6.5 to OOTP 2006 Online League Conversion Blog

[sporr]

Well, apparently some people here have been reading my blog on Baseball Maelstrom's conversion to the new version. It started as an owner request. Keep in mind that some of this info and issues are specific to our league and may not apply to yours. However, it still gives a good insight on some of the issues that you may run into when converting a league over. Feel free to visit our league forums or website for more discussion.

You can find the original thread on our site:

http://www.baseballmaelstrom.com/boa...opic.php?t=212

[sporr]

Well, I've converted a test league over to take a deeper look at stats and the College League. Here are some thoughts.

INITIAL TEST: I did one conversion, but I forgot to turn down player creation ratings, so many of the minor-league "fill-ins" were better than some team starters. (Tester's note: Since the new game requires full minor leagues, the game will create players upon conversion to fill team minor league systems.) I also forgot to turn off player movement, so the computer traded a bunch of guys around. I simmed the season anyway to see stat output and saw the results detailed below. At first, I thought HR might be too high because Topper Lundgren had 35 or so HR at the half. But he was an oddity and cooled down to only hit 50 on the year anyway.

In my 2nd test, I turned down player creation so the fill-ins were crappy. I also set all team to human manager and didn't allow movement. I turned off injuries so that I wouldn't have to micro-manage during the sim. I fixed attendance based on 1st test, and also stolen base frequency (see below).

1 - The college league players will need some adjustment. There is no way to create players with high talents and lower ratings. What really sucks is that you can't set different ratings scales between leagues in the same universe, so we can't set it to 1-5 ratings like we did before since it has to match the 1-10 that we use in our league. Anyway, if you set the Independent College League to be like Single-A, you get players with TALENTS maxed out at about 5's. If you set the Ind. League to be like the Majors, you get players with talents and ratings at Major League level. I think what we'll need to do is let everyone know that when a player comes over from the MCL, his actual ratings will be half of what is listed. That way we can have guys with normal talent levels but not major league ready like we've had in the past.

I'm going to try to find ways to do some mass editing. That way, the MCL will be less time-consuming to set up. As it is, all of the ages will need adjusted to college levels and with 16 teams x 25+ players per team, that's 400+ players to edit. (Tester's note: I have already gotten some ideas on mass editing using the import/export roster function. I still need to test the ramifications of using this each season. i.e. in 6.5 this would dump all players into the minor leagues and we don't want to force owners to move players back every season. Also, as you may notice we are expanding the College League to 16 teams) The other problem is that if we set up the league to have ML level ratings/talents, we'll have a lot more than just a few players with good enough ratings to enter our league. I'm not sure of a good way around this. I guess the other option is to set the league up to be like Single-A so everyone has 4's and 5's for ratings but then bump up a few players per team to be at higher levels. Problem is they might put up godly/unrealistic stats.

2 - Offense. At default levels, batting averages are WAY down. Down about 10 points per league in the .240-.250 overall range. HR seemed to be right on the money, but doubles, triples and base hits in general were down a lot. There were probably 4-5 pitchers PER LEAGUE with ERA's under 2.00. Runs scored were down 10% as a result of fewer hits.

3 - Errors. I also noticed that errors were way too low. Only one team had more than 100 errors on the season, and many were in the 50-60 range. Comparatively, in 2017 there were just a few teams that were UNDER 100 errors.

4 - Stolen bases were way low at defaults. The leader had about 50. I ran another season with default stealing cranked up to VERY HIGH instead of AVERAGE and it was better. Leaders around 70. But overall still about 10-20% lower than "normal" for our league. Might not be a bad thing. Also, individual stealing settings weren't in effect. Once a Ty Cobb gets cranked up to Very Often, he'll probably steal a normal number of bases instead of the 66 that he did in the test. So this is probably fine.

5 - Finances. In my first test, I left average attendance at 26000 or whatever it defaulted to. At that level, revenue was way too low and many teams saw major attendance drops. In my 2nd test, I averaged our attendance from last season and it was closer to 31000. I entered this in and the end of year finances were right on the money. I also kept $10 as the default ticket price.

