It is my opinion (however inexperienced or limited) that there ARE some minor issues within the algorithm that cause OCCASIONAL highly unrealistic occurrences. Unrealistic outcomes of individual matchups and even games. It doesn't happen often but it DOES happen. Just my opinion.

Using modern era players in a progressive (for one example) with no owner intention (AAA guys, etc) and getting game results like 57 -3 or something is just not very realistic. I don't know how you fix it!

I've noticed that when some players under perform on my team others seem to over perform. So it evens out somewhat. I've also noticed that when players start off cold they heat up later and vise versa. In a 162 game season it is the manager's job to know the players (hot & cold) as the season moves along. How some have 5 or more teams going at the same time and manage them properly, unless you're retired and living alone, is kind of a tough task.

Posted by laramiebob on 7/30/2016 1:10:00 PM (view original):
It is my opinion (however inexperienced or limited) that there ARE some minor issues within the algorithm that cause OCCASIONAL highly unrealistic occurrences. Unrealistic outcomes of individual matchups and even games. It doesn't happen often but it DOES happen. Just my opinion.

Using modern era players in a progressive (for one example) with no owner intention (AAA guys, etc) and getting game results like 57 -3 or something is just not very realistic. I don't know how you fix it!

See, in my mind this means it's working correctly. Highly unrealistic is more like highly improbable. But the SIM has run well over 100,000 seasons, compared to the 140 or so in real life. When you have a huge sample the extreme deviations *will* pop up on *occasion*. That's sort of how statistics works.

OK. I get that. How do you feel about 5 or 6 triples in a row?

Posted by laramiebob on 7/30/2016 8:30:00 PM (view original):
OK. I get that. How do you feel about 5 or 6 triples in a row?

Shouldn't happen often, but especially in a league with high quantities of deadballers and parks like Petco, would not be surprised if it happened at some point.

It has. And it was highly unlikely.

Let's review what "highly unlikely" means a bit.

Looking at baseball-reference.com, the number of triples per game has historically declined, from hovering in the .4s and .5s in early history to the current rate of under .2. I'll use .35 as a rough historical average, because I don't feel like doing detailed calculations to find the historic average.

By similar reasoning, I'll declare 1.35 as the historical WHIP. It's probably not perfect, but we're just going with rough estimates here. Figuring that rare events like HBP cancel other rare-ish events like double plays, and we can then conclude 4.35 batters per half inning as reasonable for a rough calculation.

This yields 78.3 batters per game. Thus we get a triples average of about .00447. That is to say, with a typical at-bat until perfectly neutral conditions, a triple has a probability of happening 0.447% of the time. Five triples in a row happens at a rate of 1.78*10^-12. Six triples in a row happens at a rate of 7.9*10^-15. So, very unlikely...But that's for any given sequence of six batters.

I'm playing in MLB116256 right now. I'll use that as a rough estimate of the leagues that have been played. Let's further assume the standard 24-team, 162-game schedule is a perfect average for simplicity -- smaller and larger leagues cancel. So that's 12 games per timeslot, 162 timeslots for a total of 1944 games per regular season. Let's call 4, 5.5, and 5.5 average playoff series lengths, and we get an average postseason total of 32.5 games. I'll lop it to 32 to account for some shorter seasons, as larger leagues don't generally have expanded playoffs. So that's 1976 games total, on average, per league. That's 229,721,856 games played. Use the 78.3 batters per game figure, and... 17,987,221,324.8 is the number of expected at-bats. Obviously you can't have a partial at-bat but such is the nature of estimation. In scientific notation, this is about 1.8*10^11. So we're still looking at a decent-sized gap of 10^4, or about 10000 times for six triples, but only about a factor of 10 for five triples. But this is all assuming perfectly neutral conditions. If a team builds of all deadballers, plays in Petco or a similar park, and is going up against a homerun-suppressing deadball pitcher, this factor can start to shrink tremendously. I would conclude that at one point in SIM history, a five triple run might well indeed happen. Six triples is a bit harder, admittedly, but it's still possible especially if we design to hit triples.

Posted by laramiebob on 7/30/2016 8:30:00 PM (view original):
OK. I get that. How do you feel about 5 or 6 triples in a row?

I'll believe this when I see the boxscore.

Posted by uncleal on 7/31/2016 1:00:00 PM (view original):
Let's review what "highly unlikely" means a bit.

Looking at baseball-reference.com, the number of triples per game has historically declined, from hovering in the .4s and .5s in early history to the current rate of under .2. I'll use .35 as a rough historical average, because I don't feel like doing detailed calculations to find the historic average.

By similar reasoning, I'll declare 1.35 as the historical WHIP. It's probably not perfect, but we're just going with rough estimates here. Figuring that rare events like HBP cancel other rare-ish events like double plays, and we can then conclude 4.35 batters per half inning as reasonable for a rough calculation.

This yields 78.3 batters per game. Thus we get a triples average of about .00447. That is to say, with a typical at-bat until perfectly neutral conditions, a triple has a probability of happening 0.447% of the time. Five triples in a row happens at a rate of 1.78*10^-12. Six triples in a row happens at a rate of 7.9*10^-15. So, very unlikely...But that's for any given sequence of six batters.

I'm playing in MLB116256 right now. I'll use that as a rough estimate of the leagues that have been played. Let's further assume the standard 24-team, 162-game schedule is a perfect average for simplicity -- smaller and larger leagues cancel. So that's 12 games per timeslot, 162 timeslots for a total of 1944 games per regular season. Let's call 4, 5.5, and 5.5 average playoff series lengths, and we get an average postseason total of 32.5 games. I'll lop it to 32 to account for some shorter seasons, as larger leagues don't generally have expanded playoffs. So that's 1976 games total, on average, per league. That's 229,721,856 games played. Use the 78.3 batters per game figure, and... 17,987,221,324.8 is the number of expected at-bats. Obviously you can't have a partial at-bat but such is the nature of estimation. In scientific notation, this is about 1.8*10^11. So we're still looking at a decent-sized gap of 10^4, or about 10000 times for six triples, but only about a factor of 10 for five triples. But this is all assuming perfectly neutral conditions. If a team builds of all deadballers, plays in Petco or a similar park, and is going up against a homerun-suppressing deadball pitcher, this factor can start to shrink tremendously. I would conclude that at one point in SIM history, a five triple run might well indeed happen. Six triples is a bit harder, admittedly, but it's still possible especially if we design to hit triples.

Nice analysis. One key part that is difficult (impossible?) to quantify is that WIS owners build their rosters (and ballparks) each season to fit some strategy or theme, much moreso than owners in real-life. So you get a lot of lineups that do not have good historical analogs.

Someone mentioned getting unrealistic 57-3 game scores. Probably my biggest pet peeve in WIS is when a mop-up pitcher remains in the game to give up 31 runs, or whatever. Even if an owner maxes out the pitch count and sets 'Call Bullpen' to 1, I wish the sim would make a pitching change after something closer to reality. Say, after your mopup pitcher gives up 12 runs, your backup outfielder takes the mound.

Posted by dannyjoe on 8/4/2016 11:18:00 PM (view original):
Someone mentioned getting unrealistic 57-3 game scores. Probably my biggest pet peeve in WIS is when a mop-up pitcher remains in the game to give up 31 runs, or whatever. Even if an owner maxes out the pitch count and sets 'Call Bullpen' to 1, I wish the sim would make a pitching change after something closer to reality. Say, after your mopup pitcher gives up 12 runs, your backup outfielder takes the mound.

Yeah, that would be a great addition to the next update in 2012 that we would be allowed to have a backup outfielder pitch. It would eliminate the need for a mop-up.