Prediction or Probabilities ?

[Tarquin]

@ imperfectluck

You seem to have definitively proved that you can control the ladder by playing slowly and safely.

However, you say you're finding the game boring.

You know you can be a ladder master, perhaps you should build yourself a high risk, high reward team or a team with an interesting theme and attack the game with a new playstyle and interest? It is a nice thing to do, as long as you don't mind winning and losing to risks and being happy with the risks you took, as they were for a good reason, even if they don't pay off.

P.S. I remember playing you about 3 weeks ago, I recall a close run battle. You'll be glad to know you eventually won, but it was a hard, long battle, as you said yourself after it. You seemed to show dynamic play in that battle, so maybe you can be tempted out of your safety zone xD

 

[ivar]

"Instinct" is just the word we use when we are unaware of what thought processes brought us to a conclusion.

It is impossible to stop predicting. If they have Scizor out and you use a Fire move, you're predicting that they're probably staying in (or that you can handle whatever they switch to).

QFT

Considering nearly every battle of Pokemon is just a big "Prediction". "What move is he going to use now?" "Im sure he wont switch if i keep my pokemon in play" "If i use this...." (Unless your psychic and can read your opponent move for move, which i highly doubt anyone can do). Predicting is a big part of life, not just pokemon =P.

Ya know, I said that first. But Obi's got the respect 'round this here parts that DaBossMan just does not command.

Prediction is not pure instinct, nor can anyone win without prediction. Every move you make (whether it even be in the realm of pokemon battling) is a prediction. People seem to think that prediction is only when the effective decision is an apparent toss-up (or, effectively a guess), but this is not the case.

The only real problem seems to come in the early game when (most of the time), for things like Salamence, you don't know what their moves are, and you don't know what the opponent's other Pokémon are. So, using the information that you are given (what lead are they using, did it use Reflect/Light Screen, is it not gaining Leftovers recovery, etc.), you need to deduce what you can, if vaguely, what the opponent's options are

You cannot deduce, however vaguely, the moves and EV distribution of your opponent; you can, however, induce, based on previous experience and knowledge of the present metagame. Semantics, I know... but I'm anal like that. Plus, you should know the difference, Mr. Mathematician!

 

[Yamitahuvaka]

Ya know, I said that first. But Obi's got the respect 'round this here parts that DaBossMan just does not command.

My mistake, i only read from the 3rd page =P

 

[Potential]

You cannot deduce, however vaguely, the moves and EV distribution of your opponent; you can, however, induce, based on previous experience and knowledge of the present metagame. Semantics, I know... but I'm anal like that. Plus, you should know the difference, Mr. Mathematician!

Hah, touché. Well, I guess it would be dependent on the case. Although I think that is essentially the point that I was trying to make (and failed, apparently). But I didn't say that you could deduce moves and EV's; I just meant you could find a very (emphasis on very) vague idea of the opponent's basic plan based on what the metagame is and what has happened in the battle so far; you can't just discount nonstandard sets completely. I'd agree, though, that induction is more correct/likely in this case than deduction. However, as you progress through the match, I am still convinced that there would be more and more deduction, not replacing/removing/somethinging (yeah, I can't explain myself properly, sorry) the induction, just compounding on both.

 

[ivar]

It's impossible to deduce your opponent's actions (if that is what you are postulating) during a match, since you have no necessary knowledge of your opponent's actions.

Think of the classic "will the Sun rise tomorrow?" question. Sure it will, but we can only make this claim based on induction (since the Sun has risen every day before). Technically though, we have no guarantee that the Sun will actually rise tomorrow; it's just highly probable. The same goes with pokemon: there is no guarantee of your opponent's moves, regardless of how probable a certain move is.

I may have misinterpreted you Mathematician5, but even so, I just don't see any element of battling, when considering your opponent's actions, being deductive, even if the element I am describing is not the element that you are describing.

To elaborate, with some relevancy to the OP's concerns: if you accept the claim that all elements pertaining to interpreting your opponent's actions is inductive, then you have to accept the claim that the interpretation of your opponent's moves is based on prediction (since prediction is inherent to induction).

