Chris on APBA: On Pitching part II

Chris compiled some good info on pitching last week. Here is a continuation of the discussion…

On Grading Approaches

There are five viable generations (or four and one major sub-category) of grades for the National Pastime-APBA game, of which the Skeetersoft boards are certainly a derivative.

1.         The APBA grades, which take a lot of heat but are quite good at doing what they were designed to do, replicating ERA.  They do give way to Wins and were docked for innings, but the design was to represent a player’s value to the team, not specific stats down to the nth degree.

2.         CMBA is a “cheater” system.  That’s not derogative – just couldn’t ever come up with a more descriptive name.  CMBA wants to allow more hits and make the pitchers more realistic and then capture the ERA by regulating how many of those runners score.  That’s not really so different from APBA, it just starts with Runner on First instead of Bases Empty.  It satisfied a psychological need all right, but nobody ever demonstrated its overall accuracy as being any better than the next generation of APBA grades.

This system is not recommended.  The hit adjustments on the Skeetersoft hitting cards are not in synch with the CMBA grades.  Of course, neither are the APBA hitting cards after 1994, and especially after 2006.

3.         The APBA system was officially revised (they didn’t announce it, but had been building up to it since the late ’80s) to include several ground-breaking ways to look at the application of the grading structure.  The 30 grades of the MG were elevated to the primary status and the Basic Game grades became derivatives.  Around 2001 or so the BG grades were further adjusted to account for the Platoon Factors in the batter grades.  Nobody’s ever demonstrated to me that the CMBA grades performed any better than these grades.

4.         A system for utilizing H/IP to grade the batters was introduced for BBW by the Skeetersoft consortium around 1998.  It was designed for BBW but was based on the hitting charts of the (1976) MG, around which the computer games were built.  About the same time, Miller Associates introduced a similar system into the Wizard for Windows programs.  The Skeetersoft model was a bit more sophisticated, as it included the schedule of possible opponents in the weighting system.

The Skeetersoft scheduling algorithm was later included in Joe Sweeney’s grader which combines the two CMBA grades into one — which wasn’t really an improvement to me since we could already accomplish everything CMBA tried to do by aggressively assigning the ratings in BBW.

On Walks:  14’s are generally assigned to hitters’ cards on a 1:1 ratio, meaning the walks a hitter gets are represented by the 14’s on his card.  Many APBA seasons include IBB in the totals; SKS seasons do not.  A pitcher with a plain control rating can be assumed to give up walks at the league average rate.  Z pitchers will steal about a third of the 14’s they see while W pitchers surrender walks about 1.3 times as often as a plain pitcher.

If the league average batter walks 60 times in 600 plate appearances, he should walk about 60 times in 600 plate appearances against a plain control pitcher, about 40 times against a Z, and about 80 times against a W.  Assume these basic figures whether dice or BBW.

An APBA C is generally going to yield about a hit per inning.  B’s are going to be around .8 hits per inning.  A’s are around .55 and A&C’s between .5 and .55.  D’s are usually around 1.25 to 1.30.

Pitchers’ Strikeout Ratings

Strikeout Letter        K/(BF-IBB) compared to league average

R                                 less than 0.62

No Letter                   0.62-0.92

Y                                 0.92-1.17

X                                 1.17-1.40

XY                               1.40-1.66

K                                 1.66-1.90

KY                               1.90-2.00

KX                               2.00-2.20

KXY                            greater than 2.20

Pitchers’ Control Ratings

Control Letter           BB/(BF-IBB) compared to league average

ZZ                               less than 0.30

Z                                 0.30 – 0.83

No Letter                   0.83 – 1.26

W                                greater than 1.26

Pitching Miscellany

Q.        What are the R and K ratings?

A.        The R is a pitcher rating that removes strikeouts for low strikeout pitchers.  The K is a pitcher rating for high strikeout pitchers.  The R and K were added in 2006 along with the official scorer’s option.  Any set calculated in 2006 and after has the R and K pitcher ratings.  The official scorer’s option adds or removes errors for high or low fielding percentage seasons.  They are not master symbols.  They are similar to the X, Y, W and Z ratings. 

If you have cards that have the R and K rating but don’t have the boards that use those ratings, then ignore the R and make K pitchers XY pitchers (the pitcher gets all strikeouts for the X and the Y).  The R changes board results to an out, while The K changes board results to a strikeout

Q.        Are the R and K ratings are found on the pitchers’ cards? I don’t remember seeing them, but maybe I just overlooked them. 

A.        Yes they are on the pitcher’s card, but there are probably very few in the set.  It depends on the season.  The ’65 set I have contained 4 pitchers with K ratings.  Sandy Koufax appears as A (K) (Z).  There are 8 pitchers the R rating.

The R and K ratings are not in BBW 5.75.  The computer uses a different system which is probably more accurate.  It would destroy playability if used with the cards.  The card system is more emblematic; when you see a K on a pitcher’s card, or better yet, a KY, a KX, or the top of the line, a KXY, you know you are looking at a monster.

