What Is Turnover?

In portfolio construction, turnover measures how frequently a strategy replaces its holdings over a given period. It is typically expressed as the fraction of the portfolio’s total value that is bought or sold during the period. A turnover of 100% (or 1.0×) means the entire portfolio was effectively replaced once; a turnover of 500% means it was replaced five times. For two-sided measurement, turnover counts both buys and sells; the one-sided convention counts only one direction. The specific convention matters when comparing numbers across platforms and papers.

Turnover is one of the most important reality checks for any backtest. A strategy that looks spectacular on paper may be completely unimplementable if it requires daily full-portfolio rebalance cycles. Every trade incurs costs — commissions, bid-ask spreads, market impact, and slippage — and these compound multiplicatively against returns. The relationship is roughly linear at low turnover but becomes punishing at high levels: López de Prado and others have documented that across quantitative strategies, higher turnover systematically erodes cents-per-share profitability. A useful heuristic from Kakushadze (2015) is that the number of independent alpha signals in a portfolio is bounded by volume / turnover — meaning that high-turnover strategies need vastly more liquidity to sustain their edge.

Closely related to turnover are rebalance volume (the dollar or share volume needed to execute each rebalance) and cost drag (the cumulative performance reduction from transaction costs). Together, these three metrics form a “fragility triangle” that separates robust live strategies from overfitting paper optima. A strategy should be evaluated with realistic turnover, estimated costs, and rebalance volumes before its Sharpe or CAGR are considered. The best practice is to simulate execution with volume-dependent market-impact models (e.g., Almgren–Chriss) and report net-of-cost returns alongside gross returns.

Formula

One-sided Turnover = (1/2) × Σ |w_i,t - w_i,t-1|

where w_i,t is the weight of asset i at time t

Pros

  • Directly quantifies implementation burden and cost exposure

  • Catches overfitting — overfit strategies tend to have unnaturally high turnover

  • Easy to compute from any backtest with position weights

  • Pairs naturally with cost drag analysis for net-of-cost evaluation

Cons

  • One-sided vs. two-sided conventions can cause confusion when comparing across sources

  • Raw turnover does not distinguish between high-cost and low-cost trades (e.g., large-cap vs. small-cap)

  • A low-turnover strategy is not automatically good — it may simply be stale or unresponsive

  • Does not capture timing of trades within a rebalance period (e.g., concentrated vs. spread execution)

Kakushadze — Performance v. Turnover (arXiv).

See also