Posted by dahsdebater on 10/3/2018 2:58:00 PM (view original):

Posted by shoe3 on 10/3/2018 1:28:00 PM (view original):
Seems as though we have a distinct difference in thought and expectations here. Some folks want to play competitive multiplayer games. Some folks want the system to just tell us who has the better team. I suspect Mr. Debater came closest so far to naming it out loud, lots of folks just don’t want to deal with probability “when reward dollars are on the line”.

I'm confused.

How is it less "competitive" if the better teams win more often? Take it to the extremes - best team always wins, that's a competition. Completely random, that's not really a competition at all. There's no point in trying. So which one is more competitive? Which one is a better game?

“Competitive” doesn’t say anything about what the outcome is “supposed to be.” That’s a bias toward a deterministic mindset. “Competitive” only refers to the state of competition, 2 or more sides struggling to win.

Any outcome is reasonable, if the process is reasonable. I’m competitive. I like to compete. I enjoy winning, but I don’t mind losing “if the competition is fair”. Stacking the deck in the middle of a competition is fundamentally unfair.

Posted by shoe3 on 10/3/2018 12:58:00 PM (view original):

Posted by dahsdebater on 10/3/2018 12:26:00 PM (view original):
That's a completely ridiculous analogy. Massaging the odds for a single decision is easy. Massaging the outcomes of a sequence of events is NOT easy. Your best solution is to make all the shooters less close to .500 shooters so that probability will favor the better ones? That's dumb as hell. Almost nobody has complained in over a decade about the bulk shooting percentages in the game. Why break something that works to fix a different problem?

And by the way, moving a 53% shooter to a 55% shooter reduces variance by less than a tenth of a shot per game. So that really doesn't address the problem in a meaningful way. You have to make him like an 80% shooter to really make a difference. Again - dumb as hell. Your solutions all basically involve making it less of a basketball sim. The implemented solution keeps all the basketball sim that it ever had but narrows the distribution of outcomes. Which is exactly what the player base was asking for.

Speaking of dumb as hell, let’s talk about how you think it would take making a player “like an 80% shooter to really make a difference.” You come off as a bit smarter than this, so I’m sincerely perplexed.

I think the problem is, as I said in my post to you, that you are misunderstanding the reference. You are stuck on “standard deviation”, when that probably wasn’t even really the term mully meant. The discussion is on a deviation from expected outcome, ie how much a user thinks it blows when the simulation produces an upset. If there are too many upsets - and this is a developer question, not a user question, except as far as a developer feels like listening to any given group of users - you can either address the deviation (from “expected” results) by adjusting the parameters that define the probability determining the outcomes *of possessions* (better), or you can lazily engineer the game to adjust parameters on the fly, so to skew towards “expected” final scores (what we have).

In other words, if you’re consistently seeing too many “random” upsets, then it’s a problem with the design of the parameters that determine the results of the possessions. You fix that design (adjust the parameters as needed, weight the applicable attribute discrepancies more, etc), you don’t just engineer the game to skew probability as it goes along toward the favorite. Because if you do the latter, you are no longer running a simulation, you’re engineering a god mode. Maybe lots of people want to play poker in a casino that deals you loaded hands based on how your night has gone so far. Lots of people want to play FarmVille too. I’m not obligated to join them.

You still have never explained how you think you can adjust the parameters defining the outcomes of possessions better to result in fewer upsets.

Because you can't.

The fact that you can't see this indicates you are missing some of the fundamental mathematics at play here, frankly.

The standard deviation in the outcome is the standard error/standard deviation of the mean. This is what you have to reduce to reduce the number of upsets. The standard error for a sequence of statistical events is the standard deviation divided by the square root of the number of events. So in a sim like this, the standard error in shooting percentage is standard deviation per shot divided by sqrt(number of shots). There are only two parameters there you can really change that impact the likelihood of upsets. One is number of shots. You're not going to jack up the number of shots because the optics are bad. The other is standard deviation on a per shot basis. If you don't artificially shrink this, if the shooting percentage is around .5 the standard deviation is around .5. It has to be. Outcomes of individual events are 0 or 1, they are .5 away from .5. Always. Even if you do what I said and go to either 80% or 20% shooting the standard deviation only drops to .32 and you're really only removing 1/3 of the variance. If anything I underestimated.

Posted by dahsdebater on 10/3/2018 3:06:00 PM (view original):

Posted by kcsundevil on 10/3/2018 1:13:00 PM (view original):
Shoe, I respectfully suggest you reconsider the casino analogy. 50 hands of casino blackjack - a long sequence of mostly independent events (affected by shoe size) - is fundamentally different from 1 game of HD where individual plays are supposed to be affected by what happened before.

It's not a good comparison, and nobody else seems to be buying it.