 

[Potential]

Well, I guess there's always the exception, but let me give an example of what I'm talking about (a basic one, but still; this sort of thing does occur): you have out a Choice Band Flygon and it's just fainted a previously weakened Swampert with Earthquake (and due to the damage dealt, your opponent knows it's Choice Banded) in early- to midgame. Your opponent sends out a Heatran, from which action I'm pretty sure you can deduce, unless he's bluffing, that the Heatran is Scarfed, and that it knows HP Ice or Dragon Pulse (or even simply Fire Blast; it all depends on your HP left); otherwise (again discounting bluff), why would he send in Heatran just to get killed? Once more, like I said, there is always the chance of bluff, but as the game progresses (and you haven't accidentally killed off your Flygon, in this case), you will likely know that the Heatran is Scarfed, say if it outruns your Starmie, etc., giving you more evidence (read: not exactly proof in some cases). I think this can safely be called deduction.

 

[ivar]

Well, I guess there's always the exception, but let me give an example of what I'm talking about (a basic one, but still; this sort of thing does occur): you have out a Choice Band Flygon and it's just fainted a previously weakened Swampert with Earthquake (and due to the damage dealt, your opponent knows it's Choice Banded) in early- to midgame. Your opponent sends out a Heatran, from which action I'm pretty sure you can deduce, unless he's bluffing, that the Heatran is Scarfed, and that it knows HP Ice or Dragon Pulse (or even simply Fire Blast; it all depends on your HP left); otherwise (again discounting bluff), why would he send in Heatran just to get killed?Once more, like I said, there is always the chance of bluff, but as the game progresses (and you haven't accidentally killed off your Flygon, in this case), you will likely know that the Heatran is Scarfed, say if it outruns your Starmie, etc., giving you more evidence (read: not exactly proof in some cases). I think this can safely be called deduction.

But if it were to be a deduction, it would have to be definitive, not likely (and it cannot be definitive if the possibility of the bluff exists; I've used the bluff myself many times, actually). If it is likely, then it becomes induction/abduction (for deduction is certain and induction/abduction is, at best, probable). I think you fail to to understand the fundamental difference between deduction and induction. Btw, when I say abduction I don't mean kidnapping, but investigative abduction (figuring out a probable conclusion based on the evidence).

Here's another scenario. Imagine that same CB Flygon up against Heatran (without a CS, as this was determined at some earlier point in the battle), and the Heatran is the trainer's last pokemon for the match. You would think that you would be able to deduce the CB Flygon trainer's victory, but how do you that the trainer won't purposely leave the battle while it's progressing? I know this sounds somewhat absurd, but bear with me here; there is no guarantee that the trainer even wants to win, even if every move before was seemingly designed with victory in mind. He could just be fuckin' with you for giggles and shits. So, we, at best, can only induce that CB Flygon will use EQ in order to win the match, even though it is extremely probable (as in, so probable that I've never faced anyone who did that in a similar scenario). Thus, since it is only probable, it cannot be deducible.

Again, I will reiterate: deduction, if the premises are sound and the form valid, must always be true. But no move is ever certain in pokemon. None. I hope the above example articulates exactly how uncertain even the most apparently certain moves are. Pokemon is pretty much a game of odds and the best battlers can figure out the best odds, or assess the optimal success of a particular move based on the odds and rewards (i.e. risk management, such as choosing Fire Blast over Flamethrower, for an easy to understand example).

 

[Potential]

Okay, I concede... Well, hey, I haven't gotten much education in the way of logic (I'm in 9th grade, and I had a tiny little thing about truth tables, etc. in 7th...), so most, if not all, of my definitions would be partially messed up. And I know that every move in Pokémon has some degree of uncertainty, so we were really saying the same thing... just my definitions were messed up. Well, now I know. But I should say this: more often than not, it will not be a bluff (otherwise the "bluff" will become too obvious in the general metagame/playstyle; don't pick on this point, since I can see flaws in it, but it's late and I can't think to fix it), so you can tell to (I'll just choose some arbitrary number), say, 95% that the Heatran is Scarfed, and you need to weigh if it's worth it. And that's what it really boils down to, isn't it? "Is it worth the risk?"

