Glicko-2 Rating Calculator

A plain guide to the Glicko-2 rating calculator: what the inputs mean, how rating deviation and volatility move a rating after each rating period.

9 min read3 viewsGlicko-2 Rating Calculator
— · 0 votes
Contents
  1. What the Glicko-2 system measures
  2. The inputs a calculator asks for
  3. How the formula updates a rating
  4. Choosing the system constant and period
  5. Mistakes that distort calculator output
  6. Worked sequence for one rating period
  7. Reading and checking the results
  8. Using Glicko-2 in a group event
  9. Glicko-2 Calculator Questions
  10. What does a Glicko-2 rating system calculator actually do?
  11. What is rating deviation in Glicko-2?
  12. How is volatility used in the Glicko-2 formula?
  13. Why does the rating period matter for the calculation?
  14. Can I run the calculation by hand instead of using a calculator?
  15. What settings should a group pick for its calculator?

A Glicko-2 rating calculator turns a player's current rating, its uncertainty and a set of game results into an updated rating for the next rating period. The calculator applies the Glicko-2 formula, which tracks both a rating and how confident the system is in that rating. You feed it the inputs, and it returns the new numbers along with the uncertainty that carries forward.

  • Glicko-2 stores a rating, a rating deviation (uncertainty) and a volatility value for each player.
  • The rating deviation shrinks as results accumulate and grows while a player is inactive.
  • Volatility describes how much a player's true strength swings between rating periods.
  • Results are processed one rating period at a time, not game by game into a single figure.
  • The calculator output is only as meaningful as the inputs and the results you enter.

What the Glicko-2 system measures

Glicko-2 is an extension of the Glicko rating system, and it models a player with three numbers instead of one. The rating estimates current strength, the rating deviation estimates how uncertain that strength estimate is, and volatility estimates how erratic the player's results tend to be. Together these describe a player far more fully than a single score.

A rating deviation is a confidence measure, not an error to fix. A new player starts with a wide rating deviation because little is known about them. Each rated result narrows that uncertainty, so the system becomes more willing to move the rating. A long gap between events widens the rating deviation again.

Volatility captures how much a player's underlying strength changes from one rating period to the next. A steady player has low volatility, so results shift the rating less. A player whose form swings has higher volatility, and the system responds with larger adjustments. The Glicko-2 formula estimates volatility from the player's result history.

The inputs a calculator asks for

A Glicko-2 rating calculator needs the player's current rating, current rating deviation and current volatility before any new results. These three are the state carried over from the previous rating period. Entering them correctly matters more than any other step, because every later figure is built on them.

The calculator also needs the rating period settings: the system constant that limits how fast volatility can change, and the rating deviation of a new or unrated player. Some calculators ask for a conversion scale between the internal Glicko-2 scale and the displayed rating. Keep one set of settings for a whole event so results stay comparable.

Each opponent result adds an opponent rating, an opponent rating deviation and the outcome, usually a win, a loss or a draw. The opponent's rating deviation tells the calculator how much weight to give that result. A result against a well-established opponent counts differently than one against a newcomer with a wide rating deviation.

How the formula updates a rating

The Glicko-2 formula works through the results of one rating period and produces a new rating, a new rating deviation and a new volatility. It first estimates how the player was expected to score against each opponent. Those expectations are compared with actual outcomes, and the gap drives the update.

The calculation combines two sources of change. One is the information from the games played, which tightens the rating deviation. The other is the volatility estimate, which decides how much the system allows the rating to move. The formula solves for volatility iteratively, so the result comes from repeated refinement rather than a single expression.

After the update, the new rating deviation is adjusted to account for the passage of time until the next rating period. That step is what lets a rating become less certain during inactivity. A step-by-step calculation is easiest to check by keeping each intermediate value in a table and comparing it against the final output.

Choosing the system constant and period

The system constant, often written as tau, controls how much volatility is allowed to change in a single rating period. A smaller value keeps volatility steadier and makes the rating more conservative. A larger value lets volatility react faster, which suits players whose results genuinely shift in quality.

The rating period is the block of games the calculator processes in one update. You can set a period as a single event, a week or a season, but the choice must stay fixed while the data is being compared. Mixing period lengths across one ladder makes ratings hard to interpret fairly.

Tuning these settings is a judgment call, and there is no setting that fits every group. A recreational club may prefer stable ratings that move slowly, while a fast-moving league may prefer quicker response. Record the settings you choose so future updates stay consistent with earlier ones.

