|
MLB 08 Programmer
OVR: 6
Join Date: Mar 2008
Posts: 168
|
Re: AI Follow-up Questions...ie The Boys are back in town....
Q&A Part II
Hi this is Brian, and I just created an account to make things a bit easier. Here are some more answers to AI questions for now. I'll answer more until the end of the week after I finish some other tasks. I need to get to those so I'll focus on completely new Qs as time permits. There are going to be several slider Q&As, but I want to emphasize that the game is best enjoyed by most people using default settings and a reasonable difficultly level. Every user is different and has different goals, and we leave the choice up to the user. If you're sure about using sliders, I'd give default settings at least 10 games if not 20 for the reasons listed in the original post.
If there's many posts about non-AI topics, we could have a separate thread for it so we can better adress the Q&A on AI. I'm not really qualified to answer non-AI questions and I'd prefer to only give accurate answers.
Two typo clarifications in CAPS:
Typo 1: "..you can guess both type AND zone"
Typo 2: "You have to swing a little earlier on inside pitches..to put them into play WELL". The dilemma is inside pitches tend to hit inside given matching timing, so at an extreme you'll only be able to hit a very hard inside pitch foul. This is mainly a factor for pitches off the strike zone.
Q: Pitcher stamina slider doesn't do much, if anything. Pitcher control doesn't seem to do anything either. (Garrett67)
A: There are two stamina sliders, one for the starting pitcher and one for relievers. A typical starter will change +/- 12 pitches for reaching 0 energy (though 1/4 is a more typical baseline for subs). Certain pitch types take more energy to throw (each game will be slightly different) but pitch count alone will give you a good idea. If you want to reduce control and keep the game somewhat balanced around that, I'd suggest the pitch count slider.
Q: Pitch speed slider - we know it speeds up the percieved speed, and doesn't affect the radar gun. But does it affect the path of a batted ball at all - in other words, do balls get hit harder, or are the hits the same strength? (bcruise)
A: The only thing it changes is how fast the pitch visually comes at the batter. Nothing else changes on purpose, to give the user the best isolated control. Lowering the pitch speed slider too much can be a form of cheating, but the user can decide that. Most people are best off with default settings in the long run, but certainly some people have very fast/slow reactions. Also, there is actually a counterbalance that activates to compensate steal success % (to keep it constant to the extent possible).
Q: I find that with the pitch speed slider maxed out, a fastball is more of a true-to-life speed, and greatly increases the difficulty.. (bcruise)
A: That's a very good observation. Certain sliders are more useful and allow you to make meaningful changes in preference - I'd like to focus on developing and documenting more of these in the future. Each user deserves to play their ideal game as far as development, technology, and the interests of other users allow. Those limits explain a lot of decisions, but I can't elaborate in the time and space given.
Q: My question is about the timing window slider. How exactly does the timing window slider effect the CPU batters when adjusted? (ericjwm)
A: It doesn't affect the CPU. The timing window is there to adjust for very fast/slow user reactions, as you can observe from driving around town. This is a superior slider to the pitch speed slider for most purposes.
Q: What is the diff between the AI in the demo disc and the actual game disk? (kingdevin)
A: There are some improvements, but my focus was always on the final game and I didn't keep track. As far as demos go, I'd say it's a fair representation of that part of the game. There's obviously a lot more to the game such as RTTS, CAP, Franchise, roster updates, etc.. What you'll see in the full game is that every pitcher performs and feels different, whether using him, watching him, or hitting against him.
Q: Brian stated prior to release that difficulty level makes no difference when playing Manage Only Mode. Yet, many on this board claim otherwise. (SoxChamp)
A: Yes, it makes no difference. I'd like to keep answers brief, but I'll elaborate with a riddle. Notice that in Human vs. AI, the Human's difficulty depends on his own setting and the AI's difficulty depends on the difficulty setting of its opponent. In AI vs. AI, what's the difficulty of its opponent?
