OOTP Developments Forums - View Single Post

Buane · 08-23-2016, 01:55 PM

Howdy all. The following is a bit of a lengthy analysis of some defensive metrics and game tendencies in my OOTP league. I've tried to present everything as clearly and concisely as possible, even when those two endeavors have been in opposition. I apologize in advance if any of the below is not clear, or if any of the below is not concise.

I run a longtime OOTP online league - we've been in business since the OOTP9 days, and we're about to close the books on our first season running on OOTP17. Looking over our league numbers, I've been following a trend that's been obvious from the season's first few weeks - there has been a pretty sizable spike in plays involving CFers across the entire league.

Every individual CFer I looked at was making way, way more plays than they had in previous seasons. Even bad defenders with low range who were pressed into starting or temporary CF duty were getting to far more fly balls than previous numbers suggested they should or did. Stated plainly, centerfielders in OOTP17 are getting more balls hit to them than ever before in any version of OOTP.

So naturally I had a few questions. Was this a conscious decision? Did the new model better represent the balls-in-play distribution from the real MLB? If CFers were making more outs this season, what position was making fewer? Was this just somehow a setting or a change in my league, or was it universal?

Some of these questions I've answered myself, and for others I turn to the ootpdev forums.

Let me get an important caveat out of the way first: My league converted from OOTP15 to OOTP17, so we skipped OOTP16. This change COULD have been introduced in OOTP16 but I wouldn't have seen it. For what it's worth, I would guess it's an OOTP17 change though, otherwise it probably would have come to light before now.

Ok, on to the numbers. As you probably know we don't (yet) have wonderfully complex defensive metrics in OOTP. No statcast numbers telling us how much ground was covered, what a guy's first step was, his route efficiency, etc. That means I have a choice of some VERY imperfect statistics for this analysis. Primarily, I am using Total Chances. Believe me when I say I understand all the limitations of using Total Chances as a defensive stat and/or metric. Thankfully, for the main part of this analysis, Total Chances works as a point-prover since it is only being used as a year-over-year comparative, and can be seen to be very steady from year-to-year in both OOTP and the real MLB.

Ok, actually on to the numbers now. The current year in my league is 2045, so all the significant comparisons are going to be made against that year. For the 2044 season, our previous year, Center Fielders logged 34775 defensive innings, and accumulated 9292 total chances. This is a rate of 2.40 chances per 27 outs, or 2.4 chances per 9-inning game. So far so simple. In real-life MLB the average chances/27 for a CF is about 2.6, so the CF numbers from our LAST season seem pretty in-line with the real world.

However THIS season, our CFers logged a chance rate of 3.53 per 27 outs, almost a 50% jump from the previous season. I dug far into my league's history to see if maybe Total Chances was just subject to these kind of crazy swings. The answer was an emphatic "no":

These numbers show it clearly: there has never been a time in our league when so many balls were being hit to CF.

Now, you may notice the IP totals for this year are way up as well, compared to previous years. In the interest of full disclosure, this year we underwent some league expansion, giving the league additional teams and additional innings. Believe me, my first reaction was the same as your initial reaction is now: that expansion probably had something to do with this change.

So I dug a bit deeper into this year's numbers. League ERA, batting average, BABIP, every significant stat from this season was basically right in line with our previous seasons. So could expansion really have an impact on how many balls get hit to CF and nothing else? Even though that sounds unlikely, I still wanted to disprove it, so I loaded up my league backup from our initial import to OOTP17 - prior to the expansion - and ran a full season without adding any teams. Same league settings, same players, just with the same number of teams as last year. Here are the numbers I got from that alternate season:

So even with expansion completely removed from the equation, the chance rate was still extraordinarily high.

Ok, so what about real MLB? What about the CF chance rates there historically?

While the numbers from my league don't match up perfectly with the MLB numbers (and I'm not expecting them too of course - plenty of league factors can account for these slight differences), you can see the MLB numbers are as steady as my RSL numbers were until this year. As flawed as Total Chances may be as an evaluative metric, this is clearly a number that doesn't fluctuate very much from one season to the next.

