World Baseball Association
General Category => General WBA Discussion => Topic started by: Huckleberry on October 16, 2016, 11:13:17 AM
-
Okay, I'm going to run the tests now as discussed in the 2105 Season Thread.
Step one - back up the league. Complete.
-
I just realized that because of our custom financial system that just straight simming every offseason then season might screw things up. So I am going to have to go into every team and correct their finances each offseason. My first thought is going into each team and bumping up their cash by $50M each year. That is the average revenues for every team and since I have all the revenues off in-game that will keep the financials as realistic as possible.
If that's not quite right I will look at maybe adding $25M to their budget and $25M to their cash each year, or some other split that adds up to $50M.
-
Step two - Create two new leagues. "WBA Auto-Calc" "WBA No Auto-Calc". Complete.
-
Now I'm loading up the No Auto-Calc league to run the test.
Step one - Set all teams to AI control (team financials have already been set for the first offseason). Complete.
Now simming until 4/1/2106.
-
Now simming through the 2106 season until the playoffs end at which time I'll post the stats and try to figure out what to do with financials.
-
Previous numbers from the other thread:
Going back to the issue of level of offense in the WBA now that the season is complete. Remember that the settings are based on MLB in 1995. I will use the NL numbers from that year considering we have no designated hitters in the WBA.
1995 National League: 0.263 AVG, 0.331 OBP, 0.408 SLG, 4.63 R/G
Here are the historical ABL numbers, using a slash line format in the same order as above (so the 1995 NL was 0.263/0.331/0.408/4.63)
2100 - 0.247/0.325/0.386/4.15
2101 - 0.251/0.327/0.396/4.31
2102 - 0.257/0.333/0.396/4.34
2103 - 0.261/0.340/0.409/4.67
2104 - 0.254/0.329/0.395/4.43
2105 - 0.249/0.325/0.379/4.30
IBL Numbers:
2100 - 0.247/0.323/0.382/4.16
2101 - 0.253/0.326/0.391/4.28
2102 - 0.242/0.319/0.371/4.08
2103 - 0.246/0.318/0.381/4.11
2104 - 0.245/0.319/0.382/4.16
2105 - 0.237/0.309/0.369/3.81
There are the facts. Now we can discuss potential action.
-
2106 numbers with no auto-calc:
ABL - 0.241/0.316/0.369/3.97
IBL - 0.244/0.316/0.374/4.02
-
Added $50M in cash to everyone, that may have been a bit too much but oh well. Worried about league totals, not financials.
-
Simming 2107 now. May be a bit until the next update as I have some errands to run.
-
2107 numbers with no auto-calc:
ABL - 0.241/0.320/0.374/4.10
IBL - 0.244/0.318/0.378/4.05
-
Simming 2108 now
-
2108 numbers with no auto-calc:
ABL - 0.245/0.322/0.385/4.25
IBL - 0.240/0.315/0.372/3.95
-
Adding $25M to every team's cash every offseason now.
Simming 2109 now.
-
2109 numbers with no auto-calc:
ABL - 0.244/0.320/0.379/4.09
IBL - 0.246/0.320/0.389/4.14
-
2110 numbers with no auto-calc:
ABL - 0.244/0.321/0.381/4.03
IBL - 0.250/0.325/0.397/4.32
-
Will start the tests with auto-calc next.
-
Step one - Set all teams to AI control (team financials have already been set for the first offseason). Complete.
Now simming until 4/1/2106.
-
2106 numbers with auto-calc:
ABL - 0.258/0.330/0.408/4.76
IBL - 0.265/0.338/0.418/4.92
-
2107 numbers with auto-calc:
ABL - 0.267/0.339/0.418/5.01
IBL - 0.259/0.329/0.406/4.83
-
2108 numbers with auto-calc:
ABL - 0.269/0.341/0.422/5.08
IBL - 0.270/0.338/0.427/5.15
-
2109 numbers with auto-calc:
ABL - 0.274/0.346/0.432/5.35
IBL - 0.268/0.338/0.413/4.84
-
2110 numbers with auto-calc:
ABL - 0.270/0.340/0.418/5.04
IBL - 0.275/0.346/0.431/5.26
-
So if my math is correct the grand totals are as follows:
Without auto-calc, ABL hits .243, IBL hits .245, WBA total is .244.