Mistakes that distort calculator output

The largest error in a Glicko-2 calculation is entering a stale or guessed rating deviation. A rating deviation that does not match the player's actual history makes the update move too far or too little. Recompute it from the previous period rather than copying a number from memory.

A second mistake is processing results in the wrong order or splitting one period into several updates. The formula is designed for a batch of results within a rating period, and each extra update adds noise. Collect the period's results first, then run the calculation once.

A third mistake is ignoring the opponent's rating deviation. Entering the opponent's rating without its uncertainty treats every opponent as fully known. The calculator's weighting depends on both numbers, so supply both whenever the tool allows it.

Worked sequence for one rating period

A clear worked example starts with a player's rating, rating deviation and volatility, then lists every opponent with the same details. Writing the inputs in a table before touching the calculator prevents transcription errors. The sequence below follows the order the formula uses.

The expected score for each game comes from the rating gap and the combined uncertainty of both players. A large rating deviation on either side pulls the expectation toward an even contest. This is why two identical ratings can produce different expectations when their uncertainties differ.

The final step returns three new values, and all three must be stored for the player. Keeping only the rating loses the uncertainty that makes Glicko-2 useful. A spreadsheet column for each of the three values keeps the history readable across many periods.

  • Record the player's rating, rating deviation and volatility.
  • List each opponent's rating, rating deviation and the result.
  • Set the system constant and the new-player rating deviation.
  • Compute each expected score from the rating difference.
  • Estimate the volatility from the results.
  • Produce the new rating, rating deviation and volatility.
  • Carry those three values into the next rating period.

Reading and checking the results

A Glicko-2 result is a set of three values, and each one tells you something different about the player. The rating shows where the player sits now. The rating deviation shows how much to trust that position, and the volatility shows how likely the rating is to swing next time.

Sanity checks catch most data errors before they spread. A rating that jumps far beyond the change the results justify usually points to a wrong rating deviation or a misentered opponent. A rating deviation that grows after a win suggests the period was set up incorrectly.

Compare the output with a hand calculation for a small case to confirm the calculator follows the Glicko-2 formula. If the tool shows intermediate values, check the expected scores first, since later steps depend on them. A calculator that hides its steps is harder to audit for a tournament.

Using Glicko-2 in a group event

Glicko-2 suits a recurring group where players meet again across several events. The system rewards consistent results and accounts for uncertainty, which helps when players join or skip rounds. A single knockout bracket is a poor fit, because the formula expects a batch of results per period.

Pairings still need care when ratings carry uncertainty. Matching a new player with a wide rating deviation against a settled player gives the system useful information. Spreading high-uncertainty players across the field improves the quality of every update.

Publish the settings and the update schedule so players understand how their rating moves. A short note explaining rating deviation helps people read an unexpected change without alarm. Keep the tournament record and the rating inputs together for any later review.

Glicko-2 Calculator Questions

What does a Glicko-2 rating system calculator actually do?

A Glicko-2 rating system calculator applies the Glicko-2 formula to a player's current rating, rating deviation, volatility and the results of one rating period. It returns a new rating, a new rating deviation and a new volatility. The calculator is a tool for arithmetic, so its output depends entirely on correct inputs.

What is rating deviation in Glicko-2?

Rating deviation is the uncertainty attached to a player's rating in Glicko-2. A wide rating deviation means the rating could be far from the player's true strength, and a narrow one means it is well established. The value shrinks with results and widens during inactivity.

How is volatility used in the Glicko-2 formula?

Volatility in the Glicko-2 formula estimates how much a player's true strength changes between rating periods. The calculator solves for it iteratively from the player's result history. Higher volatility allows larger rating adjustments in later periods.

Why does the rating period matter for the calculation?

The rating period defines the batch of games the calculator processes in one update. Glicko-2 expects all results from that period to be handled together, not one at a time. Changing the period length between updates makes the resulting ratings inconsistent.

Can I run the calculation by hand instead of using a calculator?

Yes, a step-by-step calculation is possible, but the volatility step requires repeated refinement by hand. A spreadsheet or calculator removes arithmetic errors and stores the three values between periods. Doing it by hand is useful mainly for checking a small example.

What settings should a group pick for its calculator?

The settings depend on how quickly the group wants ratings to respond and how often it meets. The system constant and the new-player rating deviation are the main choices. Record them once and reuse them so every update stays comparable.

Read next

Latest articles