Wednesday, December 11, 2013
6:00 PM - 8:00 PM EST
The LIVE portion of this chat will begin on
Wednesday, December 11, 2013 at 6:00 PM EST.
Another chat to discuss what you like and dislike about the new changes to 3.0. I'd also like to hear from anyone on possible enhancements that they would like to see for the 3.1 update.
Hello everyone. Thanks for joining me for another episode of GD update 3.0. Hope everyone has been having a great holiday season. Let's go ahead and get underway here. I'll start by answering some of the questions that have already been asked over the last couple of days. ( Moderator - 6:00 PM)
It seems that upperclassmen outperform underclassmen of similar attributes. Is there some kind of seniority factor like in previous versions? (slid64er - Hall of Famer - 6:01 PM)
Within the engine there is not a concept of seniority. Having said that, the senior players probably have a better handle on the formations that are being played, given the years of practice.
In the last Dev Chat you mentioned that formation IQ is used in calculating interceptions. I've noticed anecdotally that a low formation IQ for QB's is disproportionately related to int's. QB's with high form-IQ throw almost none, while those with low form-IQ throw a lot. I'm just throwing the observation out there. Any thoughts as the lead developer? (mcbethbr - Hall of Famer - 6:04 PM)
That's interesting. At first glance, I think that Formation IQ should have a significant role in the interceptions. However, the results might be a little extreme. This would be a good test to run to make sure those numbers are not inappropriate. Thanks for the observations.
Sack seem to be out of control as are the lack of interceptions....any chance of correcting these issues? (kevthekidd - Hall of Famer - 6:07 PM)
Sacks are dependent on the play call. The long or deep passes are more likely to get sacks. The numbers from my test show that on a balanced plan the sack rate lines up pretty well with what occurs in real life. They are a bit high, so I'll pull that number back in the next update, but it won't be significant. The interception rate is lower that I expected. I put out a small change to the interception rate in last night's update, but I will have to confirm the numbers before I make a bigger change.
How is tackling a core attribute for a DB? What are the main and secondary attributes for all positions. (zsap - Pro - 6:09 PM)
Tackling is very critical for a DB. They are influential on outside runs as well as any play after a catch. The other attributes are listed in the knowledge base.
If the "block" function is unselected on the formation do Tight Ends/Full Backs block on running plays? Same question could be asked of Wide Receivers on any given play. (noah23 - Hall of Famer - 6:11 PM)
Yes. Tight Ends, Full back, and WR will all block. The difference is where their blocking influence will occur. If they are set "isBlocking" they will be added to the offensive line and influence very short and short runs in the middle and to the outside. If they are not blocking they will have blocking influence based on there route specification. For example, if a TE has a 50% chance of running a medium route, a 50% chance of running a short route his influence will be split between those two locations.
I have noticed that in the past few seasons that the quality of the players have gone down when RCing. (I am in AI in WILK) They have been dumb down a bit. Now don't get me wrong, I know that there will be some players ready to play and some will take a couple of seasons to be their best. I use to see several top quality TEs but it seems that they are scarce, seeing may be one or two now if I am lucky. The QBs too. Seems AI in now what AII is or was. Is there any rime or reason for this? (fermor332002 - Hall of Famer - 6:12 PM)
In the latest update of the engine we did not change the recruiting. The quality of the players is the same in this version as it was in the previous versions.
As coach of Div II MSUM I am trying to understand your ranking system. Charleston and we both have 8-2 overall records. We have a 5-0 conference record and they are 4-1, Their loss was to us 58-17. Their SOS is 40 and ours is 38. They are ranked #18 and we are #19. What am I missing? (crunkrhinos - All-Star - 6:13 PM)
The ranking system has not changed in the latest update. If you would like more information about your specific case please feel free to add a support ticket and I'll have someone answer this in a timely manner.
