Quote:
Originally Posted by mpejkrm
Man, the AI is sooooo far ahead of where it was when I first started playing (12) or even back at 18 and 19. It's not perfect, obviously, but I don't find myself scratching my head nearly as often as I used to in the past. Player transactions make sense to me 8 out of 10 times or so, I'd guess. 4 years ago it might be 1 or 2 out of 10.
|
Improvement of trading algorithm fades compared to structural problems.
Current system is prone to imbalance. The way AI treats high salary (even if player is good) and the way it treats prospects is two different universes. Seems like it has moneyball values that override everything. And there's timing and "here&now" aspect to GM jobs that can't be ignored. Prospect might grow into great player in 2-3 years... when you will be long gone. That's why sometimes there's a need for productive vet and CAP is just there to spend this year. Multiple examples of negative consequences of that in this thread. Even if you don't engage in exploiting AI, trading block gets clogged with players that should never be there. AI ignores opportunities to use "qualifying offer" on star players that refuse to sign extension, it just puts player in last year of contract on trading block.
Revamping trading could be great improvement to playing experience and solve all these problems. Off the top of my head:
1) Differences in evaluation.
Exploiting through unlimited requests exposes underlying problem. AI never really wants different players or options, just quality (talent/potential stars provide some idea). And most of the time it's for players on position that they got covered. So what's the point of randomizing coaching/scouting ratings if there's no "AI decides to take a chance on your prospect, who you're pretty sure is 1/1.5 star bust". Dynamic potential changes&two sides of scouting mistakes can provide crazy randomness fun and increase total trades number. Right now it's just "Oh, you want Mike Trout to sweeten the deal? Yeah, you and 28 other GM's"
2) Negotiation points to increase variation of trade evaluation.
Think owner goals&timing for GM. Maybe addition for GM lifeline value is necessary to realize this. F.e. Dodgers getting hit by injury bug doesn't bother them. Strong farm means that they can make playoffs even with stars missing some time, so no need to run around and make panic trades and clog roster. Less fortunate teams&GMs have different life. Star player is done for the season and you're neck and neck in playoff race? Time to move some furniture, and prospects are not "Untouchable" anymore. Owner wanted profit and had none, owner wanted playoffs and you didn't deliver, your reputation is not good, players as assets value haven't increased under your management? This year becomes key and there's no tomorrow. So that can be modifier for actual negotiation points.
Improving in-game performance weakness: lineup creation strategy for AI can be helpful here, but it's tough to think of all variations. So just use owner goals and stats from Team Home Screen as vector. Different scouting plays a part here too for "slight upgrade" situations. Maybe they feel like their player is stronger while your scout thinks opposite. But if team lacks AVG and you offer them great contact hitter on challenged position like catcher, they can't respond with "No, our SS prospect who can't hit and will never overtake star player at his position is untouchable, offer MUCH more". Maybe adding "meteoric rise" progression scale or values for prospects is a good idea for this type of situation. If player rushes through minors or shows excellent progress in attributes, AI might avoid trade even if it makes sense and offer
other prospect of same quality. Especially since progress of player is better known to his own team. But other than that, improving weakness is a good constant for AI GMs to have. Adjusting through other available stuff (cap space, organisation mode) should be simple.
Depth (esp. positional): Outside of obvious "We have 3 good SS prospects, we don't need another one" there is another key improvement through addition. Addition of absolute/relative scouting ratings or rankings. Relative rankings (can be as simple as already existing positional ranking) helps AI exclude "minor improvements" situations. Unless player is struggling, injury prone, bad character why take a chance on similar player and risk adjustment issues or scouting mistake? To change #7 to #4? AI green lights "similar" player exchange too often, it should only be for scouting valuation differences or if he likes
other player in package deal to much to let it go.
Absolute ranking helps fix obvious problem of relative one: when #1 Catcher or SS is All-Star hitter, changing #2 for #1 might be a very necessary hustle. And it will also fix long time problem with AI being uninterested in league's best players (for both OOTP and FHM). Unless user is bored with saying no and activated "untouchable" option, he should be presented with all kind of crazy packages. 3 pitchers and 2 batters from rotation, you name it. But right now you can have GOAT and get 0 offers during his career. Funny how nobody wants that 18 WAR at top 20 salary. With absolute rankings teams would chase obvious MVPs and "Get MVP" from owner won't be just another variation of goal with "Hey this guy won it 10 years ago" counting as completion. Max of 30 runs from stealing, 30 from baserunning, 90+ from batting, defense is quantifiable through WAR too. So not impossible to give AI some benchmarks about possible production and use it as vector. Dead era = no activity triggered. Multiple GOATs era = massive activity, maybe poor guy gets drafted by struggling team and some lucky GM will steal him and resign him to "max, no questions asked" deal immediately.
Popularity, attendance, interest: Make it available in trading negotiations, since it's big factor in Owner Goals and in general (budget size). Another useful addition can be "Fan Profile". Public in Detroit or Pittsburgh might embrace small market mentality and defensive plays/players. Yankees have 40k fans wanting to see Home runs. So two very different player profiles will fit "boost to attendance" mission. Maybe make Fan Profile scout-able or dynamic, so that you can make mistakes as GM mislead by what you "know". You thought your fans just like winning, but they got really invested in popular/charismatic guy who you traded and don't care that you still win 70%.