6 - Overall hitters/teams. Overall, I didn't see any strange individual stats other than the low offense already noted. The top hitters led the league as did the top pitchers. So it looks like players should perform as we expect. Teams did as well. Our standard playoff teams all did well and the Mets won the division again...barely.

NEXT UP: I plan to tweak the error settings as well as the offensive categories to get things closer to what we expect. It doesn't have to be perfect, but the way it was, we would have seen the lowest offensive output the league had ever seen. And both tests I ran fell along almost identical lines, so it wasn't a fluke. I'll try to raise Cobb's individual steal setting as a test case to see if that gets us to the level we expect.

[sporr]

This week has been hectic, but I found a good chunk of time to dial in our stats.

Basic Stats:
Basically, I made a quickstart out of a basic Maelstrom conversion with the college league added. But, so that I could compare stats better to 6.5, none of the new Legends were included. I turned all teams to not allow AI roster movement, turned off injuries and waivers and also boxscore saving to speed things up for the tests. Then I just simmed some seasons, each time making tweaks to our league settings to get the stats closer to "normal". Since my prior two tests had produced similarly low offensive results, I started by averaging league stats between the two sims. The AL has always had more offense (averages are about 10 points higher), so I did it on a league by league basis. Fortunately, the difference between AL and NL stats was still basically in the same ratio.

The next thing I had to do was get our normal league stats. So I pulled the league hitting and pitching stats from 6.5 for the last few seasons to compare.

What I found was that our current league had 3% more hits, 30% more doubles, and 20% more triples. Meanwhile, we had 4% fewer Ks. My guess here was that these extra Ks were taking away some of the chances for hits in the new game. Finally, despite the overall low offense, HR were actually up about 7% in both of my tests. Walks were actually right where they should have been.

Now was the trick...I didn't figure I could just bump up the League Totals straight by those percents because, all things equal, if I reduced Ks, more hits would result without me having to raise hits simply because there would be more balls in play. Also, by raising doubles and triples, hits would go up anyway. However, HR had to be dropped a little which ended up making up some of that difference.

Well, my idea ended up being wrong. Apparently, the added doubles and triples were taking the place of singles, so averages stayed down. So, my next tests raised everything to where it should have been. Strangely, I had to raise hits even higher than I expected to get averages in line. But when I finally finished, I had stats that nearly matched our 6.5 results.

Here is where I ended up:
AB: Leave this alone
H: 1.050
D: 1.300
T: 1.200
HR: 0.930
BB: Left this
K: 0.960
HBP: Left this (see below)

I want to add that the new game makes changing this figures MUCH easier. Instead of dealing with the raw league totals, as you can see, you simply work with the percent increase, decrease that you want.

Fielding Stats:

Somewhere in the middle of my simming of seasons, I started wondering...I'm dialing in our stats to compare well to our 6.5 league stats, but how did they stack of to real-life data? Because even though our league breaks from reality a little, we'd still like to have some "realistic" MLB-type stats. I didn't figure we were that far off in the main stat categories.

I checked out team stats on MLB.com from the last few seasons and, as expected, we were well within normal ranges in most cases. There were a few exceptions...

As I noted in my reactions to initial testing, error rates seemed to be WAY down. But what about real-life error ratings? Well, real MLB teams average around 100 errors per season on the whole. The Maelstrom has averaged just slightly higher than that. But my test sims were coming in 30% below that number (averaging around 70 per team).

Fortunately, the new game has an engine configuration file that allows you to control this and other "hidden" factors. I opened it up and raised the error ratings to 140% (since we were 30 below where we should be, 30/70 is about 42% to get us to around 100 errors per team). Even though this was slightly below where we have traditionally been, I always felt that our error rates were just a little high, so this would bring us more in line with MLB numbers. Double-plays seemed to be a little low in the tests compared to our league and MLB, so I raised that slightly to 105% as well. These results were right on the money in the next test.