If nothing else is done to fix bugs, etc. At a MINIMUM, the developers need to explain to us how the engine works, so that we all have a basic understanding of how the various manipulations actually work. 1. Anatomy of a running play 2. Anatomy of a passing play (harriswb3 - Hall of Famer - 6:17 PM)
Actually lots of work has gone into fixing bugs, many of which were released last night, and more updates to the system to come in the following weeks. As for the anatomy pages, I apologize for not doing a better job of creating the necessary documentation on this. I know it's an excuse but I've been pretty deep in fixes until now. Once I work out fatigue and clear up another couple of questions I'll start creating the documents necessary to teach you how the plays work.
Since the attributes that are important to players have changed - especially a reduction in direct influence of TECH, and (for example) the increase of TKL to DB, are there any plans to adjust the distribution of attributes for recruits in the future? (bhazlewood - Hall of Famer - 6:20 PM)
Right now there are no concrete plans on adjusting the recruiting world. We have discussed some various modifications that we can make to the recruiting system, and that includes creation of recruits.
When can we expect a fix to the (lack of) fatigue issue? It seems to me that this fix would solve the quite a few of the game results/statistics related issues we're seeing. (examinerebb - Hall of Famer - 6:22 PM)
Unfortunately over the last couple of weeks I was working my way through some rather tricky issues with bugs. I've tracked down and fixed many of the critical bugs and it has freed up my time to finish investigating fatigue. I'm building a test suite now and hope to have a new model out by late next week.
I would like to be able to hold onto a scholarship or two some years. This happens often in real life where a coach does not use all his scholarships. Any chance for a choice of not having the computer fill out all 50 spots? (david_fla - All-Star - 6:26 PM)
I agree, I've played a few seasons now where I have had that same thought. That's a great suggestion. I'll add that into the mix of possible quick changes for 3.1.
So, if I have my offensive settings set to "run the clock" with a lead of 4 points or more with 4 minutes left, why would my team call 3 timeouts on offense in the final 2 minutes of the first half with an 8 point lead? (mattymacd - Hall of Famer - 6:29 PM)
I've recently updated the Timeout logic. It went out in yesterday's release. If you are still seeing some anomalies in the next games please add a support ticket with the details and I'll investigate.
Why is it that the OL never keeps pressure off of the QB? Will this be addressed? (slid64er - Hall of Famer - 6:33 PM)
This is a Play by Play oddity that will be addressed. The pressure on the QB is determined by time as well. So even if the OL has done a good job of protecting the QB the Play by play still says that there is pressure. I hope to address this soon. Though in addition, I have been wondering if the check down process is too conservative, so I'll be running some tests over the next few weeks for that as well.
Are there plans to adjust the default settings so RBs aren't catching an unrealistic number of passes? (Coaching/formations/offensive) (sactowngiant - All-Star - 6:37 PM)
Last night we made changes to the defaults used to target the players in the short and very short scenarios. This will have a big impact on those RB numbers.
On an inside running play, who is involved in blocking at the LOS for the offense and defense? Who is involved on an outside running play? (slid64er - Hall of Famer - 6:44 PM)
This varies as the play goes because the location will change, but for demonstration purposes here is an example. Note: the partial value depends on position and current location of the play. This information is available on the formations page in the blocking section
OFFENSE (using IFormation as example)
INSIDE RUNNING PLAY: OL2, OL3, OL4. Partial influence: RB (if blocking)
OUTSIDE RUNNING PLAY: OL1, OL5, TE (if blocking). Partial influence: RB (if blocking)
DEFENSE (using 4-3 as example)
INSIDE RUNNING PLAY: DL2, DL3. Partial Influence: DL1, DL4, LB1, LB2, LB3, SS, FS
OUTSIDE RUNNING PLAY: CB1, CB2 Partial Influence: DL1, DL4, LB1, LB2, LB3, SS, FS
In the beta, at least near the beginning of it, there seemed to be a real differentiation between what running back did well outside and the running backs who did well inside. Is this still the case. A clear difference? (noah23 - Hall of Famer - 6:49 PM)
yes, the attributes for the ball carrier are weighted differently depending on the location. If you are running inside the line the game instinct plays a large role and elusiveness plays a role on avoiding tackles, while strength and athleticsm play a role on breaking the tackles. On the outside elusiveness and speed play larger roles in tackling avoidance with str/ath still being the factors in breaking tackles.
