Risk & Distribution Shape (1-Year): VanEck Semiconductor ETF vs. Nvidia

Tail Risk & Distribution Shape: SMH vs. NVDA (1-Year)

Downside co-moves (2σ) — 1-Year

Downside co-move map: SMH vs. NVDA (2σ)

Days when both SMH and NVDA had a big down day (2σ)

Days when SMH had a big down day

Days when NVDA had a big down day

Full Comparison of VanEck Semiconductor ETF vs. Nvidia (1-Year)

Inputs & conventions

Formulas

VanEck Semiconductor ETF vs Nvidia: Frequently Asked Questions

Which has higher volatility: SMH or NVDA?

Does SMH provide diversification when held with NVDA?

How bad are the worst 5% days for SMH vs NVDA?

Do SMH and NVDA crash together on bad days?

Which has better risk-adjusted returns: SMH or NVDA?

Can SMH and NVDA be combined in a portfolio?