I actually think blackjack is a very good analogy for what we're talking about. And the fact is that your variance in expectations in 50 hands of blackjack - or even a few hundred - is quite high. This is the allure of gambling. It's why people who don't count cards bother playing blackjack in the first place. Because there is a real chance for them to win. Not even a poor chance. They are at a fundamental, mathematical disadvantage to the house, but they can win on a regular basis because of the fact that there is no control for the variance associated with near 50/50 events except to ramp up the volume.

The argument is that many, many HD players think this same phenomenon is a negative when it comes to HD. The fact that the statistical loser can win a significant proportion of the time isn't particularly appealing. If I'm playing Blackjack I expect to lose, and if I win it's a bonus. If I'm playing poker and I'm losing I can just stay longer and in most cases eventually I'll come out ahead (or I'm at a very good table, but in that case I probably won't stay). The game length in HD is fixed and, in terms of number of possessions on a statistical scale, it's very short. This is why outcomes have such high variance.

The same problem exists in RL college basketball. I remember a study (probably 10 years ago, so maybe the numbers would be a little different now) that found that 5-15 point underdogs win almost 30% of the time IRL. So it's not like WIS wasn't reproducing reality. The question is whether it's better to reproduce reality or to make a better game. At the time of the update most users on the forums seemed to agree that it was more important to make a better game, and that a "better game" was a game in which you were rewarded more often for putting together a superior team and game plan. Nobody wants to play a game where you aren't rewarded for your input.

Lets say you consistently have a 10% better team/ are a 10% better coach, and we play each other in conference 3 times per year for 20 years. If that’s all true, I should expect to win about 45% of the time, say 27 wins out of 60 matchups. Yet you would be upset with this model, because you have the “better team”, so you should win “most of the time”, ie some percentage much higher than 55%?

Thats not a very competitive attitude. I hope you understand me when I say it just appears you don’t want to actually compete, ie struggle with someone to win a game, you just want to be told who has the better team.

Posted by dahsdebater on 10/3/2018 3:06:00 PM (view original):

Posted by kcsundevil on 10/3/2018 1:13:00 PM (view original):
Shoe, I respectfully suggest you reconsider the casino analogy. 50 hands of casino blackjack - a long sequence of mostly independent events (affected by shoe size) - is fundamentally different from 1 game of HD where individual plays are supposed to be affected by what happened before.

It's not a good comparison, and nobody else seems to be buying it.

I actually think blackjack is a very good analogy for what we're talking about. And the fact is that your variance in expectations in 50 hands of blackjack - or even a few hundred - is quite high. This is the allure of gambling. It's why people who don't count cards bother playing blackjack in the first place. Because there is a real chance for them to win. Not even a poor chance. They are at a fundamental, mathematical disadvantage to the house, but they can win on a regular basis because of the fact that there is no control for the variance associated with near 50/50 events except to ramp up the volume.

The argument is that many, many HD players think this same phenomenon is a negative when it comes to HD. The fact that the statistical loser can win a significant proportion of the time isn't particularly appealing. If I'm playing Blackjack I expect to lose, and if I win it's a bonus. If I'm playing poker and I'm losing I can just stay longer and in most cases eventually I'll come out ahead (or I'm at a very good table, but in that case I probably won't stay). The game length in HD is fixed and, in terms of number of possessions on a statistical scale, it's very short. This is why outcomes have such high variance.

The same problem exists in RL college basketball. I remember a study (probably 10 years ago, so maybe the numbers would be a little different now) that found that 5-15 point underdogs win almost 30% of the time IRL. So it's not like WIS wasn't reproducing reality. The question is whether it's better to reproduce reality or to make a better game. At the time of the update most users on the forums seemed to agree that it was more important to make a better game, and that a "better game" was a game in which you were rewarded more often for putting together a superior team and game plan. Nobody wants to play a game where you aren't rewarded for your input.

First, I was wrong. Someone else accepts shoe's analogy. Duly noted.

You're a stats guy, so you know if you play 50 straight blackjack hands without adjusting your starting wager, the outcome is predictable within a narrow range at a high degree of confidence. More so if you play a few hundred hands.

Casual players bank on hitting a hot streak and leaving the table before they regress to the mean outcome. They can do that.

In HD you're at the table for all 40 minutes. Tis a pity shoe couldn't put his team on the bus at halftime and leave the table.

10% better at what? Because 10% better team, any intelligent way you want to define it, does not mean you wind up with a 55/45 winning percentage.

Posted by dahsdebater on 10/3/2018 3:20:00 PM (view original):

Posted by shoe3 on 10/3/2018 12:58:00 PM (view original):

Posted by dahsdebater on 10/3/2018 12:26:00 PM (view original):
That's a completely ridiculous analogy. Massaging the odds for a single decision is easy. Massaging the outcomes of a sequence of events is NOT easy. Your best solution is to make all the shooters less close to .500 shooters so that probability will favor the better ones? That's dumb as hell. Almost nobody has complained in over a decade about the bulk shooting percentages in the game. Why break something that works to fix a different problem?