If you have a league that has spanned OOTP versions, you can check these numbers yourself. Just export the players_career_fielding_stats.csv file and interrogate the year/league/position totals in excel. Since I changed literally none of my league's settings from last season to this season, I would imagine you will see a similar jump in your leagues as well. I would be very curious to see the numbers from some other leagues out there.

So, that concludes part one of my analysis. I think this reasonably proves there is a change in OOTP17 (or 16) that affected balls in play. What about the other positions on the field though? If more outs are being converted by CFers now, what position is converting fewer? I'm glad you asked! I did this same analysis for the rest of the positions on the field.

Let's start with the rest of the outfield. Here are the LF/RF numbers:

You can see a slight uptick in our LF and RF chances, meaning OF chances are up across the board this year (though obviously not nearly as much as they are in CF). This is a slight change, but not too significant.

Ok, here are the infield positions minus 1B (more on 1B later...):

So my first big surprise was seen here. Despite the CF numbers being up significantly, the IF numbers haven't dropped that much. I was expecting more balls to the outfield to equal fewer balls on the infield. Now, there's a slight drop off at 2B/SS/3B from where we were in previous seasons, but not a ton. And certainly not enough to make up the difference.

At this point I got to thinking that since infield outs weren't down very much, maybe strikeouts had dropped a lot? In theory, fewer strikeouts could potentially cause more balls hit to CF? I guess. Here are strikeouts league-wide, as well as pitcher defensive plays for the hell of it.

Pitcher plays remain steady, and strikeout numbers ticked down very, very slightly - this is likely the only effect of the expansion I mentioned earlier. But you can see there's no big change at all. Strikeouts are a dead end.

That leaves us with two remaining positions: catcher and first base. Unfortunately for us...catcher and first base are the two defensive positions where Total Chances is a real mess of a metric. Catcher Total Chances includes pitcher strikeouts, and 1B Total Chances includes, well, every ball another infielder throws to them. So we need to use a bit of creative license to get our comparative numbers.

For catchers, we really just need to remove strikeouts from their Chances totals. That gives us numbers that look like this:

So, no real change there, either. Leaving us with the real 1B conundrum.

See, Total Chances for first basemen contains a ton of noise, because they're involved in so many plays on balls NOT hit to them. There is noise in the Total Chances number for all positions, true (for example, on a ball hit to SS where the fielder flips to 2B for the force, both players get credit for the "chance") but I have been comfortable up to this point ignoring that noise. For the other infield positions, the vast majority of their chances do not involve one and other. But that is not true for 1B - the vast, vast, vast majority of plays a 1B is involved in are not on balls hit to them, but rather thrown to them.

So to get a workable Total Chances number, I've decided to subtract 3B, 2B, and SS assists from the 1B total chances number. Now, obviously, some 3B/SS/3B assists are not on throws to 1B. Obviously, this is NOT an accurate representation of the actual number of balls hit to first base, but it will serve us as a decent rough estimate. Again, what is more important in this study is how these totals have changed from year to year - indicating a change in the balls-in-play engine - and not necessarily with determining the precise number of balls hit to a fielder. For that, we would need a much more surgical way of extracting balls-in-play data.

So, with all that said, here's the 1B totals:

Whoops. Looks like we found those missing outs. This year, there are almost NO balls being hit to first base at all. Or, at least, no balls that first basemen are making plays on. Compared to previous seasons, when our numbers were very much similar to those of the real world, this seems to be a huge change, and almost singlehandedly makes up the difference in the spike of CF plays.

So to sum up:

In OOTP17, at least compared to OOTP15, there is a rather sizable change in the number of balls hit to & played by first basemen, and balls hit to & played by center fielders. This would appear to be either a conscious design change, or a mistake/bug.

If this was a conscious design change, then the question is obvious. Why change something when the previous results were very much in line with the real world numbers?

If this was a mistake or a bug, then let's fix it!

I'm happy to provide any more data anybody would be interested in, and would be very interested to see some of the totals people are seeing in their leagues! While I think my data is comprehensive, in the end, this is only a study of one league...