Currently, DBs cover all depths and the entire width of the field on the same play. Will this be addressed? (slid64er - Hall of Famer - 6:52 PM)
Yes, the Play by Play is showing that each DB can cover the entire field. This is partially true in that the DB has an influence in each target zone, but the Play by Play will soon use the most influential at that location when fixed.
10. Explain these 'buckets" of talent a little more so that we might figure out which "bucket" our team fits into...AND, is it even possible for a low talent team to upset a high talent team? 11. Fatigue must be fixed. 12. I want to DESIGNATE my starters, just like in 2.0 (harriswb3 - Hall of Famer - 6:59 PM)
10. The "buckets" are relative to the team you are playing. The game is broken up into matchups units (OL vs DL, WR vs DB, etc), and the advantage is calculated based on the weighted attributes used in the specific situation. Each bucket is different, but an average rating difference will change the advantage at about every 3 points. So a WR with a 50 rating will likely to have a advantage over a 47 DB. And the larger the difference, the more likely an advantage will be gained. With this setup it is possible though not likely that a team with an average talent level of 6 points or lower to win. 11. Fatigue is my next fix. 12. Starters had a different meaning in the old engine, and with the next update we'll be eliminating the value of the starters tag. Also, starters are currently set up using your depth charts.
As a follow up to my question on running backs. Is it possible to make it obvious which of the two sets of abilities is helping a running back, maybe only a few plays a game, but something that tells us about it? (noah23 - Hall of Famer - 7:01 PM)
Great feedback. I'll soon be investigating the Play by Play, and besides creating a little bit more diversity it would be nice to add more useful information. I can try to make this a little more understandable so coaches can use that information.
It use to be that players would advance in practice after they played EXgames. Now they advance very little if at all during these games. Why has this been taken away or dumb down so much? (fermor332002 - Hall of Famer - 7:02 PM)
In this update we did not change anything in the practice either. It currently acts as it did before.
How can you tell if an opponent's LBs are at Line or Cover? (slid64er - Hall of Famer - 7:06 PM)
Good question. Currently there is no way to tell. I can see if there is an easy and concise way to show this in the PBP when I start investigating it.
The only real value that I see using a "starter" tag is its value in recruiting and making that promise. How will your next update affect that "promise"? (harriswb3 - Hall of Famer - 7:09 PM)
Currently the starter tag is still used for recruiting, the starters are calculated from the plays played and the depth charts. In the 3.1 release we are investigating ways to better improve this. The most probable way to go about this is to get rid of that promise in recruiting. It's value is similar to playing time for the recruit so eliminating it would not impact any decisions of a recruit.
How do we know how much better/worse a team needs to be in order to move into another bucket? For that matter, what are the cores used in determining? How are additional players at the line counted towards the calculation? i.e. TE set to bock on a running play, LB set to line on a passing play, etc. (slid64er - Hall of Famer - 7:12 PM)
The bucket system is an internal tool to determine advantage. There is no set way to move up to another bucket. The buckets are calculated each section of each play which means that the values adjust with each step in each play. The attributes used to calculate these depend on the matchup, the play, the location, fatigue, etc. Additional players are counted two ways. One is that the number of players on the line affects pressure, and their attributes are weighted and averaged depending on the play and the location.
It seems that guessing run/pass correctly is still the biggest factor in defensive success. Will this be corrected? Will we ever have the ability to be anything besides Always Run or Always Pass? (slid64er - Hall of Famer - 7:15 PM)
From the tests that I have run the main factor in defensive success is player attributes. In addition the difference between guessing pass/run has been greatly minimized. There are some slight advantages to guessing correctly, but it is no longer a major factor. We have discussed the possibility of adding a variation on the always run/always pass scenario, but I don't have any further input on when that might occur.
