WISDOM Win Shares (points) Calculation

←  Home

For anyone who's interested in how I figured out how many points to assign a player each season

Offense = Defense

Offense = Hitting + Base Runnning (where hitting counts for 90% and base running counts for 10%)

Defense = Pitching + Fielding (where pitching counts for 80% and fielding counts for 20%)

Points (Win Shares) Allocated to Teams:

points = team's winning percentage x 200

e.g. If a team wins all of its games, it would be awarded 200 points (1x200). If it wins half of its games , 100 points (.5x200). If it wins no games, zero points (0x200).

Points (win shares) are based on a 200 game season

Using 200 points as a basis for every season levels the playing field for all seasons regardless if they are 154 games, 162 games or some lesser number due to wars or strikes.

Basing on 200 points also adds a bit of granularity to allocating win shares while maintaining the sense of a relationship between actual wins and win shares.

The points allocated to a team are then further allocated into offensive and defensive points based on the difference from the average runs scored by all teams.

Approximately 8 runs translates into a win share

An example would be instructive:

Say that the league average runs scored per team is 700 and the team in question has scored 740 runs, given up 660 runs and had a record of 92 wins and 70 losses

The 92-70 record gives them a winning percentage of .568 which when multiplied by 200 gives a total of 114 points

They are 40 points above the league average in runs scored and 40 below the league average in runs yielded.

Because they are equal amounts of runs away from the league average on both offense and defense, we split the points allocated to offense and defense equally - 57 points allocated to each

A second example:

Suppose in the above example that the team had yielded only 620 runs, not 660.

They would be 40 above the league average offensively, but 80 below the league average defensively.

Therefore, we would skew the points allocated more to the defense than to the offense.

Note: The rule of thumb is 9 or 10 runs per win, because our wins are worth less due to using 200 game seasons for calulations, wins are devalued a bit to 8 runs per win

At 8 runs per win share with the defense 40 runs better from the mean than the offense, we should allocate 5 more win shares to the defense than to the offense (40 divided by 8).

This would yield us either 60 to the defense and 54 to the offense (6 win shares difference) or 59 to the defense and 55 to the offense (4 win shares difference)

There is no way with 114 total points (win shares) for the allocations to be 5 apart (we only work with whole numbers here).

Because there's a lot of rounding off, we'll determine whether we go with a 60-54 split or 59-55 split based on how the rounding off is looking for all teams overall.

With thirty teams in the big leagues and offense equalling defense, the total number of win shares allocated to offense and to defense overall is 1500 each - we'll need to ensure that our

roundoffs for all the teams adhere to a total of 1500 for offense and 1500 for defense for all teams combined.

Using the second example, we allocate further (assuming we ended up going with a 54 offense, 60 defense split):

As will be seen below, the allocations for hitting and base running for the team are independent of the points allocated to the team on offense.

For the sake of argument, we'll say that thru OPS, the hitters on the team total 52 points and thru stolen base totals for the team's game changing base runners, a total of four points are allocated.

52 (hitting) + 4 (base running) = 56 which is two more than what was allocated to the team for offense.

Therefore, we'll need to subtract two from someone to get down to 54 total offense for the team. This two is a fudge factor called HA or Hadj (hitting adjustment).

If you're uncomfortable with the term fudge factor and want to be more scientific, we can call HA "Lineup Synergy"

On defense, we allocate 80% of the 60 points (48 points) to the pitchers and 20% (12 points) to the fielders in every case

On defense, it's very difficult to know, for ground outs, fly outs, singles, doubles, triples and inside the park home runs, who was responsible, the pitcher or the fielder. How hard was the ball hit?

Was the defense positioned correctly? If not, why not? Did the fielder exhibit good range or bad range? Was the fielder fast or slow? Etc. Based on all of the unknowns involved, we'll stick with a

constant 80-20 splt and leave it at that

Therfore, in this example, we end up with the following allocations: Hitting:52; HA:-2; BaseRunning:4; Pitching:48; Fielding:12  Offense:54; Defense:60; total:114

Hitters:

All major league hitters with more than 300 at bats are graded on a bell curve from 1 to 10 points based on their OPS (highest OPS getting 10 points, lowest OPS getting 1 point)

I was considering using 400 at bats as the cutoff for the bell curve, but opted for 300 to get a denser number of data points.