08-23-2016, 01:55 PM	#1
Buane Minors (Double A) Join Date: Mar 2006 Posts: 164	Big Spike in CF Plays in OOTP17 Howdy all. The following is a bit of a lengthy analysis of some defensive metrics and game tendencies in my OOTP league. I've tried to present everything as clearly and concisely as possible, even when those two endeavors have been in opposition. I apologize in advance if any of the below is not clear, or if any of the below is not concise. I run a longtime OOTP online league - we've been in business since the OOTP9 days, and we're about to close the books on our first season running on OOTP17. Looking over our league numbers, I've been following a trend that's been obvious from the season's first few weeks - there has been a pretty sizable spike in plays involving CFers across the entire league. Every individual CFer I looked at was making way, way more plays than they had in previous seasons. Even bad defenders with low range who were pressed into starting or temporary CF duty were getting to far more fly balls than previous numbers suggested they should or did. Stated plainly, centerfielders in OOTP17 are getting more balls hit to them than ever before in any version of OOTP. So naturally I had a few questions. Was this a conscious decision? Did the new model better represent the balls-in-play distribution from the real MLB? If CFers were making more outs this season, what position was making fewer? Was this just somehow a setting or a change in my league, or was it universal? Some of these questions I've answered myself, and for others I turn to the ootpdev forums. Let me get an important caveat out of the way first: My league converted from OOTP15 to OOTP17, so we skipped OOTP16. This change COULD have been introduced in OOTP16 but I wouldn't have seen it. For what it's worth, I would guess it's an OOTP17 change though, otherwise it probably would have come to light before now. Ok, on to the numbers. As you probably know we don't (yet) have wonderfully complex defensive metrics in OOTP. No statcast numbers telling us how much ground was covered, what a guy's first step was, his route efficiency, etc. That means I have a choice of some VERY imperfect statistics for this analysis. Primarily, I am using Total Chances. Believe me when I say I understand all the limitations of using Total Chances as a defensive stat and/or metric. Thankfully, for the main part of this analysis, Total Chances works as a point-prover since it is only being used as a year-over-year comparative, and can be seen to be very steady from year-to-year in both OOTP and the real MLB. Ok, actually on to the numbers now. The current year in my league is 2045, so all the significant comparisons are going to be made against that year. For the 2044 season, our previous year, Center Fielders logged 34775 defensive innings, and accumulated 9292 total chances. This is a rate of 2.40 chances per 27 outs, or 2.4 chances per 9-inning game. So far so simple. In real-life MLB the average chances/27 for a CF is about 2.6, so the CF numbers from our LAST season seem pretty in-line with the real world. However THIS season, our CFers logged a chance rate of 3.53 per 27 outs, almost a 50% jump from the previous season. I dug far into my league's history to see if maybe Total Chances was just subject to these kind of crazy swings. The answer was an emphatic "no": These numbers show it clearly: there has never been a time in our league when so many balls were being hit to CF. Now, you may notice the IP totals for this year are way up as well, compared to previous years. In the interest of full disclosure, this year we underwent some league expansion, giving the league additional teams and additional innings. Believe me, my first reaction was the same as your initial reaction is now: that expansion probably had something to do with this change. So I dug a bit deeper into this year's numbers. League ERA, batting average, BABIP, every significant stat from this season was basically right in line with our previous seasons. So could expansion really have an impact on how many balls get hit to CF and nothing else? Even though that sounds unlikely, I still wanted to disprove it, so I loaded up my league backup from our initial import to OOTP17 - prior to the expansion - and ran a full season without adding any teams. Same league settings, same players, just with the same number of teams as last year. Here are the numbers I got from that alternate season: So even with expansion completely removed from the equation, the chance rate was still extraordinarily high. Ok, so what about real MLB? What about the CF chance rates there historically? While the numbers from my league don't match up perfectly with the MLB numbers (and I'm not expecting them too of course - plenty of league factors can account for these slight differences), you can see the MLB numbers are as steady as my RSL numbers were until this year. As flawed as Total Chances may be as an evaluative metric, this is clearly a number that doesn't fluctuate very much from one season to the next. If you have a league that has spanned