And by the way, moving a 53% shooter to a 55% shooter reduces variance by less than a tenth of a shot per game. So that really doesn't address the problem in a meaningful way. You have to make him like an 80% shooter to really make a difference. Again - dumb as hell. Your solutions all basically involve making it less of a basketball sim. The implemented solution keeps all the basketball sim that it ever had but narrows the distribution of outcomes. Which is exactly what the player base was asking for.

Speaking of dumb as hell, let’s talk about how you think it would take making a player “like an 80% shooter to really make a difference.” You come off as a bit smarter than this, so I’m sincerely perplexed.

I think the problem is, as I said in my post to you, that you are misunderstanding the reference. You are stuck on “standard deviation”, when that probably wasn’t even really the term mully meant. The discussion is on a deviation from expected outcome, ie how much a user thinks it blows when the simulation produces an upset. If there are too many upsets - and this is a developer question, not a user question, except as far as a developer feels like listening to any given group of users - you can either address the deviation (from “expected” results) by adjusting the parameters that define the probability determining the outcomes *of possessions* (better), or you can lazily engineer the game to adjust parameters on the fly, so to skew towards “expected” final scores (what we have).

In other words, if you’re consistently seeing too many “random” upsets, then it’s a problem with the design of the parameters that determine the results of the possessions. You fix that design (adjust the parameters as needed, weight the applicable attribute discrepancies more, etc), you don’t just engineer the game to skew probability as it goes along toward the favorite. Because if you do the latter, you are no longer running a simulation, you’re engineering a god mode. Maybe lots of people want to play poker in a casino that deals you loaded hands based on how your night has gone so far. Lots of people want to play FarmVille too. I’m not obligated to join them.

You still have never explained how you think you can adjust the parameters defining the outcomes of possessions better to result in fewer upsets.

Because you can't.

The fact that you can't see this indicates you are missing some of the fundamental mathematics at play here, frankly.

The standard deviation in the outcome is the standard error/standard deviation of the mean. This is what you have to reduce to reduce the number of upsets. The standard error for a sequence of statistical events is the standard deviation divided by the square root of the number of events. So in a sim like this, the standard error in shooting percentage is standard deviation per shot divided by sqrt(number of shots). There are only two parameters there you can really change that impact the likelihood of upsets. One is number of shots. You're not going to jack up the number of shots because the optics are bad. The other is standard deviation on a per shot basis. If you don't artificially shrink this, if the shooting percentage is around .5 the standard deviation is around .5. It has to be. Outcomes of individual events are 0 or 1, they are .5 away from .5. Always. Even if you do what I said and go to either 80% or 20% shooting the standard deviation only drops to .32 and you're really only removing 1/3 of the variance. If anything I underestimated.

I’m not a programmer. I “can’t” tell you how the parameters can be adjusted, because I don’t know how they’re established in the first place. But I can tell you broadly that of course it is entirely possible to adjust parameters of each possession in a game which is based on series of possessions determined by variables. Referring to the poll, if a guy is in a matchup against a player who is roughly 9.5% worse, it’s entirely reasonable to expect a 55% shooting percentage, given enough possessions. If those parameters result in more upsets than desired, developers can weight it to 57, or even a little higher (they should stop well short of 80%). Remember, the goal is not to manufacture a result, the goal is to credibly reproduce a competitive simulation.

Any outcome is fair and reasonable, if the process is fair and reasonable. If the process is not fair, the outcome is not reasonable.

“Tis a pity shoe couldn't put his team on the bus at halftime and leave the table.”

If the game had used the same parameters and rules in both halves, I wouldn’t have been tempted to, even if I could. In real life, if a basketball game worked like that, who would want to play?

“Hey guys, Vegas says you’re supposed to be down 3 and you’re up 11. Sorry for what’s about to happen.”

Posted by dahsdebater on 10/3/2018 3:29:00 PM (view original):
10% better at what? Because 10% better team, any intelligent way you want to define it, does not mean you wind up with a 55/45 winning percentage.

How much better do you think a 55% team is than a 45% team, assuming statistically significant games played? What number do you suggest putting on it? Just go with that, and plug that in instead, if my amateur math is a stumbling block for you.

Shoe3, would u rather see the attributes for the better team -stretched- on each possession rather than be affected by prior possessions?

Posted by ncmusician_7 on 10/3/2018 3:59:00 PM (view original):
Shoe3, would u rather see the attributes for the better team -stretched- on each possession rather than be affected by prior possessions?

*If* too many upsets is a problem in the developers mind, yes. That is the approach I think they should take. Re-weight the parameters that determine the probability for each (possession) outcome incrementally, based on how the variables - player attributes, gameplan, etc - are set to play out.