With auto-calc, ABL hits .268, IBL hits .267, WBA total is .267.
So turning on auto-calc adds about 23 percentage points to batting averages.
-
auto calc may be OK; more hits for Karachi = more home runs
this not my official opinion - just a first thought. :)
-
I think I'm in favor of autocalc...my only concern is that the avgs increased every year in the test sim. I'm not too familiar with OOTP in terms of long term sims...is there a chance that continues?
BTW, I really appreciate you taking the time to do this, Huck. Sounds like you wasted most of your day on it.
-
I was off and on between other things.
My concern is that runs per game is much higher for equivalent batting rate stats than it was in real life. In 1995 the NL averaged 4.63 runs per game. With auto-calc the WBA leagues were nearly a half-run higher for similar slash lines. I will look at it a bit more closely tomorrow, but perhaps we could try a test or something with a different season. Over 5 runs per game is too high.
-
I haven't looked into it much, but it did seem to me like our ERAs have been petty normal looking
-
Yeah, I hope to have time to investigate. Seems like the best explanation is there is some combination of better baserunning and worse fielding happening in OOTP than in real life.
-
Huck, if that last number is average number of runs per game TOTAL (rather than per team), then isn't 5 too low? I would've figured the two median teams (inexact, I know) to be more than 2.5 runs per game each...
-
Stated differently, if the average team gave up 2.5 runs per nine innings, and every pitcher was identical, then every pitcher on that team would have at most a 2.50 ERA right?
Am I missing something? I must be...that seems like an all-time staff.
-
That's per team per game.
-
A suggestion --- look also at walk totals. If the auto-calc seasons had significantly more walks than the real-life seasons, that could account for the fact that more runs are being scored despite similar batting averages.
-
Walks should also be reflected in the OBP of the triple slash line right
-
That's true. Sorry, my bad.
-
Okay, thanks to this website:
http://www.baseball-reference.com/leagues/NL/1995.shtml
I know that the 1995 National League defensive efficiency was 0.688 for the year.
Here are the historical defensive efficiency numbers for the WBA from the almanac on my hard drive:
ABL -
2100 - 0.685
2101 - 0.685
2102 - 0.679
2103 - 0.677
2104 - 0.688
2105 - 0.688
IBL -
2100 - 0.689
2101 - 0.683
2102 - 0.692
2103 - 0.693
2104 - 0.694
2105 - 0.702
So it's possible some of the difference is due to that, but it doesn't seem to explain it all.
-
Although we need to keep in mind that defensive efficiency directly affects the slash lines, particularly batting average and slugging percentage. So now I'm back to wondering why with similar slash lines are we seeing so much more scoring in the WBA.
-
Okay, so I decided to look up errors per game because errors would increase runs per game while decreasing (or not increasing at least) the offensive slash line.
1995 National League - 0.764 errors per game
ABL:
2100 - 0.594
2101 - 0.583
2102 - 0.581
2103 - 0.617
2104 - 0.576
2105 - 0.588
IBL:
2100 - 0.665
2101 - 0.688
2102 - 0.665
2103 - 0.579
2104 - 0.599
2105 - 0.580
A few thoughts. First is that apparently somebody taught the IBL how to field a baseball in the 2102-03 offseason. The second is that this makes no sense whatsoever. Fielding in the WBA is actually about the same efficiency wise but with a much lower error rate. Yet our runs scored are higher relative to slash line than we would think they should be.
In order to have the same efficiency but with a better fielding percentage the real life answer is that the fielders in the league lack range. But, again, if they simply aren't getting to balls then that should show up in batting average. Now another test.
