What Scouting Accuracy Setting is for You?

[PSUColonel]

Which do you find to be the most realistic/challenging/fun?

[PSUColonel]

Also which AI evaluation settings would you use of these?

Scout Grade: 55
This Year: 20
Last Year: 20
Two Years ago: 5

or

55
20
15
10

or

55
25
15
5

[Questdog]

I usually play w/o scouting (stats only), but when I do, I play "Normal". And neither of your eval settings are what I would use. The choice comes down to making this year worth more or Ratings worth more. Last year and, especially, the year before are irrelevant.

I think if the A.I. interpreted ratings well, then 100% ratings would be the best choice for a competitive A.I., though you would see odd things like All-Star starters being sent to AAA in August. But it does not interpret the ratings very well, in my opinion, neglecting defense and not taking into account the team context (park and league stats).

100% this year's stats would seem to be the best for a consistent A.I. that should make decisions as if it was playing stats only like me, but it does not work out that way in practice. The way OOTP calculates the percentages you set for evaluation is not what I would choose. Basically, there is no such thing as a 100% stats evaluation setting.

So, after much experimentation, I have not found a setting that actually works better than the default.

[PSUColonel]

which default are you talking about? MLB, or fictional? I forget what the fictional default is. Something like 30/50/15/5 if I recall correctly.

[PSUColonel]

Has any one ever considered some other odd stuff like 30/30/20/20? or 30/30/30/10?

[PSUColonel]

Originally Posted by Questdog

I usually play w/o scouting (stats only), but when I do, I play "Normal". And neither of your eval settings are what I would use. The choice comes down to making this year worth more or Ratings worth more. Last year and, especially, the year before are irrelevant.

I think if the A.I. interpreted ratings well, then 100% ratings would be the best choice for a competitive A.I., though you would see odd things like All-Star starters being sent to AAA in August. But it does not interpret the ratings very well, in my opinion, neglecting defense and not taking into account the team context (park and league stats).

100% this year's stats would seem to be the best for a consistent A.I. that should make decisions as if it was playing stats only like me, but it does not work out that way in practice. The way OOTP calculates the percentages you set for evaluation is not what I would choose. Basically, there is no such thing as a 100% stats evaluation setting.

So, after much experimentation, I have not found a setting that actually works better than the default.

I see your argument, but great players aren't benched or moved down in the rotation because they might be going through a tough patch either.

Also...what about trading?

[Mister_G]

I have it set to normal and 40/30/20/10

[PSUColonel]

From what Questdog wrote above, I am thinking about paying much more attention to the current year. Ratings are always used no matter what... as he said. Which technically means he is correct, the current year's stats are the most important. You could even argue the second most important might be equally ratings and last year.

In this case you would have something like 20/50/20/10.

I am now also wondering what 40/4015/5 could be like, or 30/40/20/10, 30/50/15/5

[PSUColonel]

Been doing some experimenting here...One thing I DO notice, is that good free agents without much in the way of MLB stats (International Players) ask for peanuts with a stat heavy AI evaluation compared to more ratings heavy. Not sure this is a good thing.

[PSUColonel]

I am going to use 50/30/15/5

This seems to be the best of both worlds. The contract demands are in line for the talent...while the AI will use players better for depth/lineups and pitching.

[catcherjul]

I was at 35/30/20/15 but found it was leading to odd trade involving former good players worth too much.

I switch it when reading your post to 45-35-15-5

What do you think about this? Its a fictional world in his 66th season that I started with ootp 15 3 years ago

[NoOne]

I've switched things up a bit this year. i now use Low accuracy and 50ratings 25/20/5. i don't care about my perceptions of what "should be," i only care about the results these settings have on the AI's decision making.

the problem with current year vs prev yr is the sample size, imo. at least the previous year is an entire year... current year might be 1game, or some minimum before it starts kicking in.... which isn't going to fix the problem much.

that's why i keep the 2nd year nearly as important. i just don't like the AI decisions when ratings are <50% of the evaluation or too much wieght is given to current year... i've tried ratings as low as 30% with a heavy current year as well as a more balanced current/previous year.

i handle my team so much better, the last thing i want to do is hamstring the AI even more... ai evaluation doesn't affect me at all because i don't look at overall/potential in any roster decisions, ever.