Roster capacity awareness: There's limit to everything and AI should know it. Offseason especially. You have CAP, you have Free Agency, you have Rule 5 and you'll have spring training to see if somebody made a leap or will catch fire. Right now you can flood AI with players and CAP and exclude them from participating in all mentioned activities. I spotted it once and just unloaded all my high paid vets to only other team with significant CAP space, then signed all elite FAs with discount just to see if it would work. Nobody would really rush to take worse player, so maybe add
"Delay trade" option for offseason? If they get FA or run out of CAP space it's cancelled. If not, it's agreed in principle and goes through before preseason starts.
And in general, AI is not good with waivering/Rule 5 practices, so maybe have roster roles. You have expectations now, but a little different. Starter if healthy, starter (better than alternative), sub, practice in AAA&sub if necessary, 1 year away, raw prospect (probably 2 years in minors before majors possibility). Number of positions limited, number of games limited etc. So AI would have an idea of how balanced roster is. Why clog playing time and send people to waivers for no reason?
Owner input: Happened countless times IRL. Meddling owner might add a new layer of difficulty. Best use imo is to stop cutthroat manager (most of us) from trading players all the time. Or the opposite as exemplified in "Last Dance" with guy from CWS. "You think we have one more run but I have to spend and risk with projected budget in red? Not gonna happen". Even hands off owners might be interested in blockbuster trades and bring modifier to negotiation. F.e. if we want winning and prospects at the same time, he can come in and say "ok, get this star guy and top prospect goal is void/accomplished". If player unperformed, you get less punishment.
3) Timing of fan interest/reactions changes:
I wrote about it for FHM. But can we not get instant +/- on fan reactions. Until 3-way deals are possible, I shouldn't hurt in attendance because of player who haven't been on roster 1 day. Either games played counter or start of season as activation date. If I'm dealing in offseason, why would fans react to it. Guy who lead HR race gets traded - this is understandable. But not in offseason or about players that have been on roster less than X days.
4) Way of interacting:
Conditions from get go, side recognizes who is beneficiary. There are a lot of options to be suggested for interactive screen. But for simple, doable solutions I thought of these two.
Conditions can be based on many things. One of them is your reputation. F.e. now you have "counter" on drafted players WAR. Maybe have similar for traded pitchers/batters. So if you're legendary guy known for being "pitcher whisperer" AI would only let you trade for pitchers he doesn't need. Anybody of any value is off the books. Most obvious and logical is not wanting to help your division rivalries. Never help improve, but always happy to take your best players if you ran into financial problems as small market team. Dump player goal from owner can be realized here: we want you to take him but we will sweeten the deal. So if you came for other players, AI will be very stubborn unless you take big salary too (if you have CAP for it). Maybe add "I owe you" modifier, when you did somebody solid, or you had Win-Win trade in the past (to substitute whole "relationship" module headache). "We lack quality relievers, so any deal should involve one" etc. Good examples are not hard to find.
Beneficiary: continuation of salary dump idea. If AI is using you to salary dump "overpaid veteran" (they have that evaluation instrument) they become beneficiary in trade. So it means they should trade more easily. Right now it's like you owe them, because something overwrites priority. Would be nice to have sides understand who is helping who. Conditions about helping trade instead of usual one shouldn't be hard to quantify. Trading block, overpaid vet dump, improving glaring weakness, trade helps to accomplish owner goal, improvement in tight playoff race, quantity for quality (3-4 meh prospects/players for elite one should gravitate toward 4).
5) Package deals, suggest what you want.
This would be gamechanger. Either have AI tell you "You want Mike Trout? I will need two 3+ star SP, 4 star CF (to sub), and improvement RP (at least 1 star)". Or ask AI what he wants most of all. Batter, pitcher, prospect? Maybe do it optional, some GMs might not be willing to show their cards. But will listen for your offers. Get negotiation mood involved, or GM personality. Some will "Okay I listen for your offer" and hang up if it's bad or not enough (and add negative history points). Others will be patient and listen w/o getting upset. But combined with previous ideas this would be awesome improvement. As it is now, I know from experience what it would take for AI to fleece me for Top Prospect
So it's doable, and trade was Win-Win. Only thing I don't know is what it takes(have to do it manually and ask 10+ times). Bring GM character or "I owe you" points here to vary it. Some might want to ask for half a kingdom, others will tell you exactly what it takes. Maybe they give you package, but owner swooshes in and says "No way we losing him, even for 5 best players you have".
6) Multiway action:
Bumping heads like in moneyball movie. Involve other team to see who can offer better package. Desperate GMs should be more vulnerable to this, but at least there will be reason for it.
3-way trades are also a good potential improvement. One way to say "Not enough" is to say what you want and send you looking for it. Same as delaying trades in offseason, freeze the trade state and have option to go looking for 3-rd team to facilitate what's necessary. If AI needs 3 star RP, you have 3 star 2B, go find a team that looks for 2B improvement and maybe they will give you 3 star RP.
Ok, thanks for reading (hopefully)