-
Okay, I may have a promising possibility. After running the last test, I thought about double plays. A double play is simply an 0-for-1 when it comes to league batting average but it obviously creates two outs. Therefore if the WBA is turning fewer double plays than are turned in real life then that would explain the higher runs per game compared to slash lines. And my first check was promising so I'll look those numbers up now.
1995 National League - 0.917 double plays per game
ABL:
2100 - 1.073
2101 - 0.977
2102 - 1.093
2103 - 1.076
2104 - 0.913
2105 - 0.932
IBL:
2100 - 1.102
2101 - 1.010
2102 - 0.965
2103 - 0.983
2104 - 0.864
2105 - 0.837
Well that doesn't help. Although the plummeting number of double plays in the IBL could definitely be affecting things. This entire exercise has created more questions than answers so I think I'm simply going to find a season as close to halfway between our 2105 numbers and the 1995 NL as possible and then run an auto-calc test against that.
-
The season is 2010. I will run the test now.
2105 WBA - 0.243/0.317/0.374/4.055
2010 NL - 0.255/0.324/0.399/4.33
1995 NL - 0.263/0.331/0.408/4.63
-
Sorry guys, I'm going to hold off on the 2010 test. I have a new proposal I'd like to get input from everyone else on:
Modifying my era_stats.txt file to have league totals in the range we'd like to see for years moving forward.
I truly believe this is the best option for the league but it brings up questions about how to do it. Here are some of my thoughts:
- I would start out by setting the years so that our league totals are listed for the 2105 season.
- I would then slowly ramp from our league totals toward the 1995 National League totals. My first inclination is to meet 1995 in 2110 so there is a reasonable transition.
- For years after 2110 I would write a formula that somewhat randomizes league totals between what we had in 2105 and 1995 league totals. I would try to set it up so that there are cycles in the run-scoring environment instead of the league attempting to hit the exact same totals every year.
- After doing so I would run a LONG auto-simulation (probably by turning finances ON in-game so I don't have to mess with that) and then showing everybody the league totals output.
- I would be perfectly willing to post the era_stats.txt file that the league uses as obviously I have visibility there. I don't plan to actually look at it after I first write it. Even if I were tempted to do so these tests have shown that the league fails to hit the actual totals by quite a bit from year-to-year so I would consider it a waste of time as far as anyone trying to gain an advantage. That's why I think this method works.
Thoughts?
-
I am in favor of your proposal - a 5 season or so adjustment should present minimal hardship for teams built around current performance. thanks for all the work Huck!
-
I fear change, i see no reason to adjust anything....but i trust you do whatever right! (rebuilding soon)
-
Huck, you've asked for feedback, so for what it's worth, here's my two cents (and that might be ALL that it's worth).
Let me start out by saying that I enjoy taking part in the WBA and that any criticism I offer is in the spirit of fine-tuning an already enjoyable experience. At no point do I want to be misconstrued as being on the warpath about anything. Even if NOTHING changes, I'll still find the WBA to be a lot of fun. But almost anything can be improved, and with our league having just completed a season in which the batting averages were lower than anything ever seen in real life, it's clear that we have an opportunity to make some improvements, so that would seem like the logical thing to do.
You've suggested starting out with the league offensive totals from 2105 and then slowly increasing them over the next 5 seasons until they're as high as the 1995 NL totals. My opinion --- our 2105 league offensive totals were fine with regard to home runs, but historically low with regard to batting average, even lower than the Deadball Era, so I wouldn't want to see the 2105 totals used as a starting point; I'd like to see more offense RIGHT AWAY. And instead of a gradual 5-year increase, how about increasing the batting averages over a 3-year period instead? (My desire here, obviously, is to get the league batting averages up to a historically normal figure sooner rather than later.)
Also, I'll reiterate that it's just the batting average that's the problem, not the all-around offense. As far as I can tell, we're already at a historically normal level with regard to home runs, walks and stolen bases. It's just the batting averages that are out of line with real-life baseball.