if they've aged or w/e, their ratings will show this.. they will be benched even if results are slightly better than expected for that player - if their is suitable competition on that team, which is the only time it matters. the point is for the cream to rise to the top and that most of the best players are playing regularly.

best players are almost always the ones rated better... a significantly stronger correclation to success than any 3yr percentage combination you can possibly create in the settings.

in real life all the best players do not play... some get lost in the shuffle, some are not understood, and reputation/experience plays too strong factor in how a player is choosen vs another. some may be ruined during development (self-destructive behaviour or poor management doesn't matter relative to this topic)

So, by adding stats and less accurate scouting you are making it more "realistic," but also handicapping AI decisions (and yours if you focus on overall/potential, as opposed to individual ratings like contact). know the actual impact of these decisions not what you feel about them. There are no wrong settings, but there can be a mismatch of what you think they do and what they will do, and only in those cases i'd argue you are using the wrong settings.

how often do you want a superior player sitting behind an inferior one?

how often do you want the AI to make bad decisions?

the more inaccuracy you add to the equation (scouting and or stats due to sample size is inaccurate info) :: the more poor decisions the AI will make. regardless of what we want, this is fact. what are the repercussions in the game? are they as you want them to be? that's all anyone should focus on.

[NoOne]

Originally Posted by catcherjul

I was at 35/30/20/15 but found it was leading to odd trade involving former good players worth too much.

I switch it when reading your post to 45-35-15-5

What do you think about this? Its a fictional world in his 66th season that I started with ootp 15 3 years ago

if the results fit what you want, then they're great! -tony the tiger

[PSUColonel]

The real issue is how to balance actual performance vs. perceived talent. It is important to do this because obviously you want to judge players by how they actually perform...that's a given. It's also very important because percieved talent DOES have a place in OOTP. It helps as a guideline for fair contract values, trades, and expectation.

After a lot of experimenting, it seems to me (and this may be unfortunate) anything with ratings under 50 creates some very odd results when it comes to lineup/rotation selection, as well as salary demands.

The idea is to find that sweet spot, where perception and expectation are present, but not on such a level where it ignores performance. I used to think last year's stats were a somewhat important part of the evaluation equation. That is utter rubbish. I have learned that current year's stats are the primary driver in creating a realistic experience.

What I have learned:

1 Ratings need to be at 50 or higher.

2 Current years stats needs to be substantially higher than last year and two years ago.

3. With the latest updates, normal scouting accuracy is very good, and presents a good fog of war challenge.

4. The trade AI is also very good now on average. The development teams has been fine tuning some things. I do feel you need a healthy perception/expectation part of the evaluation.

So,

Ratings (50) = Expectation/perception

Current Year (30) = performance which keeps players who are playing better higher up in the lineup and rotation.

Last Year (15) = a back up measure to reaffirm (or not) ability and performance, but not to an extent where the results will be too screwy. Think of this as a "tie breaker"

2 Years ago (2) = almost nothing. It is only here to throw a bone to players who have had a statistical anomaly one year, or for players who have either progressed or regressed since then.

[Questdog]

Originally Posted by NoOne

I've switched things up a bit this year. i now use Low accuracy and 50ratings 25/20/5. i don't care about my perceptions of what "should be," i only care about the results these settings have on the AI's decision making.

the problem with current year vs prev yr is the sample size, imo. at least the previous year is an entire year... current year might be 1game, or some minimum before it starts kicking in.... which isn't going to fix the problem much.

that's why i keep the 2nd year nearly as important. i just don't like the AI decisions when ratings are <50% of the evaluation or too much wieght is given to current year... i've tried ratings as low as 30% with a heavy current year as well as a more balanced current/previous year.

i handle my team so much better, the last thing i want to do is hamstring the AI even more... ai evaluation doesn't affect me at all because i don't look at overall/potential in any roster decisions, ever.

if they've aged or w/e, their ratings will show this.. they will be benched even if results are slightly better than expected for that player - if their is suitable competition on that team, which is the only time it matters. the point is for the cream to rise to the top and that most of the best players are playing regularly.

