Preparation, Part 2

I didn't intend to wait this long before starting my 1901 replay, but I was interrupted by a couple of house projects and work-related issues. Most of that is behind me, plus winter is here, so I am finally ready to light the fires on this replay.

Now that part one (link) of the replay set-up is complete it is on to step two: reviewing the player transactions and making additional card adjustments in the area of errors.

Baseball-Reference.com (BBR) includes listings of major league debuts, major league finales, as well as a list of player movement that occurred before, during, and after the season. The ATMgr player transaction file comes with its own list of player transactions during the season, most of which overlap with what BBR has. The three BBR listings were dumped into a spreadsheet, their formats massaged so they could be lumped into a single file, and then additional tweaking was completed so the BBR file could be merged in with the ATMgr file. Now it was time to walk through each entry, remove duplicates and merge notes to come up with a comprehensive list. The final step here was to then walk through the data one more time and ensure the dates specified for adding or dropping a player from a roster corresponded with what the text stated.

I mentioned this previously, but someone had done a lot of research into old Sporting News and Sporting Life magazines (among others) from this time to come up with a large amount of interesting and useful notes. Suspensions, fights, details on injuries, as well as relevant supplemental minor league information all of which help to flesh out the 1901 season and the players.

The next thing to do was to review error totals. After having completed my 1930 and 1949 replays I noted that both seasons had an almost similar number of errors (and not enough of them). However, given the similarity in the number of games played and the similarity in the distribution of player defensive ratings, that really is what I should have expected all along. Further, in both replays, the number of double plays came up short when compared to the actual count.

What this means is that the given fielding ratings were simultaneously too good to reach the number of errors that actually occurred but also not good enough to reach the number of double plays expected. I was curious what the count of errors and double plays looked like through the ages, so here is a chart:

Actual
Season	Games	Errors	DP	E/G	DP/G
1901	1110	5327	1530	4.8	1.4
1910	1249	4436	1730	3.6	1.4
1920	1234	3494	1967	2.8	1.6
1930	1234	3035	2482	2.5	2.0
1940	1236	2834	2338	2.3	1.9
1949	1240	2295	2700	1.9	2.2
1960	1236	2169	2404	1.8	1.9
1970	1944	3327	3695	1.7	1.9
1980	2105	3609	4114	1.7	2.0
1990	2105	3180	3792	1.5	1.8
2000	2429	3447	4711	1.4	1.9
2010	2430	3030	4395	1.2	1.8
2018	2431	2792	4094	1.1	1.7
Replay
Season	Games	Errors	DP	E/G	DP/G
1930	1234	1845	2010	1.5	1.6
1949	1240	1915	2428	1.5	2.0

As you might expect the number of errors per game has routinely dropped through the ages. Better equipment, better playing fields, better practice, better skills, and just better players overall are reasons why this is true. I would have expected to see the changes in the number of double plays per game be comparatively linear as well, but that isn't quite the case.

The 1901 card set has two error numbers on all player cards (located at 53 and 21). That should double the number of error outcomes in a replay, but as the chart above shows it still won’t be enough to even come close to the total number of real errors. I would need to add  a third error number on all cards because 1901 has such a large number of errors but does so in a lesser amount of games as well.

So here was my compromise. I decided to assign a third error number to 50% of all plate appearances, meaning there should be 2.5 error numbers present per plate appearance throughout the replay. It just happened to work out that all players with an on-base percentage below .333 plus all of the XC's players I created account for 50% of all plate appearances.

As far as error number distribution, the APBA model is that each error number should be represented in a normal lineup. They carried that model through when adding the second error number to player's cards. I tried to mimic that in that catchers all got the same third error number and so on through the different positions. Once the criteria was set the next step was to walk through the cards themselves and update the players disk accordingly.

Note: APBA assigned a 23 as the second error number to 1-2 outfielders per team. I had always counted 23 as a rare play number previously, as do I think most APBA players.

I also regraded the pitchers using Weighted Pitching Averages. This normalizes pitching grades across the two respective leagues and eliminates some of the old-time APBA pitching grade idiosyncracies. The grades were released with the original card set ~30 years ago, and APBA has tweaked their own procedures in this area with recent re-issues card sets. I used this method in my 1930 replay with good effect and wanted to use it again here.

Now that this is completed the next step is to start to load the lineups and transactions into ATMgr. The tool is very particular, and that's good. Do the names provided match the names on the player disk? Are players in the lineup active on the roster that day? Am I trying to put a player on two different rosters at once? All of this is now sorted out I should be ready to start play.

I will be posting weekly updates as part of this blog as I have done previously. I like to include pictures of players, just to give a view into a different time. Pictures of players from1901 might be a bit problematic, but I will see what I can find.

I am still reading through the Reach and Spalding Guides covering the 1901 season and want to provide a pre-season preview before I actually get started. These guides have player and team pictures that I might use, plus there are plenty of interesting advertisements from that ERA that I will incorporate.