Methodology

How Performance Grades are calculated

bridgeranking.net calculates a Performance Grade for every player across participating clubs. This page explains the full algorithm — the maths, the design decisions, and how the system differs from England's National Grading Scheme on which it is based.

1. What a Performance Grade measures

A Performance Grade is a number — currently scaled roughly 39 to 70+ — that reflects how well you play bridge right now, relative to the other players in the system. It is not a career average, a masterpoint count, or a measure of how long you have been playing. It is an ability rating, recalculated after every session.

The grade is built from the difference between how you actually scored and how you were expected to score given the current grades of your opponents. Beat stronger players and your grade rises; lose to weaker ones and it falls. The size of the move depends on how many boards were played and how well-established your grade is.

Grades are only published once a player has accumulated at least 150 boards. Before that threshold, the grade is calculated internally but treated as unreliable.

2. Session performance score

After each session, every player receives a session performance score — a percentage that captures how much better or worse they played than expected.

Expected score

For each board, a player's expected score is derived from the current grades of all their opponents across the field. The higher the opponents' grades, the lower the expectation. The formula computes a weighted average expected matchpoint percentage for the whole session.

Performance score

The session performance score is simply:

P = actual_percentage − expected_percentage + 50
Where 50 is added so that exactly meeting expectations yields a score of 50. A score above 50 means you outperformed; below 50 means you underperformed.

This score, P, is what gets fed into the rolling grade calculation.

3. The rolling grade formula

Performance grades don't simply average all your session scores. Each session's contribution decays over time — recent boards count more than old ones. This is the core of the NGS methodology.

Board weight

Each board played in session s is assigned a weight based on how many total boards have been played since it, across all sessions:

w(b) = max(0, 1 − (boards_since_b / 2000))
A board played today has weight 1.0. A board played 1,000 boards ago has weight 0.5. A board played 2,000 or more boards ago has weight 0 — it no longer contributes anything to your grade. At roughly 8–10 boards per session and 2–3 sessions per week, 2,000 boards represents about nine to twelve months of regular play.

Session weight

The weight of an entire session is the sum of the weights of each board played in it:

W(s) = Σ w(b) for each board b in session s

Weighted grade

The raw Performance Grade is the weighted average of all session performance scores, going back as far as those boards still carry weight:

G_raw = Σ ( P(s) × W(s) ) / Σ W(s)
P(s) is the session performance score (the adjusted percentage from §2). Sessions so old that all their boards have zero weight drop out of the sum automatically — there is no hard cutoff date, just a smooth fade to zero.

4. Shrinkage

A player with very few boards could have a wildly inflated or deflated raw grade just from a lucky or unlucky run. Shrinkage corrects for this by pulling the grade toward 50 in proportion to how little data there is.

total_weight = Σ W(s) (sum across all sessions still in window) shrinkage_factor = min(1.0, total_weight / 2000) G = (G_raw × shrinkage_factor) + (50 × (1 − shrinkage_factor))
A player with 2,000 boards of weight has a shrinkage factor of 1.0 — their grade is taken as-is. A player with 200 boards of weight has a factor of 0.1 — their grade is 90% pulled toward 50. The grade is only published at ≥ 150 boards (shrinkage factor ≥ 0.075), to avoid showing meaningless numbers to new players.

Shrinkage also protects the system: a very new player cannot single-handedly move anyone else's expected score by a meaningful amount, because their grade sits close to 50 regardless of a few strong sessions.

5. Grade bands

Each Performance Grade maps to one of 16 named bands, based on the card ranks used at the bridge table. The bands are the same as those used in the EBU's National Grading Scheme, with one characteristic: only the top band — Ace — is split by suit, giving four distinct Ace bands for elite players.

Ace ♠67 +
Ace ♥65 – 67
Ace ♦63 – 65
Ace ♣61 – 63
King59 – 61
Queen57 – 59
Jack55 – 57
Ten53 – 55
Nine51 – 53
Eight49 – 51
Seven47 – 49
Six45 – 47
Five43 – 45
Four41 – 43
Three39 – 41
Twobelow 39

The band labels are intentional: every bridge player already knows these names and what they represent at the table. There's nothing to learn.

6. How this differs from England's NGS

The EBU National Grading Scheme (NGS) is the system this is based on. It has been running in England since the early 2000s and covers hundreds of thousands of players across thousands of affiliated clubs. The methodology here follows NGS v17 closely — the same decay window, the same shrinkage approach, the same band structure.

The key differences are scope and context, not algorithm:

Aspect EBU National Grading Scheme bridgeranking.net
Coverage All EBU-affiliated clubs in England — hundreds of thousands of players Currently under review at two clubs in New South Wales. Designed to expand to any Australian club using Compscore.
Governing body English Bridge Union (EBU) Independent. Not affiliated with the ABF or any state body.
Data source EBU affiliated club results via their own systems USEBIO XML export from Compscore (the most common Australian club scoring system)
Cross-club comparisons Yes — your NGS grade reflects performance across all clubs you play at Not yet. Each club's players are currently graded within that club's field. Multi-club grading is the next step once more clubs are on board.
Core algorithm Weighted decay over 2,000 boards, shrinkage to 50, 16-band scale Identical
Band structure Two through King (plain), Ace ♣ ♦ ♥ ♠ (four suit bands at top) Identical
Publication threshold Varies by EBU policy 150 boards minimum
In short: if you have an NGS grade from playing in England, you would expect your bridgeranking.net grade to be in a similar range — the algorithm is the same. The main difference today is that your grade here reflects only your performance within your local club's field, not against a national population. That changes as more clubs join.

7. What the grade is not

Not a percentage average. Your club percentage average rewards consistency over decades and punishes you for playing in strong fields. The Performance Grade rewards you for playing well against stronger opponents and treats old results as less relevant than recent ones.

Not a masterpoints count. Masterpoints measure participation and event calibre, not ability. A Performance Grade can go down if you play poorly.

Not permanent. If you stop playing, your grade fades toward 50 as your results age out of the 2,000-board window and shrinkage takes over. Come back after a long break and you'll need to re-establish it.

Not a ranking within a session. The grade is not your position in last Tuesday's results. It is a rolling measure of your general ability, updated after each session but built from all recent history.

8. Questions and feedback

The system is in active development. If you have questions about the methodology, find an error, or want to discuss whether it's right for your club, get in touch.

admin@bridgeranking.net