What's the difference between a 5-2 and a 4-3 with an LB set to line? What changes vs the run and vs the pass? (slid64er - Hall of Famer - 7:16 PM)
There is no difference except for the personnel on the field. An extra OL vs extra LB. This could matter for different plays. For example a 3rd LB will likely have more speed that will be more beneficial against an outside run. The passing will stay the same.
Is there a difference between a LB set to line and a LB who blitzes? It seems so based on the small sample I have so far. I've been burned on blitz plays, but not nearly as often when the same person is set to line. (noah23 - Hall of Famer - 7:20 PM)
No difference. The only difference is that if a LB would have been in coverage and is set to blitz he will now be added to the line. Just sample size problem that you are seeing.
Do LBs that are set to LINE, drop into coverage on a pass play? (harriswb3 - Hall of Famer - 7:21 PM)
No. they are part of the offensive line at that point.
Can you give us an update on your work on how player fatigue is affecting the game? (mcbethbr - Hall of Famer - 7:24 PM)
Fatigue is the next pressing issue that I'll be addressing. I've played with the numbers a bit and have a method for going forward, but I'm in the process of creating a testing suite that will make sure the new update will not have an adverse affect on other aspects of the game. I've scheduled a week to verify this, and hope to have an update next week.
I've noticed Sim AI teams winning much more often and winning against superior opponents. I think this is problematic for the game because nobody likes being upset by a Sim. I am against shoe-horning results, but if you are going to "throttle" anything, this should be it. You mentioned in the last chat that "The current system is set up in slices…". Maybe an automatic extra "slice" advantage for the human team playing a Sim could be built in to reduce the current number of Sim upsets. Any thoughts? (mcbethbr - Hall of Famer - 7:29 PM)
Interesting. A while ago I toyed with the idea of creating a human influence on the games. I ultimately decided against this because of the information that was available to me in the engine. The SIM AI teams at this point has no inherent advantage or disadvantage. I agree though, the game is better functioning if humans are playing humans. Perhaps as time becomes available I can investigate this further. Thanks for the input.
What factors cause penalties and what comparative factors cause me to commit more penalties than my opponent? (moreron - All-Star - 7:33 PM)
There could be a couple different reasons that you might be committing more penalties than your opponents. First, it could just be bad luck. One season is a pretty small sample size for something that doesn't really occur that often. Second, the variation in penalties called comes mostly from formation IQ. If you are running a formation that your players haven't practiced as much, they are more likely to get called for a penalty.
If I understand correctly, TECH is no longer a primary attribute of either WRs or DBs (especially Cornerbacks). Can you confirm that? Also please explain what effect TECH has regarding WR/DB matchups. (harriswb3 - Hall of Famer - 7:36 PM)
It's not the primary attribute used, but it is still used in coverage and for consistency, which determines the variation of results. More consistency means more likelihood to get a better result. If technique of a receiver is better than the technique of a db then the result of the decision is more skewed toward the receiver.
Does the hands attribute have any affect on the OL and DL (fla_dave - Hall of Famer - 7:37 PM)
No, hands do not affect any of the lines.
You have addressed this to some degree earlier in the chat, but it appears that its almost impossible to throw DEEP now. Can you explain the checkdown process (using whatever assumptions you need to make)? (harriswb3 - Hall of Famer - 7:43 PM)
The checkdown process is not what is causing the lack of deep throws. The tests that I run simulate the same play 1000 times. The way that I calculated the completion percentage was to create an average yards per play. For the most part if you were to run a game against an equal opponent and pick just one type of play (deep pass, run outside, medium pass) they would all come out to about the same average yards per play. The difference being that for deep passes interceptions increase and sacks increase. The completion percentage decreases and the yards per reception is increased.
