Metrics

What is System Quality Number (SQN)?

Reviewed by Karthikeyan Anbarasan (Founder, TradeLyser) · Updated 19 Oct 2026 · Editorial policy

DefinitionThe System Quality Number (SQN), developed by Van Tharp, scores a trading system by dividing its average R-multiple by the standard deviation of R-multiples and multiplying by the square root of the number of trades. Fifty illustrative trades averaging +0.34R with a standard deviation of 1.56R give an SQN of about 1.54; at ₹2,000 risked per trade that average is ₹680 a trade.
Formula
SQN = √N × (Average R-multiple ÷ Standard deviation of R-multiples)
Example
√50 × (0.34 ÷ 1.56) = 7.07 × 0.218 = 1.54

What goes into SQN

SQN combines three things you can read from a trading journal. The average R-multiple is expectancy in units of risk. The standard deviation of R-multiples measures how scattered the outcomes are. The square root of the trade count rewards a larger sample. The result looks like a t-statistic: it asks how far the average sits above zero compared with the noise around it. Every trade must be logged as an R-multiple: profit or loss divided by the amount risked at the initial stop. A trade that risked ₹2,000 and made ₹4,000 is +2R; one that lost ₹2,000 is −1R.

A worked example: 50 trades

ResultTradesR total
−1R27−27R
+1R8+8R
+2R9+18R
+3R6+18R
Total50+17R
  1. Average R = +17R ÷ 50 = +0.34R. At ₹2,000 per R that is ₹680 a trade before charges.
  2. Standard deviation of R: the sum of squared R values is 125, minus 50 × 0.34² (5.78) leaves 119.22; divided by 49 that is 2.43, and its square root is 1.56R.
  3. Square root of trade count: √50 = 7.07.
  4. SQN = 7.07 × 0.34 ÷ 1.56 = 1.54.
Average R
+0.34R
Standard deviation
1.56R
SQN at 50 trades
1.54
Same stats at 100 trades
2.18

The last tile shows the catch. With the same average and spread, doubling the trades lifts SQN from 1.54 to 2.18 purely through √N. For that reason Tharp’s later work caps N at 100 when scoring systems, so a strategy that trades very often is not ranked higher simply for trading more.

Comparing setups in a journal

Opening range breakout84 trades+0.32RVWAP reclaim41 trades+0.11RExpiry-day scalps63 trades−0.18R
Figure 1. Illustrative expectancy per setup tag: opening range breakout +0.32R over 84 trades, VWAP reclaim +0.11R over 41 trades, expiry-day scalps −0.18R over 63 trades. If each tag’s R-multiples had a standard deviation of 1.5R, SQN would be about 1.96, 0.47 and −0.95.
Below 1.6
Tharp regarded systems here as hard to trade; the edge is small next to the noise.
1.6 to 2.9
His bands run from below average (1.6 to 1.9) through average (2.0 to 2.4) to good (2.5 to 2.9).
3.0 and above
Excellent in his scale, with higher bands beyond 5. Scores this high on a short record usually mean the sample is too small.

These bands are Tharp’s rules of thumb, not a standard, and they assume R is measured consistently. They are most useful for ranking your own setups against each other on the same journal data.

SQN

Average R per trade over its spread, scaled by √trades. Built from trade results.
Unit: trades, in R

Sharpe ratio

Average excess return over its standard deviation, usually per month or year. Built from account returns.
Unit: time periods, in %
Common mistake
Big winners raise the standard deviation as well as the average, so trend-following setups with rare large wins can score lower than steady small-win setups with similar expectancy. Read SQN next to the R-multiple distribution, not instead of it.
For F&O traders in India
Compute R after charges. On a Nifty option trade risking ₹1,950 (a ₹30 stop on one lot of 65), ₹80 of brokerage, STT and exchange charges is about 0.04R per trade, which matters when the average is only 0.1R to 0.3R. Split Nifty, Bank Nifty and stock F&O into separate tags before comparing scores.

Reference guide

ContextValueReading
SQN from 30 trades or fewerToo earlyBoth the average and the spread of R are still unsettled; one large winner can change the score a lot
SQN rising only because trade count grewRead with careThe √N term grows with every trade; check that the average R itself has held up
SQN compared across different R definitionsNot comparableOnly compare systems where 1R means the same thing: the planned loss at the initial stop

Questions

How is SQN calculated?

SQN = √N × average R-multiple ÷ standard deviation of R-multiples, where N is the number of trades. Fifty trades averaging 0.34R with a standard deviation of 1.56R give about 1.54.

What is a good SQN score?

In Van Tharp’s scale, 2.0 to 2.4 is average, 2.5 to 2.9 good and 3.0 or more excellent, with N capped at 100. These are rules of thumb; compare your setups with each other on the same data.

How many trades do I need for SQN?

At least 30 for a rough figure, and closer to 100 for one you can lean on. Below that, a single large winner or loser can move both the average and the spread.

What is the difference between SQN and expectancy?

Expectancy is the average result per trade. SQN divides it by how scattered the results are and scales by the trade count, so it also reflects how consistent and how well evidenced that average is.

Related terms