Reducing accuracy ratings for all QB’s in Madden 21

[khaliib]

Quote: Originally Posted by tg88forHOF

Why so low? In 2019, 13 NFL QBs finished with a completion percentage higher than 65%. 29 finished with a completion percentage above 60%. The only starting QBs to finish with a completion percentage below 60% were Andy Dalton (59.5), Baker Mayfield (59.4) and Josh Allen (58.8). In 2018, the average numbers were even higher....19 QBs finished above 65%. Both seasons had 3 QBs finish above 70%, and not all of them are ones you'd necessarily consider elite. Brees finished above 70% both years (like you'd expect), but some of the other names are guys like Derek Carr, Ryan Tannehill, and Kirk Cousins. You have to go all the way back to 2013 to find a season where even 10 starting QBs finished with a completion percentage under 60% If you're trying to replicate a specific historical era, I'd understand the lower numbers. But trying to force the majority of the league into the 50-55% range doesn't match the modern league. Citations: 2019 passing stats 2018 passing stats

This ideology has been around for several years now.

It’s just impossible no matter how many mathematical formulas are used, for the AI to meet a static statistical metric while being dynamic/variable against multiple User play styles/difficulty level/experience.

What’s this is asking for is “Skynet 2020” to force outcomes/animations during gameplay against all gamers to meet such a metric.

This is something that would affect every Madden User vs AI gamer and someone would always have an issue, esp if it’s a QB/Team their in favor of

[tg88forHOF]

Quote: Originally Posted by khaliib

This ideology has been around for several years now. It’s just impossible no matter how many mathematical formulas are used, for the AI to meet a static statistical metric while being dynamic/variable against multiple User play styles/difficulty level/experience. What’s this is asking for is “Skynet 2020” to force outcomes/animations during gameplay against all gamers to meet such a metric. This is something that would affect every Madden User vs AI gamer and someone would always have an issue, esp if it’s a QB/Team their in favor of

Agree completely.

So many of the endless arguments around here around what is and isn't "sim" (and what "sim" even means) is driven by expectations, and results that don't fall within the range of those expectations.

And forcing results into a predetermined range is deterministic by definition. I'm not familiar with a definition of "simulation" that includes determinism....especially in this sport.

Completion percentage is where I see that phenomenon pop up most often. That's not to say "robo QB" isn't possible, but nearly every time I see somebody complain about it, they're talking about individual games at 75-80%, or seasons at 65-70%, when not only do both of those fall well within the range of current NFL norms, artificially forcing the game to NEVER produce them would be at least as unrealistic.

[tg88forHOF]

In terms of individual games, in just week 1 alone of the 2019 NFL season, 8 different QBs had a completion percentage at 75% or higher, led by the "elite" Gardner Minshew III (a backup in his first NFL action) at 88% (22/25 for 275).

I'm fairly certain a Madden game that saw a QB of Minshew's caliber completing 88% of his passes would have at least a few people muttering "Robo QB".

And that's just 1 week of last season.

[CujoMatty]

Quote: Originally Posted by tg88forHOF

Agree completely. And forcing results into a predetermined range is deterministic by definition. I'm not familiar with a definition of "simulation" that includes determinism....especially in this sport.

All sports games do this though or maybe I'm not understanding your point?

It's why NBA 2k has realistic shooting percentages. It's why MLB the Show has realistic batting average. The CPU forces results all the time thru no fault of the user.

I think I'm just not getting it.

[tg88forHOF]

Quote: Originally Posted by CujoMatty

All sports games do this though or maybe I'm not understanding your point? It's why NBA 2k has realistic shooting percentages. It's why MLB the Show has realistic batting average. The CPU forces results all the time thru no fault of the user. I think I'm just not getting it.

I think you're confusing "what I expect to see" with "realism".

