A raw closing price chart lies to you in a specific, avoidable way. When a company splits its stock two for one, every share doubles and every historical price on the chart before that split has to be halved retroactively, or the chart shows a price crash that never actually happened. The same problem shows up with dividends: a stock that pays out cash reduces its own price by roughly that amount on the ex-dividend date, which looks like a loss to anyone reading raw prices even though the shareholder received that value directly. The fix is the adjusted close, a price series recalculated backward from the present using a cumulative adjustment factor built from every split ratio and dividend payment in the stock’s history, so that a straight percentage change between any two adjusted prices reflects the actual return an investor earned rather than an artifact of corporate bookkeeping.
Once the price series itself is trustworthy, the next distortion is inflation. A nominal return of 8% in a year when prices across the economy rose 4% is not really an 8% gain in purchasing power and the two returns should never be compared as if they were on the same scale. The real return corrects for this using the Fisher equation, r_{real} = \frac{1 + r_{nominal}}{1 + i} - 1, where $i$ is the inflation rate over the same period. An 8% nominal return against 4% inflation works out to a real return of about 3.8%, not the 4 points that simple subtraction would suggest, because inflation compounds against the return rather than simply subtracting from it. This is also why market-moving economic releases, jobs reports, GDP and retail sales, are published seasonally adjusted: a statistical procedure strips out predictable calendar patterns like holiday hiring so that the number reflects genuine economic movement rather than the time of year, which is exactly the same instinct as adjusting a stock price for a split.
With a clean, inflation-aware return in hand, the more interesting statistical question is whether that return was actually good, which depends entirely on how much risk was taken to get it. The Sharpe ratio answers this by dividing excess return over the risk-free rate by the standard deviation of that return, S = \frac{R_p - R_f}{\sigma_p}. A portfolio returning 10% against a 2% risk-free rate with 15% volatility produces a Sharpe ratio of about 0.53 and that single number is what makes it possible to rank two portfolios with completely different return figures on equal footing, since a higher, more volatile return can easily score worse than a lower, steadier one.
Standard deviation punishes upside volatility exactly as much as downside volatility, which is a distortion in its own right, since no investor is actually bothered by a return that occasionally jumps higher than expected. The Sortino ratio corrects for this by replacing total standard deviation with downside deviation, calculated only from returns that fall below a minimum acceptable threshold, so two portfolios with identical Sharpe ratios can have very different Sortino ratios if one of them earns its volatility mostly on the upside.
Beta adjusts for a different kind of risk: not a stock’s own volatility, but how much of the broader market’s movement it amplifies or dampens. The capital asset pricing model uses beta to set an expectation for what a stock should have returned,E(R) = R_f + \beta(R_m - R_f), so a stock with a beta of 1.2 in a year when the market returned 8% against a 2% risk-free rate has an expected return of about 9.2%. Two more adjusters follow directly from that expectation: the Treynor ratio, which divides excess return by beta instead of standard deviation to isolate reward per unit of market risk and Jensen’s alpha, which is simply the actual return minus the CAPM-expected return, so a stock returning 11% against a 9.2% expectation has an alpha of about 1.8 percentage points of performance that beta alone cannot explain.
None of these adjustments individually settles the question of whether a stock or a portfolio actually performed well. What they do collectively is strip out, one at a time, every source of distortion that has nothing to do with genuine performance: corporate actions from the price series, inflation from the return, market-wide risk from the comparison and volatility’s asymmetry from the risk measure itself. A number that survives all four adjustments is about as close to a clean measure of skill as market statistics get.