OOTP versions, you can check these numbers yourself. Just export the players_career_fielding_stats.csv file and interrogate the year/league/position totals in excel. Since I changed literally none of my league's settings from last season to this season, I would imagine you will see a similar jump in your leagues as well. I would be very curious to see the numbers from some other leagues out there. So, that concludes part one of my analysis. I think this reasonably proves there is a change in OOTP17 (or 16) that affected balls in play. What about the other positions on the field though? If more outs are being converted by CFers now, what position is converting fewer? I'm glad you asked! I did this same analysis for the rest of the positions on the field. Let's start with the rest of the outfield. Here are the LF/RF numbers: You can see a slight uptick in our LF and RF chances, meaning OF chances are up across the board this year (though obviously not nearly as much as they are in CF). This is a slight change, but not too significant. Ok, here are the infield positions minus 1B (more on 1B later...): So my first big surprise was seen here. Despite the CF numbers being up significantly, the IF numbers haven't dropped that much. I was expecting more balls to the outfield to equal fewer balls on the infield. Now, there's a slight drop off at 2B/SS/3B from where we were in previous seasons, but not a ton. And certainly not enough to make up the difference. At this point I got to thinking that since infield outs weren't down very much, maybe strikeouts had dropped a lot? In theory, fewer strikeouts could potentially cause more balls hit to CF? I guess. Here are strikeouts league-wide, as well as pitcher defensive plays for the hell of it. Pitcher plays remain steady, and strikeout numbers ticked down very, very slightly - this is likely the only effect of the expansion I mentioned earlier. But you can see there's no big change at all. Strikeouts are a dead end. That leaves us with two remaining positions: catcher and first base. Unfortunately for us...catcher and first base are the two defensive positions where Total Chances is a real mess of a metric. Catcher Total Chances includes pitcher strikeouts, and 1B Total Chances includes, well, every ball another infielder throws to them. So we need to use a bit of creative license to get our comparative numbers. For catchers, we really just need to remove strikeouts from their Chances totals. That gives us numbers that look like this: So, no real change there, either. Leaving us with the real 1B conundrum. See, Total Chances for first basemen contains a ton of noise, because they're involved in so many plays on balls NOT hit to them. There is noise in the Total Chances number for all positions, true (for example, on a ball hit to SS where the fielder flips to 2B for the force, both players get credit for the "chance") but I have been comfortable up to this point ignoring that noise. For the other infield positions, the vast majority of their chances do not involve one and other. But that is not true for 1B - the vast, vast, vast majority of plays a 1B is involved in are not on balls hit to them, but rather thrown to them. So to get a workable Total Chances number, I've decided to subtract 3B, 2B, and SS assists from the 1B total chances number. Now, obviously, some 3B/SS/3B assists are not on throws to 1B. Obviously, this is NOT an accurate representation of the actual number of balls hit to first base, but it will serve us as a decent rough estimate. Again, what is more important in this study is how these totals have changed from year to year - indicating a change in the balls-in-play engine - and not necessarily with determining the precise number of balls hit to a fielder. For that, we would need a much more surgical way of extracting balls-in-play data. So, with all that said, here's the 1B totals: Whoops. Looks like we found those missing outs. This year, there are almost NO balls being hit to first base at all. Or, at least, no balls that first basemen are making plays on. Compared to previous seasons, when our numbers were very much similar to those of the real world, this seems to be a huge change, and almost singlehandedly makes up the difference in the spike of CF plays. So to sum up: In OOTP17, at least compared to OOTP15, there is a rather sizable change in the number of balls hit to & played by first basemen, and balls hit to & played by center fielders. This would appear to be either a conscious design change, or a mistake/bug. If this was a conscious design change, then the question is obvious. Why change something when the previous results were very much in line with the real world numbers? If this was a mistake or a bug, then let's fix it! I'm happy to provide any more data anybody would be interested in, and would be very interested to see some of the totals people are seeing in their leagues! While I think my data is comprehensive, in the end, this is only a study of one league... __________________ Commissioner - Rising Star League Congratulations to the 2082 Champion Montreal Harfangs!