Lastly, you suggest that, after 2110, you'd write a formula that randomizes league totals between what we had in 2105 and the 1995 NL totals. The purpose of that would be to ensure that there are up-and-down cycles with regard to offense. I'd vote against that, for several reasons. First, it seems like a lot of extra work for you. Second, there will always be some degree of up-and-down cycles, due to available talent, managerial choices and just random luck, so there's no need to force those cycles. Third, if you force those cycles so that the offense is at 2105 levels one year, then at 1995 levels the next year, it would make it almost impossible to operate our teams properly; at the 2105 level, a starting player who is hitting .250 is doing just fine compared to other players (since overall batting averages are so low at that level), but at the 1995 level, a starting player who is hitting .250 is a candidate for replacement. It would add confusion rather than enjoyment. I'd rather just reach a historically normal level and then stay there, letting the up-and-down cycles happen naturally, as they do in real life. That would let us focus entirely on running our teams, rather than worrying about whether the current season is a high-offense season or a low-offense season.
That's how I feel about the situation. I hope it's more helpful than controversial. I'm really not trying to stir up controversy.
-
I see no reason to change anything.
That said, I approve of the commish proposal as stated.
If changes are made to it, I reserve the right to retract my approval.
-
Coop, that's exactly why I'm proposing this method. Given that our home runs are close to right the fact that we will be using environments somewhere between our 2105 output and real-life 1995 output means that they will stay right in that range. I'm not applying a particular percentage increase across the board to every stat, each stat will vary between what it was in 2105 and what real life was in 1995.
So if our batting average is 0.243 in 2105 and real-life was 0.263, then we are going to target batting averages between those two numbers. Our 2105 home runs per at-bat number was 0.026 while 1995 was 0.028, so we are going to stay right around there moving forward.
-
Okay, I have scaled the 1995 National League stats in order to insert into the era_stats.txt file, and I also pulled the same numbers for the WBA 2105 season. Here are some of the major differences, scaled to the number of at-bats in the WBA:
Stat - WBA/1995
Batting Average - 0.243/0.263
Strikeouts - 0.239/0.193
Runs Per Game - 4.052/4.632
Double Plays that are GIDP - 0.913/0.801
OF Putouts per Out - 0.346/0.307
ERA - 3.73/4.18
K/BB Ratio - 2.368/1.996
So we have too many strikeouts and too many fly ball outs (although we're not turning enough of those into double plays). Unfortunately I think I've read that this is an issue overall with OOTP somewhere, so our changes will only be able to do so much.
-
I'd noticed that the strikeouts were high but wasn't sure if it was just my team. For example, my rookie first baseman, Bang Luo, had 289 at-bats this year and struck out 113 times. At that rate, if he'd played every game and gotten 600 at-bats, he would have struck out 234 times (the all-time real-life record is 223, by Mark Reynolds).
OOTP limitations are undoubtedly going to make it impossible for us to fine-tune every stat to the point where we're right in line with the real-life major leagues, but at least it sounds like you have a good game plan for moving things in the right direction. Hopefully we won't have any more seasons where one of our leagues has a batting average of .237 (and only one .300 hitter in that entire league).
-
I fear change, i see no reason to adjust anything....but i trust you do whatever right! (rebuilding soon)
not sure what the propsal is...but it doesnt matter much, i say change nothing...but if people think more BA helps, im ok with that.
-
Just to let everyone know, I'm still working on the new era_stats file but I hope to have some tests on that tomorrow.
Also, we still need a new owner for the Scottish Claymores.
-
I am OK with anything that is decided, but I am still of the opinion that we have very few good contact hitters in the league, hence low batting averages. We have maybe 35-40% of the 7+ rated contact hitters that other leagues I am in have. The pitching on the other hand looks to be very similar in overall ratings, which will again result in lower batting averages.
-
I am OK with anything that is decided, but I am still of the opinion that we have very few good contact hitters in the league, hence low batting averages. We have maybe 35-40% of the 7+ rated contact hitters that other leagues I am in have. The pitching on the other hand looks to be very similar in overall ratings, which will again result in lower batting averages.