Other Stats:
Offense, pitching and fielding were all right around where they needed to be and the results had the familiar feel of the stats that we have been used to seeing. However, I had read on the OOTP boards that one of the Dream Team testers had noted very high Wild Pitch totals in his converted league. Apparently, WP ratings for pitchers (and HBP ratings as well) are based on 550 batters faced (supposedly a full season). The problem is that most pitchers face double this number of hitters in a season, so the game was generating some pitchers that would have very high totals because of this.

But did this impact us? Here's where real MLB data came into play again. Looks like teams average just under 50 WP per season. Looking at my tests, teams were averaging around 60! 10 may night seem like a lot, but that's a 20% increase from normal MLB rates. Again, this is where the config file came in handy. I very easily turned down our WP rates, I think by about 30-40%. Interesting to note that a few pitchers in the league had over 40 WP on the season before I made the adjustment. There's still an outlier or two, but I noticed there was in 6.5 as well! For these seasons, we've had freakishly high WP numbers and I've never noticed. Well, fortunately, that will be fixed for most players.

HBP totals were next on my radar, but the overall numbers seemed to be pretty close, so I figured we'd just leave them alone. Though, again, there were a few pitchers that broke the mold. When Tim Sabia is pitching, I don't want to be anywhere close to the plate. He hit over 40 batters last season and he averaged around that in my test sims as well!

Passed balls were close, but I adjusted them down a little to get them more in line. [Tester's note: I don't have the config file here at work with me, but I'd be happy to post the details for those interested]


Stat Recap:
After all of my testing, the totals were finally right where they needed to be. Even some secondary totals like WP and PB were in line with real MLB data. As it stands, the only thing that's a little off is Complete Games. We're consistent between 6.5 and OOTP 2006, but the AL is nearly double the MLB average. Even so, I think this is fine. We have some Legend pitchers in the league and we certainly don't want to tone down their endurances. Besides that, reliever/closer usage is already maxed out. I'm comfortable with those totals. One step closer to "real" conversion...

[sporr]

Alright. Well, I guess it's fortunate that I took today off from work. I have spent a total of 6-7 hours getting all of the minor leagues set up as well as the College League.

I knew that there was a lot that you could edit after the league was already "started", so I left a lot of the details for later. The problem is that the game had some strange ideas about the newly created minor leagues.

MINOR LEAGUE STRUCTURE
The game correctly created enough minor league teams (one affiliate at each level for each Major league team). However, it didn't seem to care what kind of divisions it set up, mixing AL and NL teams in a seemingly random fashion within the structure. You might remember in 6.5 that the minor league structure matched the structure of the Majors. Not necessarily so in 2006. So I did some research on the minor leagues. It was also a good time to correct some minor league team names since many of them were outdated.

In the real-life minors, AL and NL affiliates are also shuffled amongst the leagues, so this didn't seem to be as big of a problem as I first thought. I just figured that I would use the real minor league structure and problem solved. Well, yes and no...

First, the Majors have more teams than we do, so some minor league affiliates simply didn't exist in our league. Combine that with the Tokyo Giants and Homestead Grays and there were some decisions to make. I started with the same basic structure of AAA, with the Independent and Pacific Coast Leagues being subleagues of the Maelstrom AAA system. However, there needed to be some reshuffling to get a decent number of teams in each division. So I ended up with four 4-team divisions in the PCL and three 4-team divisions in the Independent League.

AA was similar...except that there are 3 leagues in the real-world AA system. Unfortunately, there is no way to add a league in OOTP 2006, so I would have to work with the two that I had initially created. I decided to do a simple geographical split, with a Western and Eastern League, each containing a North and South division. This broke with the real-life Eastern, Southern and Texas Leagues.

A became a little bit of a problem as well. MLB has three levels of Single-A: Advanced Class-A, Class-A and Short Season Class-A. Well, most of our teams were from Advanced Class-A, but we had a few from Class-A as well (mostly the Midwest League). Well, even Advanced Class-A had 3 Leagues, so again there had to be some adjustment. I decided to use the California League and Florida State League since there was so much representation with teams from those states. I did a West and East for each and placed teams based on geography as much as possible.

