Amaroq
Posts: 1100
Joined: 8/3/2005 From: San Diego, California Status: offline
|
I've been thinking about "in-game management" recently, separate to Steely Glint's posts on the topic, and then I promised him I'd post on the subject if he didn't. I never got into PureSim's implementation of in-game management; there wasn't enough there to 'grab' me. I don't think I've played an in-game management computer game since the days of Earl Weaver and what was the SSI game? I'm not sure. I do play Statis-Pro baseball, the board game, but it feels too 'formulaic' to me. Thinking about it, I've long had the sense that an "at-bat" based mechanic is a bit sterile for a 'management' simulation. In real baseball, you select strategies on each pitch - and the changing pitch counts affect the strategies you might want to select. Little things like whether the first baseman is holding a runner on first, or playing at the edge of the infield grass because the hitter is known for his ability to bunt for a base hit, or guarding the line, or positioned normally have measurable effects on the odds of the batter getting a hit, or getting a hit for extra bases, or being able to bunt for a single. So, a game which did an "ideal" implementation of in-game management would need to present the user with the option of making strategic changes on each pitch. Some users wouldn't want this - it should probably be an 'option' which is user customizable. Some users would want to give a new strategy each pitch. Some would prefer the old stand-by of "I give one strategy for an at-bat." An intermediate point might be "I give one strategy which is in play until the count reaches something which invalidates it." For example, a 'bunt' strategy would be left in play until there were two strikes, then the user would be prompted for input again. A 'take' strategy makes sense only for one pitch. And, obviously, GM players would want to continue having the AI manage their games all around. What would the ideal in-game management functionality be, on a pitch-by-pitch basis? Batter: - must swing - be aggressive - green light (use your discretion) - swing for fly ball (sac fly) - contact play (ground ball) - protect the plate - sit on a perfect pitch - take - show bunt but take - bluff a bunt, then swing away - bunt for sacrifice - bunt for a hit Runner: - must steal on this pitch - green light (steal if you feel like) - run with the pitch - run on contact - hold on base General baserunning: - aggressive - normal - conservative specifiable for each runner and each batter Pitcher: - right down broadway - go after him - nibble at the corners - pitch around - pitch out - intentional walk - pick off - use your discretion - pitch inside - pitch away - might also include pitch selection - full windup vs stretch Catcher: - aggressive, tries pick-offs, throws out runner, etc - normal, throws out runner if going - conservative, throws out runner only if good chance of getting him - defensive indifference (let runner steal 2nd) Defensive positioning: - for each player, 'deep', 'normal', 'close', 'in'. - for each player, 'left', 'normal', 'right' - for corner infielders, 'guarding the line', 'on bag' - if runner on second, does SS or 2B hold runner? Defensive plays: - the rotation play, corners charging, as a bunt defense - the shift (anti pull-hitters) Defensive philosophy: - aggressive (throw out lead runners, etc) - normal - passive (don't take chances, get sure out) ... Obviously, these need to affect the outcome of each pitch. The closer an infielder is positioned to the plate, the more likely a ball hit in his direction is to get past him for a hit, but the more likely he'll be able to get runner out. There would be a real interaction between the batter's instructions and the pitcher's instructions, which should be able to override the hitter's natural tendencies or the pitcher's natural tendencies. Different choices and different matchups would utilize different aspects of the players attribute range: you could turn even a great control pitcher into a walking machine by having him pitch around everybody, or turn a patient hitter into a strikeout machine by forcing him to swing at every pitch. You'd be faced with decisions like "Do I want my hitter to shorten his swing and protect the plate with two strikes, or do I trust him with a green light?" "It's a three balls, no strikes count: do I put one right down broadway and hope he's taking the pitch, or do I continue to nibble and chance walking him?" Obviously, throwing a first-pitch fastball right down broadway against a power hitter should have a good chance of yielding a home run. Injuries should be affected also - a team playing very relaxed on the basepaths is less likely to suffer injury than one playing very aggressively; an infielder positioned 'deep' is a little less likely to be hurt than an infielder who is 'in'. ... You'd have to have the AI utilizing these sorts of things, and reacting appropriately. For example 'bluffing a bunt' on one pitch should result in the third baseman being positioned a little closer to the plate on the second pitch, which makes it just that much easier for the player to get a hit up the third base