Will pbp be expanded to show what is happening along the lines? i.e. who is blocking well on a play, who is defeating the block and creating pocket pressure, etc. (slid64er - Hall of Famer - 7:45 PM)
It is in my schedule to make the Play By Play more descriptive, unfortunately in the near term there are no stats about who is blocking well and who is not. My colleagues and I have discussed implementing better stats for offensive and defensive lineman. This has been added to the list of possible upgrades for 3.1, but I'm not sure what the end list will be for that update.
On a pass play, if the quarterback passes on his first option, what determines his second option? (slid64er - Hall of Famer - 7:49 PM)
Each new step of the play will calculate the targeted area. For basic uses, the targeted area will always chose an area based on the weighted distribution on the play. For example, if the play calls for a 50% chance of long and a 50% chance of deep each time a new target is calculated it will choose long half the time and deep half the time. There are other factors in this as well depending on the other things going on in the play. For instance, a QB cannot target a long or deep route early in the play. If the defense gets pressure on the QB the target will depend on the QB's ability to assess to the best target area.
Any possibility of being able to control the clock during the 1st and 3rd quarters? Just like we can during the 2nd and 4th quarters? I'd like to be able to "shorten" the game sometimes against specific opponents by "running the clock" the entire game...vs just 2 & 4 quarters. (harriswb3 - Hall of Famer - 7:50 PM)
This is another quick concept that might be doable for 3.1. Thanks, I'll add it to the list of things to investigate.
How much does home field help? (fla_dave - Hall of Famer - 7:51 PM)
There is not real way to quantify it, but home field offers a slight advantage.
You mentioned that you test plays 1000x. That's about a seasons worth of plays for D1A. Shouldn't you be using a much smaller test size? It may look good a 1000x, but in the course of a game, it may not mean anything. This would give the impression of having no cause/effect. Thoughts? (slid64er - Hall of Famer - 7:54 PM)
You are correct. I actually have two test suites, one that will do an aggregate across a large number and another that is geared toward a single season. The second one I run at 15 games and look for averages and standard deviations across those games. I tweak the numbers across both of those sets so that the results seem to be inline.
so over time the gameplans should follow your input even with checkdowns existing? (noah23 - Hall of Famer - 7:55 PM)
Yes. Over time the distribution should still line up with your existing game plan even through the checkdown processing.
If GI is not a core for DBs, why is it a core for other defensive positions? Is GI only used heavily in the running game? (slid64er - Hall of Famer - 7:58 PM)
GI and ATH are both used coverage for CBs. The reason it is not given more value in our system is because we estimated that besides pass coverage many of the DBs were also heavily involved in the tackling of most plays. The attributes are difficult to simply say "here is what is used the majority of the time." It depends much on your game plan and that of your opponent so the core values are the ones that we estimate are used more often.
All kinds of fixes are promised for 3.1. When will this targeted for release? What will be done in the meantime? (slid64er - Hall of Famer - 8:00 PM)
We are finishing up the last remaining list of things to do in 3.0 before we come up with our schedule for 3.1. Also the fixes for 3.1 are not promised, we have not come up with final set of things to do. We will sit down with the suggestions and prioritize the ones that we can get to in a timely manner that we feel will make the game better.
Am I correct that all pass defenses play zone/? If that is true is there any possibility in the future to give us an option on playing man to man instead of zone? I would like the ability to try to neutralize the other teams best pass threats (runnrun - Hall of Famer - 8:02 PM)
Correct, the way this works is that defenses play zones. We have toyed with allowing man to man, but have not come up with a reliable way to add this to the engine. Unfortunately as cool as this would be I don't think it will get into the game any time soon. Thanks for the idea though.
I'm out of time ladies and gentlemen, but thanks for listening to me. I hope I got to the major questions. I know there were a few that I didn't get to. If you have bugs or concerns please feel free to create a support ticket and if you have other ideas please post to the forums. Thanks everyone. ( Moderator - 8:04 PM)
Thanks everyone for all of the questions. Hopefully I answered all of the burning questions that you had.