The other issue is that in MLB, each of these leagues works independently, even if they are at the same level. In our world, the AAA or AA or A sub-leagues will play each other for the championship and All-Star games. That said, unlike 6.5, we will have AL affiliates playing NL affiliates. But, there is still no inter-subleague play even at lower levels. So, while not 100% accurate to real-life, it's what I'll call "historical fiction". I figured this would give us the flavor of real-life even though it won't mimic it perfectly.

TEAM SETUP - MINORS AND MAJORS
Well, before I could even setup the structure above, I had to correct all of the minor league team names. Each of them is true to life (apart from Homestead which had some creative liberties taken).

Once the structure was in place, I realized that I had to make sure all of the home cities were correct for all of these teams. The game did a decent job of pulling in the correct city in cases like the "Louisville..." or the "Akron...". However, it was lost with teams like the "Iowa...". For those, I had to manually enter the city names and select the correct matching city in cases where there were cities in multiple states with the same name. I figured I'd doublecheck all team cities to be sure they were accurate. Well, in some cases the game had chosen a matching city in the wrong state. For example, the Ottawa Lynx were not assigned to the Ottawa in Canada, but one in the States. I had to change the home country to get it to pull the right city. In other cases, the city from the correct state had to be chosen. (The Salem Avalanche are from Salem, VA NOT Salem, MA for example).

While I was at it, I figured I'd check out our Major League cities as well. Sure enough, some teams had been assigned to the incorrect city or no city at all. Neither Chicago team or New York team had been assigned since it tried to find "Chicago (A)" or "New York (N)" and not simple "Chicago" or "New York". This was easy to fix (though time consuming to check them all!).

Ahhh...but there was another minor issue that needed fixed as well. The game had generated all of these stadiums for each minor league team. The problem was, I had just changed their city and/or team nickname. Well, in most cases the stadium was named for one of these. So I had to rename all of the stadiums.

Last, but not least, all of the College Teams needed corrected as well. Commish Dugan came up with 8 more colleges to use to round out our expanded MCL. Well, I still had to research to find the correct home cities and nicknames...and then had to put them in divisions based on geography again. On top of all of this, the generic logos needed regenerated for everything I had done to this point since team initials had changed in almost every case. That's another button click on 100+ screens.

So, am I done? Well, I'd still like to have team colors that match real-life (and team logos too for that matter!)...and stadium names that match. But that is minor and can be done at another time.

THE COLLEGE LEAGUE
The basic structure was the "easy" part. There still had to be a way to get player ages in the Maelstrom College League to be accurate. As it was, there were many players in their 30s in there along with players with various amounts of service time.

Using the Export Roster function in League Setup, I got the MCL roster into Excel. Doing a find and replace, I corrected all birthdates to put players between 18 and 22. Sure there might be a real-life exception to these ages, but this is close enough for us. I also zeroed out service time. But that still left another issue...

Originally, Commish Dugan and I figured that we'd have to leave on League finances for the MCL to allow players to be signed by our league. So salaries and income had been adjusted down appropriately. In addition, players were to become FA after 4 years to simulate graduation. However, the game had randomly generated contracts for the players in the league. So some 22 year olds had 3 year contracts while 18 year olds had 1 or 2. This just wouldn't work. The only way was to turn finances off completely. Players could still be assigned to teams when they were drafted using Commish powers. Still some work to be done every season adjusting ages for entering players and retiring seniors that don't enter the draft, but that's not too bad and beats having to edit every player's age manually.

PARTING THOUGHTS
I was happy to see some Major League quality players already in the MCL (game generated) despite the AA level assigned to the league. There will still need to be some manual adjustment to get things to where we would like them to be though. That part isn't done yet.

New Legends still need added.

Oh, I noticed when looking through our options that the game had set average media contracts at $5M. Well, our actual average is around $38M, so I changed that so we don't have any issues in a few years.

Injuries are a little strange. Everything carried over from the prior version. However, since we jumped back to January 1st, players have an extra 3 months of recovery during the off-season.

Hopefully, all of the other things will be cleaned up this week and we'll be ready to roll! Stay tuned!