Posted by shoe3 on 10/3/2018 3:40:00 PM (view original):

Posted by dahsdebater on 10/3/2018 3:29:00 PM (view original):
10% better at what? Because 10% better team, any intelligent way you want to define it, does not mean you wind up with a 55/45 winning percentage.

How much better do you think a 55% team is than a 45% team, assuming statistically significant games played? What number do you suggest putting on it? Just go with that, and plug that in instead, if my amateur math is a stumbling block for you.

Shoe3, if u use an excel spreadsheet and have one team shoot 50% and the other team shoot 55% -no other variables- and each team takes 50 shots, the better shooting team wins about 70% of the time.

Posted by ncmusician_7 on 10/3/2018 4:10:00 PM (view original):

Posted by shoe3 on 10/3/2018 3:40:00 PM (view original):

Posted by dahsdebater on 10/3/2018 3:29:00 PM (view original):
10% better at what? Because 10% better team, any intelligent way you want to define it, does not mean you wind up with a 55/45 winning percentage.

How much better do you think a 55% team is than a 45% team, assuming statistically significant games played? What number do you suggest putting on it? Just go with that, and plug that in instead, if my amateur math is a stumbling block for you.

Shoe3, if u use an excel spreadsheet and have one team shoot 50% and the other team shoot 55% -no other variables- and each team takes 50 shots, the better shooting team wins about 70% of the time.

That’s fine, but I’m not saying anything about shooting percentage, though. Just generally 10% better, or whatever number you want to use for the difference between a team that wins 45% and a team that wins 55% in a statistically significant amount of games.

And there are presumably lots of factors that go into whether or not an HD shot is made anyway. It’s not just the shooter, it’s defender/defense, settings, distribution. This is kind of tangentially the point, too. Real players don’t have have numbers representing how good they are. So this whole idea of a “better” team, or a “better” gameplan, or a “better” simulation is all subjective anyway. The “better” team is the one that wins most often over a statistically significant period of time. Sometimes that team loses. The game should not be engineered to prevent that from happening.

Programmer here for like 40 years, and I can tell you that making it work like shoe wants would by totally complex, take massive amounts of code/CPU cycles to just SIM a single game (nevermind 10 worlds with 500ish games going). I wouldn't want to code review those algorithms either.

And just to be clear, shoe are you advocating adjust shooting %s on a per possession basis (assuming the previous ones are outliers) instead of just doing it at halftime (and only giving it a single adjustment)? How is that different?

All this because you think your team should of won that game because yes your team got lucky for the first half, got an outlier result (holding your opponent to 16 points), and they wouldn't get a similar outlier result (61 points of which only 46 were non-intentional FTs) plus you previously lost to the same opponent earlier in non-con.

A team winning 55% of the time is almost not significantly better in HD and most coaches who did win 55% of the time wouldn't say its out of line to role switch and be the 45% winner. Its the Team A should win 80% of the time and in terms of HD programming they are getting the shaft in the first half due to the variances of sim engine of only having some 40 possessions that don't fall within expectations.

Posted by buddhagamer on 10/3/2018 4:24:00 PM (view original):
Programmer here for like 40 years, and I can tell you that making it work like shoe wants would by totally complex, take massive amounts of code/CPU cycles to just SIM a single game (nevermind 10 worlds with 500ish games going). I wouldn't want to code review those algorithms either.

And just to be clear, shoe are you advocating adjust shooting %s on a per possession basis (assuming the previous ones are outliers) instead of just doing it at halftime (and only giving it a single adjustment)? How is that different?

All this because you think your team should of won that game because yes your team got lucky for the first half, got an outlier result (holding your opponent to 16 points), and they wouldn't get a similar outlier result (61 points of which only 46 were non-intentional FTs) plus you previously lost to the same opponent earlier in non-con.

1. Not buying it. Unless you’re telling us that right now the simulation consists of one single outcome (final score), and the rest is just window dressing, re-configuring the parameters that determine possession outcomes isn’t any more complex (at least it wouldn’t have to be) than adjusting the “player roles” from the user settings page.

2. No. I’m not talking about shooting percentage. I’m talking about adjusting the parameters that determine the outcome of a possession. My goal is not to produce a simulation with x number of upsets. My goal would be to produce a reasonable and fair process of simulating college basketball games.

3. Call sour grapes if you want. You’re incorrect, but it’s cool, I understand. I’ve talked about it before, this is the most recent, and most egregious example. And as I told mully, the whole point of bringing it up here is that once folks understand what is actually happening, and can spot it, they will feel “royally screwed”. Because forced or speeding up regression is fundamentally unfair. It is re-loading the deck. It is turning a coin into an unfair coin once it shows heads 3 times in a row.