line. (A's third baseman Eric Chavez says he hates facing Ichiro, because you have to come in close to respect his bunting, but he can really rip one past you.) Its no use having a 'bluff the bunt' option if you can't use it to adjust the AI. Likewise, the AI needs to be able to use each option, and use it well, otherwise a human player managing for his team will have a dramatic advantage. I think it would play well with 'manager philosophies', e.g., giving each manager a discrete weighting towards different possibilities (and depending on score and situation, obviously) so that you would get a definite 'feel' from different managers: one manager using the bullpen like there's no tomorrow, another trusting his starters too long. One manager might feel very aggressive on the basepaths, while another is more conservative. You 'profile' the human manager, generating what you think of as his 'weights' if you were trying to model his behavior. As you generate data indicating which strategies are successful, 'young' managers entering the game emulate the successful managers of the past in some areas, while trying their own strategies in other areas. Ideally, you might have different umpires with narrow or wide strike zones, giving you a sense of how frustrating it is to nibble on a narrow strike zone, etc. 'Control' would begin to affect the pitcher's location of pitch - if you said 'go after him', a poor control pitcher might turn that into a mistake pitch right down the middle, or might happen to turn it into a nibbling-at-the-corners pitch. ... You'd also need a bunch more possible play outcomes. Throwing errors are key - its one of the negative aspects of your catcher being aggressive, for example. He might throw it away on a pick off or a 'throw out runner' play. Likewise, the decision to have outfielders throw or not, one of the 'negatives' of attempting a throw should be the chance of a throwing error. You'd need 'out stretching for a double' as an outcome, to discourage aggressive baserunning. You'd want to have the 'triple play' possibility, as it usually would occur if the manager is 'sending the runners' with nobody out; it needs to be a possibility in that situation. Likewise, the strike-em-out, throw-em-out double play needs to be a potential result for a 2-strike steal attempt. The outfield "in" would be less likely to give up a single, and more likely to give up extra-base hits. The outfield "deep" would be more likely to give up singles, and less likely to give up extra-base hits. It should have the possibility of cutting off a triple (holding the runner to a double), and even cutting off a double (holding the runner to a single). Different 'park effects' would become more important. I'd like to see the outfield walls have a large number more impacts on the game. High, close: can turn a home run, triple, or double into a single. Can turn a pop fly into a home run. No possible triple. High, normal: can turn a home run or triple into a double, can turn a deep fly into a home run High, deep: can turn a home run, triple, or double into a deep fly out. Can turn a double into a triple. With poor-range fielder, can turn a triple into an inside-the-park home run. Medium, close: can turn a double or triple into a home run. Can turn a pop fly into a home run. No possible triple. Medium, normal: doesn't change anything Medium, deep: can turn a home run, triple, or double into a deep fly out. Can turn a double into a triple. With poor-range fielder, can turn a triple into an inside-the-park home run. Short, close: can turn just about anything hit to the outfield into a home run. Can turn a triple into a 'ground rule double' Short, normal: Can turn a double or triple into a home run. Can turn a triple into a 'ground rule double' Short, deep: can turn a home run, triple, or double into a deep fly out. Can turn a double into a triple. With poor-range fielder, can turn a triple into an inside-the-park home run. ... It's a lot of work. I don't know if PureSim is the right game for it, and there are a lot of things I've glossed over: what the GUI would look like for this, whether it would remember your strategy from the last pitch or 'reset' each time, what the 'break-off' point for each strategy is (in terms of pitch counts, etc.) It would be particularly hard to get the right percentages on each pitch - e.g., what's the percentage of pitches that are called strikes, swinging strikes, fouled off, and taken for a ball? These numbers certainly aren't available on a per-pitcher and per-batter basis from Lahman, and most baseball games reduce things to an 'at bat' granularity so that they can better achieve statistical reliability. What are hitter's batting averages with two strikes against them? How does on base percentage change if the batter gives up a first-pitch strike? Tough to get 'right'. But, that's what it would take, I think, to really put a game 'above the crowd' in terms of in-game management options. ... That's a big mouthful - anybody want to post "What I would like added to PureSim, without the rewrite Amaroq's just posted?" I think somebody had done a very good post on that topic on the old forum, maybe somebody can bring it over?
|