Agreed, but that's another battle against the engine as the player creation modifiers are all at default.
-
Here are the results of the test run of the era generator I wrote. I just picked runs per game as the representative stat, the line graph would look the same for any stat:
(http://i.imgur.com/hFoUqBY.png)
-
Agreed, but that's another battle against the engine as the player creation modifiers are all at default.
I'm not trying to shoot this idea down or anything, but if the problem is with the player creation mods., then shouldn't that be what we are looking at. If other major changes are made and the 'player creation' starts going in a more hitting direction then we will have tons of hitting instead.
I guess I am repeating myself now, so I won't say anymore. I am playing regardless!
-
Huck, if I'm reading your graph correctly, it looks like this test gave you what you were aiming for. In a post from yesterday (listed as Reply #38), you showed that WBA teams in 2105 scored 4.055 runs per game, NL teams in 1995 scored 4.63, and NL teams in 2010 scored 4.33. My understanding was that you were shooting for something in the NL 2010 range, something around 4.33. The graph shows results that are generally in the 4.3 to 4.5 range, so I guess that nails it.
By the way, what was the league batting average from this latest test (over, say, the first five years)?
-
That wasn't the simulation yet. That was setting up the era file. Now I have to run the simulation soon. Obviously trying to hurry as I don't like the dead spot we're in so I'm planning to get this done ASAP.
-
I'm good with whatever gets decided, but I'd vote for keeping it the same. I agree that there aren't many great contact hitters. I notice there are a lot of well rated hitters coming. The drafts have been loaded. I worry about it getting too offensive soon. But maybe eventually the auto-calc would regress the offensive numbers down if need be.
-
Huck, my advice would be to run some final tests then make an executive decision. You have always been fair so there are no worries there- i'm concerned we are going to start losing league interest with too long of a stagnant period.
-
Agreed. My final step is going to be to run two long-term sims. One with the revised era_stats file and one without. Then I'll pick what looks best.
I do agree with the hesitation to meddle with the modifiers, though. I'd almost rather have improved player creation modifiers but the biggest issue is our league is simply too young to really know if this is a small sample size situation or a structural issue.
-
I would also put my vote on figuring out what the best course of action is, and just make an executive decision. The long break is scaring me also, I do not want people to lose interest.
-
Okay, I will make a decision this weekend after running sims. Found this thread:
http://www.ootpdevelopments.com/board/ootp-16-general-discussions/257544-why-league-offense-so-low.html
Those numbers in the OP sound a lot like our numbers in the WBA. So I'm starting to lean toward using the auto-calc button.
-
One thing I just thought of... How much of a difference do you think a DH would make to league offensive numbers? Has to be at least a small difference
-
One thing I just thought of... How much of a difference do you think a DH would make to league offensive numbers? Has to be at least a small difference
I am only comparing to the National League so that we can remove that as an effect. Typically speaking the DH does add somewhere around 6 points in batting average and about 0.3 runs per game to the league average.
-
UPDATE ON LEAGUE PROGRESS:
Our next sim will be Sunday evening. I will continue running the tests in the meantime (my PC is finishing up a 100-year sim with the current settings right now, they just take a long time). I don't have to have this figured out until right before next regular season, so we're moving forward.
-
Results of the 100-year sim with current settings (nothing changed)
(http://i.imgur.com/dSOlo1f.png)
(http://i.imgur.com/bpIQVh8.png)
(http://i.imgur.com/xu1n4mz.png)
(http://i.imgur.com/ATSz9LD.png)
-
Personal opinion --- this doesn't look good. It shows the IBL's batting average going down under .235 in the near future, and runs per game dipping to 3.8. Those are the kinds of numbers they would have gotten in the Deadball Era if they'd replaced baseballs with lumps of granite. Let's not go there.
I'm confident that we'll see much more realistic numbers when the test is run with the auto-calc setting.
-
That one is running now.
Hopefully done around 7:00.