Q.        How does Skeetersoft handle the G/H/L/M Home Runs Allowed ratings for pitchers?  Is it the same in the computer version?

A.        The HRA letters are only utilized in Skeetersoft NP3 version but every homer or double that results from the cards and boards is evaluated if the pitcher has an applicable letter.  In other words, if a double is hit against an L or M pitcher, dice are rolled to see if it’s converted to a homer.

Additionally, the Grades are assigned as a set percentage of plate appearances and are balanced for their league.  The effect is slightly different than APBA’s as SkS ratings regulate individual pitcher stats whereas APBA’s are designed to affect the individual pitchers ERA which is a slight, but powerful, distinction.

Q.        Does BBW incorporate R and K, maybe ZZ?  I know the computer game does not recognize dual batting characteristics.  I’d like to know any other limitations as well.

A. The ZZ doesn’t exist in BBW, but a player averaging under 1.5 walks/9 will perform similar to a ZZ.  Under 0.75 BB/9 and he will perform another level better.

A.     BBW, in its initial form, does not officially recognize those ratings as they didn’t exist at the time that Roswell programmed the game in 1984.

When the game migrated from the APBA Computer Baseball to BBW, it was a smooth transition that did not require much additional programming.

BBW actually had a ZZ and a ZZZ-type of rating, before the board version did.  This was done through an evaluation, and a manipulation of statistics from the player cards.  It wasn’t called Z, or ZZ ratings, but they existed.  They weren’t based on the same formula that the Z and ZZ ratings are based on in the board version, but they are designed to do the same thing.

They did do something to ‘help’ pitchers who had a few more strikeouts than the XY was delivering, but I don’t think that it was much.  It certainly wasn’t much, as no one could have envisioned the amount of strikeout numbers that exist now.  Never saw a study of the amount of strikeouts that a XY versus a K would yield.  A XY is pretty strong, and I don’t know that that many more strikeouts are going to come from the pitcher’s card as opposed to the batter’s card.

BBW has nothing at all to do with a R-rating, or anything close to it.  This one did not exist, and wasn’t on the radar as far as I know when the shut down the business.

In game terms, there are plenty of manipulations going on, but many of the ratings simply don’t exist.  One example is the Steal Allowance Letter.  The game will allow an R-rated player to steal as much as an A.  Why?  Because the game doesn’t recognize those letters, but the Micro manager program does.  So, you can use the micromanager program to define any type of formula that you want to make it work however you want.  That is what governs the usage, not the game.

Now, BBW is programmed to make the Hit-and-Run play, and the Bunt play, less effective after a few times of usage.  The exact number is not known.  Bunts were more prevalent back in the day, so the era that is being played sets those parameters based on that.  Generally, for the H&R play, you will see things be less successful after about 3 attempts or so.  In both cases, you will begin to get more foul ball results.

The game itself was designed so that it didn’t have specific restrictions because it needed to appeal to many leagues that use many different rules.  The micromanager program is where all of those restrictions are individually applied, in whatever manner that you desire.

Q.        BBW may have finally hit the wall.  The number of pitchers who strike out more than 10 per nine innings is quite a list.  Then there is opposite scenario.  One guy walks about 7 per nine.   Someone needs to program a new version of BBW.

A.        As long as it was just one or two pitchers at a time, pitching significant innings with some off-the-charts number, it was probably ok.  You could sort of work around Feller or Byrne or Score or Ryan but now every team has a couple Billy Wagner’s or Frankie Rodriguez’s or a Chris Sales.  At the time the guys were working on BBW, there were pitchers who had exceeded the board game parameters but after the game became locked in place, the numbers really exploded.

I don’t believe that BBW has hit the wall at all.  The strike out totals of replays, at least from 2014 and on, are still very normal.  Sure, they have been growing, but so are the number of 13’s on the batting card.  The batting results are supposed to come from the batting cards.  The pitching ratings are only supposed to regulate those results.  In theory, all they do is raise the number of strike outs for known strike out pitchers, and lower the strike out totals for the others.  The game still does that.  I don’t see what you think the problem is.  If you were to replay a season, I believe that the strike out totals, etc., would be just fine.

Q.        BBW assigns a random grade to starting pitchers.  There is a -3 to +3 random movement on these.  Can we assume that a Grade 15 would never get upgraded or does it affect all grades?  And is this just for starting pitchers not relievers; or both?

A.        The randomization of grades occurs on a batter-to-batter basis and it came about when BBW was created.  The original DOS version did not have it.  The randomization applies to ALL pitchers at all times.  One of the important things that it does is to allow (or at least it initially did) the L/M/G/H to be applied to any pitcher because of the grade movement.  If you had the boards, which I always did when I was testing for them, I would look to see if the pitcher’s home run modifier could come up in certain situations.  It would be similar to picking a Grade 5 reliever over a Grade 6 because the Grade 6 had a L on his card and that could become a modified home run.  So, it was better to have the Grade 5 pitcher instead.  The randomization sort of changed that.

