Alpha is a measure of how much a portfolio, fund or trade outperforms its benchmark after allowing for the market’s movement. Positive alpha means it did better than expected given the market backdrop. Negative alpha means it fell short.
In plain terms, alpha is the bit of return that cannot be explained by riding the market. It tries to isolate the manager’s selection skill, timing or strategy edge, rather than simple exposure to an index.
What investors mean by alpha in everyday use
When investors say a fund has alpha, they usually mean it has added value beyond what a broad market index would have delivered. The benchmark could be a share index, a bond index, a sector basket or even a blend of indexes that reflects the fund’s remit. If a global equity fund earns 9% while its global equity benchmark earns 6%, many people will call the 3% gap alpha.
Strictly speaking, finance textbooks define alpha as the intercept from a regression of portfolio returns on benchmark returns. That version controls for the portfolio’s sensitivity to the market, known as beta. In practice, both ideas appear in reports. Some teams publish simple outperformance, others publish modelled alpha that adjusts for beta and the risk‑free rate.
How alpha is calculated, from quick maths to CAPM
There are two common ways to compute it.
Quick estimate: Alpha equals portfolio return minus benchmark return over the same period. If your strategy returned 4% in a month and the benchmark returned 2%, the simple alpha is 2% for that month.
Model-based estimate: Alpha equals the realised portfolio return minus the return you would have expected given the risk‑free rate and the portfolio’s beta to the benchmark. Using the Capital Asset Pricing Model, the formula is:
Alpha = Rp − [Rf + Beta × (Rb − Rf)]
Where Rp is the portfolio return, Rb is the benchmark return, Rf is the risk‑free rate and Beta measures how sensitive the portfolio is to the benchmark. If a fund earned 10% in a year, the benchmark earned 7%, the risk‑free rate was 2% and the fund’s beta was 1.2, the expected return would be 2% + 1.2 × (7% − 2%) = 8%. The alpha would be 10% − 8% = 2%.
Analysts often estimate beta using a regression over daily or weekly returns, then take the regression intercept as alpha. The point is to separate what came from market exposure from what came from decisions unique to the strategy.
Picking benchmarks, horizons and units
Alpha is only as meaningful as the benchmark and period you choose. A UK small‑cap manager should not be measured against a global large‑cap index. Multi‑asset funds often use a custom blend of equity and bond indexes. Even for a single-stock trader, the chosen yardstick can change the story. Comparing a tech-heavy portfolio with a broad market index can flatter or punish the result depending on what the sector did.
Time horizon matters too. Alpha can swing around month to month, especially for concentrated or illiquid strategies. Many managers publish rolling 12‑month or 36‑month numbers to smooth noise. Results are commonly annualised so that different periods can be compared on the same footing.
Gross or net of fees also matters. Alpha quoted after fees is usually what end investors care about, since fees and trading costs reduce the value actually delivered. Some factsheets show both.
Alpha, beta and how it differs from other metrics
Alpha and beta go together. Beta captures how much a portfolio tends to move when its benchmark moves. A beta near 1 suggests market‑like sensitivity, more than 1 suggests amplified moves, and less than 1 suggests damped moves. Alpha is the extra return that is not explained by that sensitivity.
Do not confuse alpha with absolute return. A fund can post a loss yet still have positive alpha if it lost less than you would expect given a falling market and its beta. Likewise, a fund can make money but still have negative alpha if a hot market did most of the lifting.
Alpha also differs from the Sharpe ratio. The Sharpe ratio measures return per unit of total volatility, without reference to a benchmark. Alpha is benchmark‑relative and focuses on what is left after controlling for market exposure. Both can be useful, but they answer different questions.
Where you will see alpha quoted
You will meet alpha in fund factsheets, manager letters, risk reports and performance attribution. Quant teams often talk about alpha at the level of signals or factors, for example a value or momentum tilt that aims to add a few percentage points a year beyond the index. Hedge funds that run market‑neutral portfolios focus almost entirely on alpha, since their goal is to cancel out beta and leave only stock selection or strategy returns.
In crypto trading, the same logic applies. A trader might hedge an altcoin long with a short in a broad crypto index or in bitcoin to target a beta near zero, then try to harvest alpha from relative value, liquidity provision or event‑driven bets. The benchmark choice could be bitcoin, a sector basket or a composite index, depending on the strategy.
Examples that show alpha in action
Example 1, simple difference: A thematic equity fund gains 12% over a year. Its benchmark index rises 9%. Using the quick method, alpha is 3% for the year. If that fund had a beta very close to 1 over that period, the model‑based alpha would likely be close to 3% as well.
Example 2, beta‑adjusted: A concentrated growth fund returns 5% in a quarter. The benchmark returns 2% and the fund’s beta has been 1.5. Ignoring the risk‑free rate for simplicity over a short period, the expected return would be roughly 1.5 × 2% = 3%. The fund delivered 2% more than expected, so quarterly alpha is about 2%. Annualising that would depend on compounding and is not just four times the number.
Example 3, market‑neutral: A long short strategy earns 1% in a month with a measured beta of 0.1 to the equity index. If the index rose 3%, the expected contribution from beta was about 0.3%. The remainder, about 0.7%, is alpha attributed to selection.
Limitations and common traps when reading alpha
- Benchmark fit: A poor benchmark can make alpha look better or worse than it really is. Multi‑factor models can help, but they add assumptions.
- Changing betas: Beta is not fixed. If a strategy’s sensitivity shifts over time, a single beta estimate can misstate alpha.
- Short samples: Alpha measured over a few months is mostly noise. Statistical confidence builds with longer, cleaner data.
- Fees and frictions: Trading costs, taxes and borrow fees eat into realised alpha. Check whether the number is net or gross.
- Survivorship and selection: Looking only at funds that still exist exaggerates historical alpha. Include defunct peers when doing research.
- Data snooping: Strategies engineered on past data can show great alpha in backtests that fails to appear in live trading.
Alpha is a useful lens, not a verdict. It tells you how a result compared with what a chosen yardstick would imply, after allowing for market exposure. Read it alongside beta, volatility, drawdowns and the investment process itself to judge whether any edge is likely to persist.