Hitters with between 200 and 299 at bats with relatively high OPS are awarded points but less than those with over 300 at bats -part time players lose points for not being able to hit all types of pitching

Hitters with between 100 and 199 bats with high OPS are awarded points scaled down even further than those with 200-299 at bats

Hitters with under 100 at bats get no points unless there is a rare case of exceptionally high OPS

Base Runners:

We'll preface this with I believe that existing base running models fall short of the impact of base running on the game because they're into counting bases a player advances but don't really

account for what game changing speed does to pitchers facing subsequent hitters and to fielders once a fast runner is on base. It's difficult to quantify effects several hitters down the road.

The best formulas are linear regressions that create formulas that, while accurately predicting runs scored, are too complex to be intuitive. We're looking more for stuff you can wrap your head

around here rather than decimal accuracy i.e. we're trading off a bit of accuracy for a lot of intuitiveness in this app.

Points are awarded for base running only for players with game changing speed. Up to the top 10% of base stealers may be considered to have game changing speed. 10% of all total offensive points

are awarded for base running. The top base stealers are awarded points graded on a bell curve based on the number of stolen bases they have. The lower limit is one. There is no set upper limit to

points awarded for the top base stealers. The benefits of game changing base running don't accrue so much for the runner himself, but for those who follow him in the lineup and for the team

in general. Speed causes fielders to make more errors as they must rush their fielding and throwing. Pitchers' focus on the subsequent hitters is divided and mechanics are rushed when a fast runner

is on base. Defenses play out of their normal positions when fast runners are on base, thus helping the stats of subsequent hitters in the lineup. The effects are difficult to quantify. Runs scored

gives some indication, but there are so many factors involved in runs scored that it is problematic and complicated to try to isolate how many extra are due to the speed of the runner.

Pitchers:

Pitchers are paid to win. Historically, your best pitchers are your starters. Relievers aren't deemed as good (save for closers). Management values a pitcher who gets them wins.

Points are awarded to starting pitchers based on wins. Losses aren't counted in this model. A starting pitcher is paid to win, if he doesn't, no matter whether it's a loss or no decision, he didn't win.

A normal argument against using wins as a yardstick is that wins are based on how well the team's hitting supports the pitcher. However, in this model, the points a team gets for the number of

games they win are allocated to offense and defense based on how many runs the team scored and gave up during the season. Therefore, if the team scored a lot of runs or gave up a lot of runs, the

lions share of the points allocated to the defense and, therefore, to the pitchers will be reduced. Conversely, if the team doesn't score much or doesn't give up much, more points will be allocated to

the defense and, therefore, to the pitchers.

Once the pool of points is determined for the pitchers, it is divided up amongst the team's pitchers based on the following criteria:

Starting pitchers (those with more than 150 innings pitched) get the lions share of points based on the number of wins they have. Secondary factors are innings pitched and ERA.

Innings pitched are a huge factor in the allocation of points (win shares). Management values innings eaters highly. Conversely, if a manager trusts a pitcher (particularly a starting pitcher),

he will keep that pitcher out there for more innings. Also, the more effective a pitcher is, the more likely he'll rack up more innings pitched.

Starting pitchers are awarded for their ability to eat innings. They have the skill set, diverse repertoire and stamina to be able to pitch through a batting order effectively multiple times during a game.

Middle relievers, on the other hand,  generally are less skilled and have a less diverse repertoire than starters.

Pitchers with between 100 and 149 innings pitched are awarded relatively less points than starters. They get a scaled down number of points based on their win totals than do starters, but they have

pitched enough innings to be considered a cut above other relief pitchers.

Middle relief pitchers with between 50 and 99 innings pitched may get points awarded if they have high win totals, low ERA's or saves.

Relief pitchers with less than 50 innings pitched don't get points unless they have exceptional win totals, low ERA's or saves.

Closers are a special brand of relievers. They generally don't have a diverse repertoire of pitches, but the ones they do have are exceptional. Closers are the guys who get saves and closers are

awarded points on a par with starting pitchers. Closers also pitch in the highest pressure situations and the successful ones have a special psychological makeup in order to succeed.

