Mahjong Average Placement | Calculation, First-Place Rate, and Last-Place Rate

Calculate average placement from five game results

First, suppose your finishes in the last five games were “1st, 3rd, 2nd, 4th, 2nd.”

A lower average placement means you recorded more high finishes during the measured period. However, an average of 2.40 over only five games tells you neither your skill nor how often you took each place. With fewer games, a single 1st or 4th moves the value substantially.

The same average placement can also be calculated from placement rates

If you know each placement rate in four-player mahjong, you can also calculate it as “1 × first-place rate + 2 × second-place rate + 3 × third-place rate + 4 × fourth-place rate.” Convert percentages to decimals for the calculation. If the rates are 30%, 25%, 25%, and 20%, respectively, then 1 × 0.30 + 2 × 0.25 + 3 × 0.25 + 4 × 0.20 = 2.35.

This formula should match the value calculated from the recorded placements. If it does not, check whether the reporting periods differ, incomplete games were included, or displayed percentages were rounded.

What 2.50 in four-player and 2.00 in three-player mahjong mean

Assuming every player takes each place equally often, average placement is the mean of the placement numbers. This symmetrical distribution gives 2.50 in four-player mahjong and 2.00 in three-player mahjong.

You also cannot directly compare 2.50 in four-player mahjong with 2.00 in three-player mahjong. Because the possible number of placements differs, evaluate each against records with the same number of players and rules.

When using an external benchmark, check whether its service, room or player tier, game format, rules, measurement period, and number of games match your records. A figure without stated conditions cannot simply be adopted as a personal target.

Read four-player average placement relative to the center

Read a four-player average placement by its distance from the mathematical center of 2.50. The table is a guide to that distance, not a set of skill tiers.

The example value of 2.40 is 0.10 below the 2.50 center. That difference alone does not mean that a player with 2.40 is strong. A single finish moves a small sample substantially, and results from different opponents or rules cannot be compared on the same basis.

Average-placement distribution of all 40 M.LEAGUE players

The official M.LEAGUE 2025-26 regular-season page lists results for all 40 players. Its “average finish” field corresponds to the average placement used in this article.

This distribution groups the values displayed on the official page into five bands and confirms that 3 + 12 + 10 + 11 + 4 = 40 players. The article neither recalculates nor rerounds the official average-finish values.

Individual match counts range from 20 to 39. This is a player distribution that gives each of the 40 players equal weight, not a distribution weighted by matches played. It therefore cannot establish an overall player average or a standard of individual skill.

Compare official results for three M.LEAGUE players

The following table selects three players with different relationships between average placement and placement distribution from the official M.LEAGUE 2025-26 regular-season results. Top-two rate is the percentage of matches finished in first or second.

Shimoishi played 38 matches with an average placement of 2.07, a 36% first-place rate, a 71% top-two rate, and an 84% last-place avoidance rate.

The percentages are the values shown on the official page converted to percentages, and the three players also played different numbers of matches.

Shiratori and Takizawa differ by only 0.02 in average placement, but by 22 percentage points in first-place rate and 13 points in last-place avoidance. These three records alone cannot establish playing style, the causes of the differences, or long-term skill.

Average placement and M.LEAGUE points use different calculations

Average placement is the arithmetic mean of the placement numbers. The official M.LEAGUE rules, however, use placement points that are not evenly spaced.

M.LEAGUE points also include raw score based on each player’s final points. Two players with similar average placements can therefore have very different individual scores when their placement distributions and raw scores differ.

Questions answered by five supporting metrics

Average placement alone does not show which placements increased or decreased, or how large the point differences were. Match each of the following five metrics to the question it answers.

Deal-in rate (houjuu rate) is the percentage of hands in which an opponent won by ron on your discard.

Even if average placement worsens, that number alone cannot prove that you attacked too much or defended too much. First, separate the changes in placement using first-place rate, top-two rate, and last-place rate, then check point differences using average score. Deal-in rate only tells you the percentage of hands in which you dealt in. To investigate the cause, review your tiles, your opponent’s actions, the turn, and other details from those hands in your game records or notes.

Average score and actual net results are also different figures. Net results include the conversion rate, chips, table fees, and other factors. The Mahjong Results Calculation Guide explains those conversions.

Same average placement, different distributions

Compare two players with an average placement of 2.50 over 20 games. Both have a placement total of 50, but their placement distributions are very different.

A has both a first-place rate and last-place rate of 40%, while B has 10% for both. Although their average placements are identical, A has more 1st- and 4th-place finishes, while B is concentrated in 2nd and 3rd. Which is preferable depends on placement points and your goals, so keep the full placement distribution when evaluating results.

Two patterns that improve average placement

Average placement can improve through more high finishes or fewer low finishes. Over 20 games, converting two fourth-place finishes into thirds reduces the placement total by 2 and improves the average by 0.10. Converting two second-place finishes into firsts has the same numerical effect. In the first case, last-place finishes decreased; in the second, first-place finishes increased.

Place the month-over-month changes in first-place and fourth-place rates next to average placement to show which direction drove the change. If the number of games differs by month, compare percentages rather than raw counts and always include the denominator.

Conditions to align before comparing results

Before comparing two sets of results, align the number of players, game length, rules used, measurement period, and number of games. Combining records from different conditions makes it impossible to separate changes in play from differences in the environment.

Use the same number of players

Because three-player and four-player mahjong have different placement ranges, track them separately.

Use the same game length and format

A hanchan usually includes both the East and South rounds, while a tonpuusen usually includes only the East round. Names and end conditions can also vary by ruleset, so keep records consistent with the format actually played.

Use the same scoring and end conditions

Note any day when placement points, red fives, early-end conditions, or the player pool changed. Even with the same average placement, the value of points and the distribution of attainable placements may differ.

Use the same measurement period and number of games

Short-term figures include random fluctuations. Do not impose one universal threshold for a sufficient sample; state both the period and the number of games collected under the same format and rules.

One game has a large effect on a small sample

In four-player mahjong, changing one fourth-place finish to a first reduces the placement total by 3. Divide 3 by the number of games to find the change in average placement.

This table is an original calculation example, not an external empirical dataset. Even when average placement is shown to two decimal places, a small-sample difference cannot establish a long-term skill gap.

Research on mahjong skill estimation also identifies the potentially high variance of estimates based only on average placement. This article sets no universal “minimum number of games”; compare short-term and lifetime figures and check whether the trend persists as the sample grows.

Separate short-term and lifetime results

Display the period used to assess recent changes separately from lifetime results used to assess long-term trends. Even when using a fixed window such as 30 games, do not treat that number itself as a skill threshold, and include only records with the same format and rules.

How to review your monthly results

1. Align the number of players, game format, and rules
2. Check the target period and number of games
3. Review average placement and the placement distribution
4. Check whether the first-place rate, top-two rate, or last-place rate changed
5. Check average score and net results separately
6. Review game notes and rule changes during periods with large shifts

If you have not chosen a recording method, compare paper, spreadsheets, and dedicated apps in Choosing a Mahjong Record-Keeping Tool.

Alongside the figures, note specific situations from the period. For example: “came from behind twice as dealer in the south round” or “opponents won three times in hands where I pushed unsafe tiles against riichi.” These notes narrow down which situations to review next. MemoJong does not automatically evaluate push-fold decisions or dangerous tiles; revisit the recorded situations and assess your own decisions.

Read average placement together with the placement distribution

Average placement alone does not reveal the balance between firsts and lasts or the size of point differences. Use the placement distribution and supporting metrics collected under the same number of players, game format, rules, and period to identify which figures changed.