I think the best way to serve the community is to focus on questions which can save time and improve enjoyment/understanding. It's not practical or arguably useful for me to respond to each report about "Veteran is harder than HoF" or "Pressing X makes you run faster". There are better questions to answer, and most of my time needs to be focused on the game itself. I can sometimes explain why "Too many X" or "Not enough Y" is realistic given context and more info, but our best strategy is to let the game speak for itself and be tested by time. Gameplay is designed for realism and longterm quality, not first impressions. That's why the more you play the game, the more realistic it will be in context. A game with reversed priorities would show more flaws than merits as time goes on. Each strategy has its place in business, but I only enjoy working on the first type of game. The latter fails to deliver value for the consumer or myself.
Q: Any suggestions for maximizing the number of hits for the CPU on HOF difficulty? I have CPU contact and timing window maxed out. (seveb)
A: How many games have you played, and who were the pitchers you used and teams faced? These are the factors that count. Betting odds reflect how much influence these have. Some people are better at pitching than hitting, and many people are the reverse. Much of this has to do with your interests as well as strategy vs. execution skill. But in the long run statistics dictate a good varitey of games for practically everyone regardless.
A: The timing window has no effect on CPU, and I really wouldn't recommend changes to the contact/power sliders past a few ticks unless you're sure of it. Contact and power are not isolated sliders, and have unavoidable side effects at extremes. In other words, difficulty changes maintain realism (i.e. the worst MLB team vs. the best is still realistic), but giving an aluminum bat to players because they're in a slump or have bad technique produces unrealistic gameplay longterm. Note that if the game always produced "expected stats" for every user from beginning to end, that would mean skill/effort/strategy/learning didn't matter and the game is not realistic (in fact it would be trivial). What's more important is if the game produced realistic results in the long run given each set of actions, including relatively unorthodox actions.
Q: I'm a sim freak and the main thing that I constantly long for is more walks. I've toggled with sliders, and I still feel that I am no where near where it should be, getting 1-2 walks a game feels like an acocmplishment. (Nez477)
A: On the batter side, there's a learning curve to this and there are tips in the original message. Assuming you've peaked the learning curve, there are two sliders you can try. You can decrease CPU pitcher aggressiveness, which is a good isolated slider but might make the AI sometimes foolish in extremes. Or you can try the pitch count slider, which is not isolated but attempts to rebalance relevant factors (which is a lot) to the degree possible. Depending on your goal, that may be exactly what you want. This year's focus on quality gameplay allowed limited time for sliders, and I intend to focus more on slider documentation, organization, and effectiveness in the future.
Q: My questions are about swing influence with the Right Stick. Does it really help getting more doubles, triples, HR´s and a better hit variety etc.? Could I get the same result without using it? When does it work most effectively?
A: The right stick is very optional and influences flyball/groundball ratios. It's best for sac fly, double play, and hit and run situations, but I prefer to influence manually with the left stick in those situations. You're sacrificing AVG for more FBs or more GBs. If it's FBs, you might net a small increase in HRs/RBIs with a small decrease in 2Bs and significant decrease in Hits. The exchange rate doesn't pay off unless getting a FB/GB is a dominant priority, and if you're skilled you can do the same thing manually with L stick as well or better. So it's there for those who are strong at strategy but weak at reaction/execution or know a clutch situation is overriding. Also, the AI pitcher is adjusting for those situations (ex: more low pitches with high pitches pitched around as setup so you can't narrow the strike zone).
Q: Do you believe the power attribute for individual players is accurate? I am consistently hitting opposite field bombs with "non-power" guys....and this is strictly using a contact swing with the L stick in the down position.
A: Yes, the power attribute is fairly accurate under a realistic model that the game uses. Pressing select while at bat will show you the power rating. A bomb has a lot to do with the stadium in question with wind as an influence. If your question is about hit power for a given direction I'd look at the homerun distance instead without regard to bombs, and consider the type/location of the pitch against your input. A 350 foot hit that easily clears a shallow part of the a stadium (ex: near the foul) isn't really a bomb even if the game's camera indicates there's no doubt, and what I'd also look at is the bell curve of your distances to get context. A good meatball can be hit well opposite field, they just don't happen very often and would be hit even harder same field. Superior pitch recognition would increase that frequency especially on Veteran. I'm assuming you're not playing Veteran and have sliders at default. How do your high and medium power hitters compare to your low power hitters?