Again, the reasoning behind this is that Kenneth didn’t want anyone playing APBA-ball.  He didn’t want someone holding the card and looking at it and saying, “This one has three nines, and two eights, so if I pick this pitcher I can eliminate one of those hit numbers.”  He wanted it to be more like real baseball, rather than a board game.  I might add that I was initially against it.  In fact, Kenneth did not want the card play result numbers to be displayed, so he never created that function.  There was a guy who did create that option for the original DOS version called POPBALL.  By hitting some particular key strokes the card would pop up on the screen.  I loved that program.  Kenneth later disabled that user add-on with the creation of BBW.  The new version of BBW, 5.75 has it.

The way that the randomization works is that it essentially adds the possibility that the expected grade could be changed for that batter.  I have forgotten the exact percentages, but I want to say that it was something like 40 percent that the original grade would not be changed; 15 percent that it would go up one grade; another 15 percent that it would go down one grade; 10 percent that it would go up two grades; 10 percent that it could go down two grades; 5 percent that it would go up three grades; 5 percent that it would go down three grades.  (I believe that these were the ranges, but it’s been a long time.)

Say the original breakdown was 5 * 10 * 15 * 40 * 15 * 10 * 5 – a Grade 12 pitcher, could be anywhere from a 15 to a 9, before any other modifications.  It kept you from playing APBA-ball, to a certain extent.

Further testing shows the actual breakdown to be 2 * 8 * 15 * 50 * 15 * 8 * 2.  It’s entirely possible that somewhere along the lines they decided to trim the ends a little bit.  In fact, if you look real close you will see that the -1 and +1 categories at both set a 15 percent.  So, it appears they took out a collective 5 percent from each end of the last two categories (-3/-2/+2/+3) and gave that 10 percent back to the middle.  That would make sense and would make the base grade occur 50 percent of the time.

Q.        Does BBW actually show you the adjusted grade?  Does a 12 suddenly show as a 10 or 11, or a 14-15 or do you just know behind the scenes that the +3 to -3 adjustment can occur?

A.        No, it does not.

Q.        Does the pitcher adjustment only happen on base hits or does it occur prior to the batter roll?

A.        Example 1:  A Grade 9 pitcher is on the mound.  A SA-1 batter comes to the plate.  The calculations begin.  The Grade 9 pitcher’s grade can be randomized.  It can stay the same, or possible go up or down, as much as three grades.  So, that pitcher’s new grade could be anywhere from a 12 to a 6, with the most likely grade being somewhere around the middle, at 9.  Then, the SA-1 (if it is the same side pitcher) is added.  If the grade had been elevated by one to a 10 by randomization, it will now be raised another point due to the SA-1 platoon factor.  The final grade as the result is provided is an 11, in this example.  Pitching grades, for those who don’t have the board game, can affect the hitting numbers 1-11.  Everything else has no bearing on the pitcher’s grade.

            Example 2:  A Grade 6 pitcher has just come into the game as a reliever.  The team pinch hits with a batter that is a PL-2, but is not hitting from the same side as the pitcher.  A runner is on at third, and the pitcher is pitching from the stretch.  Okay, here we go:  The Grade 6 reliever’s grade gets randomized.  As mentioned before, it can go up or down by as many as 3 points.  Let’s say that the randomization lowers it by 1 this time.  So, after the randomization, he is now a 5.  Next, the 5 points are added for the first batter grade adjustment factor, so he is now a 10.  Since there is no platoon factor in play due to the batter and pitcher being opposites, that modification is not added in, so the Grade remains as a 10.  Finally, since the pitcher is pitching from the stretch, thus disabling any possibility of a steal of home, his grade is reduced by two more grades.  So, his final grade for this first batter is an 8.

            Hypothetically, had there been no stretch, he would have been a 10; had there been a positive randomization number of 2 instead of a negative 1, then his grade would now be a 13; and had the platoon factor been in play it would have raised it two more points, so he would be a 15 for this batter.  Of course, this was the extreme on the other side.

            It really doesn’t matter what sequence the grades are altered.  They will come out the same in the end.

            In answer to your question, it is believed that they all are figured before the Play Result Number is known, but we can’t be certain of that.  Since the DOS game had to deal with these possible grade changes, before the randomization was later added, I would imagine that it was in ahead of the Play Result Number.  I do know that it doesn’t matter as far as the result is concerned.  Today’s computers can easily handle these subtle grade manipulations.

The game was first created and published for machines with an 8K memory, so there wasn’t a whole lot to it in the DOS stage.  It’s come a long way since then, however.

BBW’s Relief Pitchers have certain batter allowance numbers.  Like the starters, when that number is exceeded, the grade will start to decrease as the pitcher begins to surrenders things like hits/walks/runs, etc.  The decline is really fast, too.  So, if a Grade 13 relief pitcher has exceeded his limit, and then gives up a hit, the program will drop his grade (it will not be seen) to an 8.  If he gives up another, it will drop to a 3.  So, you need to take him out!