Value Decomposition
Baseball Player Won-Loss Records
Home     List of Articles

Value Decomposition

In the top right of the first two tables on basic player pages are a link to the Value Decomposition page. The Value Decomposition table decomposes a player's value into its constituent parts, by season and for the player's career as a whole.

Decommposition of Player Value
The first two tables on the Value Decomposition page decompose Player wins over replacement level into its constituent parts.
Clicking the "Value Decomposition" link on the pWins table decomposes pWORL; clicking the "Value Decomposition" link on the eWins table decomposes eWORL. One can switch from one to the other by adding or subtracting "&p=1" to the end of the filename: including "&p=1" will open the pWORL version of the Value Decomposition page, excluding that term will open the eWORL version. The first table here shows the player's wins relative to average in the four aspects at which a player can earn Player wins: Batting, Baserunning, Pitching, and Fielding. The headings of these four columns link to the player's Batting, Baserunning, Pitching, and Fielding pages, respectively. These tables are described here. For batting and baserunning, the "average" against which these are measured is non-pitcher average. For pitching and fielding, average is simply a .500 winning percentage.

The final column of this table sums the totals for these four aspects. For the version of this table based on eWins, the results here are context-neutral and teammate-adjusted. For the version of this table based on pWins, these numbers are context-dependent.

The second table, then, converts raw wins over average from the first table into Wins over replacement level (WORL).
The first column of this second table, headed "Positional Adjustments", adjusts for the player's positional average. Positional averages adjust for differences in average player performance across different positions. A negative number indicates a positional average over .500 (so that a player's wins over positional average are less than his wins over .500); a positive number indicates a positional average below .500. For the eWins version of this table, the numbers in this column also incorporate expected contextual adjustments. So-called "context-neutral" eWins and eLosses do not incorporate no contextual adjustments, but, rather, they incorporate expected contextual adjustments. Specifically, they incorporate expected context and expected win adjustments.

Expected Context is calculated based on a player's position and incorporates both inter- and intra-game context. Expected context differs most strongly from 1.0 for pitchers. Expected Win Adjustment measures a player's expected intra-game win adjustment based on the player's (context-neutral) winning percentage. Expected Win Adjustment helps adjust for the non-linear effect of player performance on team winning percentage.

For the pWins version of the table, the numbers in the first table incorporate the actual context in which the player's performance took place so there are no additional contextual adjustments which need to be taken account of.

The second column of the second table shows Wins over Positional Average (WOPA), which is equal to the sum of the last column of the first table and the first column of the second table.

The next-to-last column, then, presents "Replacement Value". Replacement Value is simply the difference between WORL and WOPA. In effect, this measures the raw value of playing time.

The final column, then, is equal to the sum of the two preceding columns, Wins over replacement level, WORL.

The final table builds on the page which I have constructed which allows one to build a so-called "uber-statistic". The first link in the previous sentence links to a page which allows one to specify weights for pWins vs. eWins, for wins versus WOPA versus WORL versus WO* (Wins over star - set at one standard deviation above positional average), by position played, and several other considerations.

Specifically, the final table on the Value Decomposition page applies my default set of "Uber Weights" - which were originally developed for and described in my second book, Baseball Player Won-Lost Records: 150 Players, 50 Years - to the player's record by season and for his career. For now, this table is tied to the default weights I developed in my book. But even if you prefer an alternate set of weights, I think this table can still be helpful for observing career arcs, identifying a player's best season(s), and for providing a tool for filling in gaps in a player's record.

One word of caution: this final table may take a little time to appear as all of my tables are calculated upon request and the calculation here is fairly complicated.

All articles are written so that they pull data directly from the most recent version of the Player won-lost database. Hence, any numbers cited within these articles should automatically incorporate the most recent update to Player won-lost records. In some cases, however, the accompanying text may have been written based on previous versions of Player won-lost records. I apologize if this results in non-sensical text in any cases.

Home     List of Articles