Fielders:

There are formulas that try to isolate pitchers from fielders by rating pitchers on how they do with outcomes that don't involve fielders. It's a nice try, but it's hard to quantify when a pitcher misses

his spot and, therefore, his fielders are out of position or how consistent a pitcher is (e.g. will he cause a bloop one pitch and a screaming liner the next?) or how hard a ball is hit (i.e. who's

responsibility). It's a nice try to try to isolate, but I don't buy it intuitively.

I don't see a good, simple definitive way of allocating runs to fielders. Therefore, we'll just use the existing stamdard run formula that's out there and apply it to this model.

Rule of thumb is 10 runs make a win share, either plus or minus. Because our wins are worth 80% of real wins due to using 200 games per season, our rule of thumb is 8 runs make a win share.

We use a top down approach to fielding points - 20% of the total defense points allocated to the team go to the fielders no matter what the fielders' individual run totals look like.

We allocate fielding points to players based on three criteria: do they play a skill  position?; the number of innings played at the skill positions; runs saved (plus) or given up (minus)

The three skill positions that happen to be the spine of the defense are catcher (4 points); shortstop (3 points); centerfield (3 points).

A good catcher is invaluable for his leadership (many become managers when they retire); pitch calling; pitch framing; pitcher handling; fielding (bunts, stolen bases, pickoffs, pop ups, foul tips, etc.)

Even if he doesn't hit a lick or is slow as molasses, if a catcher possesses the above skills, he's worth his weight in gold, or worth at least 4 points.

Your best infielder plays shortstop. Unlike first, second or third bases , you can't bobble, no time for second chances, you must field the ball cleanly the first time,. You also must have a cannon for an

farm and throw accurately to boot. You are also, as a shortstop, the defensive leader of the middle of the field

Your best outfielder plays center. You must be sure handed, have good range, be able to cover a ton of ground, know the hitters, be able to come in on short flies, be able to go back on long ones, and

coordinate amongst the outfielders.

If more than one player played at a skill position for a team during the year, the points for that position are divided based on innings played at that position.

The eight runs per win share formula is applied to fielder runs. There is potential for a lot of fudging here as we need to shoehorn the runs formulas into the points allocated for fielding.

Designated Hitters:

If a player is a full time DH, he gets -3 for fielding.

Other fielders will get his points to make up the difference based on the one win share per runs saved formula.

If a DH were one of your good fielders, he wouldn't be DH, hence we assume bad fielder. We make up the minus points for the DH with other fielders because most managers, with a DH, feel free to

beef up their defense when they've got a DH because the DH is taking care of offense - balances things out. Also, prior to interleague play,when the AL only was a DH league, we scored hitting from

1 to 11 rather than 1 to 10. The reason for this was that the team's run scoring was now split amongst nine guys rather than eight throughout the league. Therefore, due to the marking of hitters on a

curve, each hitter's accomplishments were being downgraded 11% with all of these DH's (rather than pitchers) skewing the curve downward for everybody else.

HAdj / Hitting Adjustment:

HA = "Offensive points allocated to the team" - "Hitting" - "Base Running"

In example two above HA = 54 - 52 - 4 == -2

If HA is a plus number, we allocate it to the team's best hitters based on OPS and at bats

If HA is a minus number, we allocate it to the team's worst hitters based on OPS or on at bats or both. In this case, we'll subtract the closer a hitters total number of at bats is to the 300 AB cutoff line

We give credit for good hitters adding positive synergy to a lineup and for bad hitters adding negative synergy to a lineup

Good hitters make the hitters ahead of them better because those guys get to see better pitches as pitchers don't want to give away free passes to hitters with a stud coming up behind them

Similarly, bad hitters make the hitters ahead of them worse because those guys get to see nothing as the pitchers will pitch around better hitters to get to the bad ones.

A position player's total points = Hitting + HA + Baserunning + Fielding. Note that even tho' fielding and HA can go negative, we never allow a player's total points to go less than zero. The reason is

is that we are counting win shares and negative wins makes no sense.