-
Those IBL graphs were definitely not pretty. Strengthens Twinkletoes' HOF case though.
-
Results of 100-year sim with era_stats file:
(http://i.imgur.com/CR5fSlv.png)
(http://i.imgur.com/Ky0NQNj.png)
(http://i.imgur.com/38Uc1Zn.png)
(http://i.imgur.com/9molB43.png)
-
Batting averages look good to me. Runs scored have a weird spike early on, going as high as 5.4 at one point, but then they settle down to 4.6. I wonder what's causing that big spike? Anyway, in general this seems to look pretty good, I think.
-
Yeah, not sure what that spike is about. May run a few more short-term tests tomorrow just for 10 years to see if it happens again.
What I have learned for sure is that for equivalent league batting averages there is more scoring in OOTP.
-
It might be something that's virtually impossible to fine-tune or to identify in standard statistics. Like for example maybe in real life, a runner scores from second base on a single 50% of the time, but in OOTP he scores 75% of the time. That sort of thing would account for more scoring despite similar batting averages and on-base percentage. Just a thought. Anyway, it looks like something we just have to live with.
-
It might be something that's virtually impossible to fine-tune or to identify in standard statistics. Like for example maybe in real life, a runner scores from second base on a single 50% of the time, but in OOTP he scores 75% of the time. That sort of thing would account for more scoring despite similar batting averages and on-base percentage. Just a thought. Anyway, it looks like something we just have to live with.
Not a bad theory...gives me an idea.
-
Never mind. Found the source of the scoring spike. Fielding percentages tanked and errors went through the roof. Fielding percentages were around 0.960 for several years instead of 0.980 with errors per game doubling from about 0.76 to about 1.56 in that period. The fielding percentages eventually stabilized but I will be looking at the modifiers to get that figured out.
1995 NL runs per game were 4.63 with a league ERA of 4.18 As an example, that crazy 5.45 runs per game in the chart for the ABL in 2109 was with a league ERA of 4.47. So while scoring was a bit high, errors were the real problem.
-
Okay, I'm in the middle of a more slow-paced long term sim where I manually push the auto-calc button each year. This allows me to correct the ridiculous position modifiers that OOTP is applying automatically. The error rates are being set to absurd values (between 3.000 and 4.000 is common) so each season I will set those manually to try and get the correct fielding percentage.
-
Update on this. Using auto-calc can cause extreme individual stats depending on league roster makeup. In 2122 during my slow sim, we have several 50+ home run and 200+ strikeout players in each league each year because many of the rosters are set up with high contact, low strikeout strategies by the AI. Because of our smaller league sizes (only 10 teams in each league), this means the league totals have to come from somewhere.
With that in mind I will continue my research but I now have reason to modify the approach I thought we would take.
edited to add - I'm thinking that I will cap the modifiers at certain levels. This may strike a balance between crazy modifiers and individual performances and leaguewide offensive explosions/droughts. It will also allow our league to develop its own personality and environment. I will test these caps for a few seasons, I think I'll start with 0.500/2.000 modifier min/max. Some of them have gotten up to 4.000 and as low as 0.300 in the future seasons.
-
And never mind again. Hahaha. The bad news is I didn't know anything about how OOTP really worked before, the good news is I'm learning.
I had checked the automatically import historical player creation modifiers. The problem is that importing league total modifiers is based on the era_stats file I modified, but the player creation modifiers are a separate file. Ugh. So I had players in the future that were created based on real-world 1915, for example, then they were trying to match the league totals in my modified file. Disaster predictably ensued.
I will have to start this one over.
-
Well, at least you're gaining a lot of knowledge that will undoubtedly prove useful in allowing you to set up the WBA in such a way that the teams and players perform as desired. I'm sure it's time-consuming and aggravating now, but it'll pay off for years afterward. And it's something that NEEDS to be done, since your tests with auto-calc set to OFF showed that the league's current ultra-low batting averages are going to continue unless changes are made.
-
Honestly, just reading this makes my head hurt. Kudos to you, Huck