Definition

Review results relative to your recorded initial risk.

Formula. Average of net trade P&L ÷ recorded risk amount. Only closed trades with a positive recorded risk amount are included.

Same money, different quality

Row C is why R matters: a loss larger than 1R is evidence about your execution, not about the market.
TradeRisk at entryNet resultR
A$200+$600+3.0R
B$2,000+$600+0.3R
C$200−$600−3.0R — a stop was not honoured

How it goes wrong

Not recording initial risk
Only closed trades with a positive recorded risk amount are included. Orderblock will not infer risk from a stop you might have moved, because an inferred R is worse than none.
Recording risk after the fact
Risk written down after the outcome is known is not risk, it is rationalisation. Log it at entry or leave it blank.
Ignoring losses beyond 1R
The distribution of R on losing trades is the cleanest available measure of discipline. Any cluster beyond −1R deserves a journal entry.

Questions

What is a good average R?

Positive, with losses tightly clustered near −1R. The average matters less than the shape: a +0.2R average with no losses beyond −1.1R describes a controlled process, while +0.5R with occasional −4R losses describes an uncontrolled one that has been lucky.

Why are some of my trades excluded?

Any closed trade without a positive recorded risk amount is left out of the average, because there is nothing to divide by.