SQN Calculator (System Quality Number)

The System Quality Number (SQN) is a metric developed by trading educator Van K. Tharp to score the statistical quality of a trading system. It combines a system’s expectancy per unit of risk (R) with the consistency of its trade outcomes and the number of trades in the sample. A higher SQN means the system produces its average return with less variation — the edge is more reliably expressed trade-to-trade. SQN is dimensionless, making it useful for comparing systems across instruments, timeframes and position sizes.

SQN = (Expectancy_R / StdDev_R) × √min(N, 100). Expectancy_R is the average profit per trade in R-multiple units (E_R = p × payoff − q, where p is win probability and payoff = avgWin/avgLoss). StdDev_R is the standard deviation of the R-multiple distribution. The √N term scales with sample size up to a cap of 100 trades (Van Tharp’s SQN100 convention). Algebraically, SQN is the one-sample t-statistic of the R-multiple sample tested against a null mean of zero.

Van Tharp’s published qualitative bands are: below 1.6 — negative or insufficient edge; 1.6–1.9 — poor but tradeable; 2.0–2.4 — average; 2.5–2.9 — good; 3.0–4.9 — excellent; 5.0–6.9 — superb; 7.0 and above — possibly the Holy Grail. Most professional traders target a minimum SQN of 2.0 for a live system. These bands assume a sample of at least 30–100 trades; scores from fewer trades are unreliable.

A score of 2.0 or above is generally considered acceptable for a live trading system according to Van Tharp’s framework. Scores in the 2.5–3.0 range are good, and anything above 3.0 is considered excellent. However, SQN alone does not tell you the system is suitable — you also need to consider maximum drawdown, position sizing and whether the historical edge is likely to persist. A very high SQN on a small sample (fewer than 30 trades) may be misleading.

Van Tharp introduced the SQN100 convention so that systems with very large trade histories (hundreds or thousands of trades) cannot inflate their SQN score purely by sample size. Without the cap, adding more trades indefinitely increases the √N term and the SQN, even if the underlying edge is unchanged. Capping at N=100 (so the √N factor tops out at 10) allows fair comparison across systems with different trade counts. For N ≤ 100 the formula is mathematically equivalent to the one-sample t-statistic; for N > 100 it departs from strict statistical theory but serves Van Tharp’s comparative purpose.

Expectancy is the average profit per trade in R-multiples (E_R = win rate × payoff ratio − loss rate). SQN divides that expectancy by the standard deviation of trade outcomes and multiplies by √N. Expectancy answers 'how much do I make on average per trade?' SQN answers 'how consistently and reliably do I make that amount, scaled by my sample size?' A system can have high expectancy but low SQN if its trade results are highly variable (e.g. a few big winners scattered among many small losers). High SQN signals that the edge is robust and repeatable, not just driven by outliers.

Yes. If the expectancy E_R is negative (the system loses money on average), SQN will also be negative. A negative SQN means the strategy has a losing edge — it is destroying capital on a risk-adjusted basis. Van Tharp’s interpretation bands are designed for positive SQN values; a negative score simply means the system should not be traded live, or that the inputs need review.

SQN is mathematically undefined when the standard deviation of R-multiples is zero — that is, when every trade has the same outcome. In the two-outcome model this occurs when the win rate is exactly 0% (all losses) or exactly 100% (all wins). In either case there is no dispersion to measure quality against. The calculator returns null for SQN and displays it as undefined rather than emitting an infinity or error. Any real trading system with a mix of wins and losses will have a non-zero standard deviation and a computable SQN.

Van Tharp recommends at least 30 trades as a minimum for the score to be statistically meaningful, and considers 100 trades to be the target for a reliable assessment. This mirrors the t-distribution’s small-sample behaviour — with fewer than 30 trades the score can fluctuate wildly between samples and may not reflect the system’s true edge. Our calculator flags N < 30 as potentially unreliable. For live system evaluation, a fresh out-of-sample test of 100+ trades provides the most credible SQN estimate.

They are conceptually similar — both divide an average return by a standard deviation — but they are not the same. The Sharpe ratio uses currency returns relative to a risk-free rate and is annualised. SQN uses R-multiple returns (returns expressed in units of initial risk) with no risk-free subtraction, scaled by √min(N, 100). SQN is trade-count-based rather than time-based, making it instrument- and timeframe-agnostic, which is why Van Tharp preferred it for evaluating trading systems across different markets.

An R-multiple expresses each trade’s profit or loss as a multiple of the initial risk (R = the amount risked on the trade, typically entry-to-stop distance × position size). A 2R win means the trade made twice the amount originally at risk; a 1R loss means the stop was hit and the trade lost exactly the risk amount. Using R-multiples makes trade statistics currency-agnostic and position-size-agnostic — the same system traded with $1,000 or $100,000 per trade produces identical R-multiples. This is why SQN, which is computed from R-multiples, can be compared fairly across accounts of different sizes.

Profit factor (gross wins ÷ gross losses) measures the gross monetary ratio of winners to losers without any adjustment for sample size or consistency. SQN incorporates the dispersion of trade outcomes (via the standard deviation of R-multiples) and scales with the number of trades. Two systems can share the same profit factor but have very different SQNs if one achieves its profit through a few large outlier wins (high variance, low SQN) and the other through many steady winners (low variance, high SQN). SQN therefore gives a more complete picture of system quality than profit factor alone.