Q: (Regarding exhibition vs. season statistics) My question is - is this something you would expect, is it just coincidence or is it just 'me'? I'm not looking for you to unveil any 'trade secrets' but is there something different going on in franchise play? (sasman71)
A: There aren't any secrets here. I'd suggest playing on Allstar if you aren't already. Some people have a shorter learning curve on pitching than batting and that can temporarily influence low run games until they catch up on hitting, especially on Veteran. Before your post there are a number of comments saying most of the game is fine (presumably including your area) but they feel detail X should change. I'd imagine there's enough overlap to gaurantee that most people think most details are good, but there are always some people who think some things need changing.
Q: ..could you comment on the stolen base issue (way to many) and what we should do to fix it? (traumaip)
A: Have you personally observed too many steals, or is this based on what you've heard? How many steals / games have you observed? Have you compared this to MLB numbers? Let's discuss it with that starting point.
A: Statistically, it's gauranteed that some users will perceive too many steals over a few games. What's important is the longterm stats of representative users. I don't imagine that people who see no issues will spend proportional time commenting on each "too many X" or "too few Y" threads. You'd have to be very patient and have time apart from the game to participate that extensively.
A: If you do want to reduce steals, reduce the Steal Aggressiveness attribute. Note that this also plays a factor in discounting surprise - Podsednik and Reyes Jose don't lose as much effectiveness for repeated steals whereas most stealers do. These go hand in hand so it's still an isolated adjustment. Very few runners have high steal abilities - it's a sharp half bell curve. Stealing ability depends on speed but it's an additional ablity separate from speed.
Statistics Part 2
*) I did an illustration of HRs with the Red Sox, so let me do one for steals. I'll pick the Angels, first 5 games of 2007 (real life). Their first 3 games were vs. Rangers, and I'll use them too. Game 1: 2 SB by Rangers, 1 SB + 2 CS by Angels. Game 2: 1 SB by Rangers, 1 SB by Angels. Game 3: 0 SB by Rangers, 2 SB + 1 CS by Angels. Game 4 Rangers: 2 SB, Game 4 Angels: 1 CS. Game 5 Rangers: 0 SB, Game 5 Angels: 1 SB. Rangers were 5 for 5, Angels were 5 for 9. If the Rangers were your AI opponent, it's easy to think that steals are cheap because they succeeded 5 for 5 (that streak actually kept going IRL). If the Angels were your AI opponent, you might think steals were balanced fine but the AI just steals so often (almost twice a game not including batter contact). Over 162 games in real life, these same teams had opposite streaks. So 5 games is not sufficient to draw any broad conclusions on stealing.
*) With this sample of 2 out of 30 teams and other more extreme examples, imagine the statistical variety you'd see in a million users playing the game. If you saw these same real life results in your game (9 steals / 5 games, 5 for 5 steals), would you draw any longterm conclusions about steals? How do these examples compare to the examples cited by other users in the game? How many games did they observe? It would actually be trivial to design a game that gave you expected results, but it would be very unrealistic and pointless. As mentioned, our game is not designed for first impressions due to focus on quality and correct behavior.
*) It's worth noting that steals are valuable but the cost of being caught are roughly double (slightly less with strong strategy). In the long run, if a team scored 139 SB and 55 CS like the Angels - they'd be expected to net perhaps 1 extra win out of their hundreds of steal attempts. That doesn't mean they won't win 4 or 5 games due to steals, but you have to subtract off the 3 or 4 games you lose in the long run due to caught stealing (you'll never know which ones out of the 55 mattered).
Due to time, I won't be able to discuss examples like this for every stat in the future. But these resources are freely available to the public.
|
Operation Sports
03-18-2008, 08:45 PM
Linear Mode