best players are almost always the ones rated better... a significantly stronger correclation to success than any 3yr percentage combination you can possibly create in the settings.

in real life all the best players do not play... some get lost in the shuffle, some are not understood, and reputation/experience plays too strong factor in how a player is choosen vs another. some may be ruined during development (self-destructive behaviour or poor management doesn't matter relative to this topic)

So, by adding stats and less accurate scouting you are making it more "realistic," but also handicapping AI decisions (and yours if you focus on overall/potential, as opposed to individual ratings like contact). know the actual impact of these decisions not what you feel about them. There are no wrong settings, but there can be a mismatch of what you think they do and what they will do, and only in those cases i'd argue you are using the wrong settings.

how often do you want a superior player sitting behind an inferior one?

how often do you want the AI to make bad decisions?

the more inaccuracy you add to the equation (scouting and or stats due to sample size is inaccurate info) :: the more poor decisions the AI will make. regardless of what we want, this is fact. what are the repercussions in the game? are they as you want them to be? that's all anyone should focus on.

Your points are valid, except for one thing: OOTP does not work like you are expecting it to. Allowing the A.I. to only evaluate players on ratings would seem to be the logical way to get it to field the absolute best team at any given moment. But this is not how it works in practice. You need to have this year's stats included in the formula and to a pretty high degree, in order to compensate for it's inadequate evaluation of the ratings.

As for this year's stats throwing a monkey wrench into the equation because of small sample size, this is not a problem in OOTP, because this year's stats mean nothing until an adequate sample size has been achieved.

I do not like this. A better solution, to me,would be to count all results from the last 3 seasons, regardless of sample size and just weigh each plate appearance in the seasons differently. For instance, this year's stats each count 5 times more than last years, which count 5 times more than the year before. If a player had 500 AB last year, then last year's results will prove to be more important on decisions until he has 100 AB this year, at which point the two seasons will be equal in importance and as the season progresses, this year's stats will continue to grow in importance.

But that is not how OOTP works. It is all or nothing. Not enough stats and they have no value regardless of the settings.

[PSUColonel]

I lol me that idea...in another thread I proposed something for scouting and accuracy. It's based on PA and the amount of resources you're willing to invest in scouting different players.

[JasonC23]

Originally Posted by PSUColonel

So,

Ratings (50) = Expectation/perception

Current Year (30) = performance which keeps players who are playing better higher up in the lineup and rotation.

Last Year (15) = a back up measure to reaffirm (or not) ability and performance, but not to an extent where the results will be too screwy. Think of this as a "tie breaker"

2 Years ago (2) = almost nothing. It is only here to throw a bone to players who have had a statistical anomaly one year, or for players who have either progressed or regressed since then.

Thanks for researching this, PSU! Just to verify, should the "2 Years ago" number be 5? If it's 2, that only adds up to 97.

[PSUColonel]

yes, 5...mistype sorry

[endgame]

I'm not disputing nor advocating anything being put forward in this thread. I would caution, however, making any specific judgments or claims about what does what or the significance of these evaluation settings. Experience will convey, no doubt, a few inclinations, but relentless and countless comparisons side-by-side against the same data and player movements are the only way to reach anything resembling a reasonable conclusion. Experimenting only conveys speculation.

For instance. when examining setting comparisons like 50/20/20/10 versus a very close 45/30/15/10, I seriously doubt it would be anything less than extremely time consuming to find significant and demonstrable differences between these two. It would take, IMHO, tedious collection of a ton of data under various circumstances to actually prove their effects. These aren't Madden-like sliders, adjusting the use of slants or average yards per carry, they're filters, altering a myriad number of the ways a player is viewed and valued in an equally large number of circumstances, not just trades or scouting, for example, but the extremely complex world of roster management; how we view and interpret our own players and the nearly mystical fashion with which the AI uses them to manage its organizations.

Play your way, experience it your way, perceive it your way, but fact-finding is quite a different path than simply sharing impressions.That's entirely different than presenting findings.

[MizzouRah]

Originally Posted by PSUColonel

I am going to use 50/30/15/5

This seems to be the best of both worlds. The contract demands are in line for the talent...while the AI will use players better for depth/lineups and pitching.

I've used these settings for quite some time now.