In the real world, dynamic environments like sports...where you have a wide array of variables that can influence any individualplay, game, or even season (multiple individual athletes with variable levels of performance on both offense and defense, coaches and gameplans with variable degrees of effectiveness, weather, variable referee performance, etc)...should have a wide variety of individual outcomes, many of which fall well outside what we expect to see as "normal".

That's probably more true in football...arguably the ultimate team sport, the one with the highest number of players performing simultaneously, and with the largest impact from team strategies...than any other sport.

And a sports video game, where user influence adds even more variability, should make outlier results even more common.

If all you ever see are results in an expected range (that is, numbers that line up with what you expect to see going in), where outlier results are artificially eliminated and forced into the expected range, that's actually less realistic. It might feel more realistic based on your expectations (and that's ok...perception and immersion are real things), but if you eliminate all or most of those outliers, you're not actually emulating reality.

You're right, other sports games artificially limit results to fit user expectations all the time. If that's what people want, cool...there's nothing wrong with it.

But that doesn't make those artificially limited results "realistic".

Basically, reality includes outliers. The NFL includes a lot of outliers. Pointing at an outlier and saying "that number isn't realistic" is backwards.

[tg88forHOF]

Quote: Originally Posted by CujoMatty

All sports games do this though or maybe I'm not understanding your point? It's why NBA 2k has realistic shooting percentages. It's why MLB the Show has realistic batting average. The CPU forces results all the time thru no fault of the user. I think I'm just not getting it.

I guess, let me ask you this: if you were playing Madden...let's say you were simming a full game in broadcast mode...and you saw a rookie backup QB come off the bench and complete 88% of his passes, would you consider that result realistic?

Edit: It's perfectly cool if your answer is "no"...that is an outlier result, and if that kills the feeling of accuracy from the sim for you, I totally get that answer.

But the point is, that's a real thing that really happened...just last year, and from a QB who (probably) isn't the second coming of Drew Brees or Tom Brady. If the game had forced that completion percentage into a more normal range, it would be actually factually making the game less realistic.

[Madden08PCgmr]

Quote: Originally Posted by tg88forHOF

I guess, let me ask you this: if you were playing Madden...let's say you were simming a full game in broadcast mode...and you saw a rookie backup QB come off the bench and complete 88% of his passes, would you consider that result realistic?

Depends..

Was he 13 for 15 for 88 yards on a bunch of dumpoffs.

Or 13 for 15 for 288 yards and 3 tds…

… and is his name Peyton Manning?

.. some of this stuff is hard to quantify. What if I flat out suck at shooting in 2k? Will the game make sure I hit 40% regardless? I have ALWAYS struggled to hit in MLB. I can't seem to find a medium. But I know damn well the game take a few out of the yard for me here and there, when I don't make great contact.

.. I don't think we'll ever get past some of this stuff.

[tg88forHOF]

Quote: Originally Posted by Madden08PCgmr

Depends.. Was he 13 for 15 for 88 yards on a bunch of dumpoffs. Or 13 for 15 for 288 yards and 3 tds… … and is his name Peyton Manning?

Actually, he was 22/25 for 275 yards and 2 TD. That's 11 YPA, so they certainly weren't dumpoffs. And he most certainly was NOT Peyton Manning.

Now, he was playing a Chiefs defense that was objectively terrible in all phases in Week 1 of 2019. The pass defense talent was essentially the same as the defense that played well down the stretch and helped win a Super Bowl, but those guys were still adjusting to Spagnuolo's scheme, and were...not good at football yet. In Madden terms, let's say their ratings were solid, but their morale was terrible.

But that's my point...there are so many variables in real-life football that we should expect to see a lot of weird, outlier results in a game that accurately models the NFL. If they're all shoved into a predetermined range, realism is decreased, not enhanced.

I do hear you on the "going back and forth" bit...I see completion after completion from an average (at best) QB and sometimes get frustrated myself. Like I said, expectations are a real thing, and if they impact somebody's immersion and perception of "real", I'm not going to argue that effect.

It's just sometimes useful to take a step back and think about what kind of "real" we want to see.