 

[husk]

Taking no risks is by far the easiest way to keep a firm grip on the top of ladder, as I have proved over and over again.

Having the patience to ladder with your team is a talent in itself...lol
I can understand using it in tournaments but I find pokemon boring when I'm not actively challenged. Not that that means I spend more than 5 seconds per move, but you get what I mean...and in a way I think that's why you find pokemon boring.

but which peak is easier to reach, the peak in teambuilding or in prediction? generally most average players can predict. very few players can build really good teams.

Prediction is easily the more challenging peak to reach. Since prediction is relative to your opponent, in order to reach the peak of prediction you must play the second greatest player (since you are the greatest) and dominate him/her. Have you ever played a perfect game versus a good opponent? I've been playing for 4 years (at a high level, in my opinion, for 3 or so) and I've only ever done it twice and I promise I could never replicate the feat on call.

Actually, before anyone responds let me describe a perfect game...it does not involve winning 6-0 but it could. It involves knowing exactly what your opponent will do and then correctly responding to it for ever single move of the game. You might lose a pokemon because CB mence outraged your scizor to death with a crit but as long as you were aware that outrage was coming and that scizor could BP mence for the KO it's fine. This probably happens quite often vs lesser players but can anyone say they've done that versus an extremely good player (someone candidate for that #2 spot)?

In my two instances I knew both of the player's teams before hand as we've played quite often and in our previous battles I had lost some and won some with imperfect games played by both us...that's not to say they weren't good games though :). A real perfect predictor could do it without knowing the teams or the other player's styles...I can't really explain how though.

If you wanted to, you could learn to build teams by simply studing how teams are built and working really hard at it. A flawless team in this metagame is easy to come by. In fact I could just borrow someone else's flawless team if I were terrible at team building. I believe that reaching the peak of prediction is something that will never occur...unless you can play perfect games vs anyone at any time.

 

[Stathakis]

I believe that reaching the peak of prediction is something that will never occur...unless you can play perfect games vs anyone at any time.

exactly. the "peak" of prediction is impossible to reach. however, there is no area between average predicting and perfect predicting. you may be right in most instances, but against any given player you can predict that they will do something. but they can predict that you will predict that they will do something and in turn do something else. you can, however, predict that they will think in that way and do something else. it's like reverse psychology. you can predict a step ahead of your opponent, and depending on where they decide to stop and where you decide to stop, whoever gets lucky and guesses right wins. there is no way to be above your opponent in this area unless you can in fact read minds or have a little gnome looking at their computer screen and telling you what they are doing. once people hit that level of prediction, since it is impossible to predict perfectly 100% of the time (a fact we both agree on), it is very difficult to get above this stage. in other words, it is very difficult to outpredict a good player. I can, however, build a team superior to theirs and think correctly in the longterm, taking out what I need to take out and winning without having to play any extreme mindgames. since everyone hits the "just below perfect" level of prediction so easily, it's easier (and more reliable) to win by beating your opponents in the other areas of pokemon

 

[paramylodon]

Here's another scenario. Imagine that same CB Flygon up against Heatran (without a CS, as this was determined at some earlier point in the battle), and the Heatran is the trainer's last pokemon for the match. You would think that you would be able to deduce the CB Flygon trainer's victory, but how do you that the trainer won't purposely leave the battle while it's progressing? I know this sounds somewhat absurd, but bear with me here; there is no guarantee that the trainer even wants to win, even if every move before was seemingly designed with victory in mind. He could just be fuckin' with you for giggles and shits. So, we, at best, can only induce that CB Flygon will use EQ in order to win the match, even though it is extremely probable (as in, so probable that I've never faced anyone who did that in a similar scenario). Thus, since it is only probable, it cannot be deducible.

I can actually recall at least four specific times that has happened, if that can further strengthen your point on deduction versus induction. When I first wrote this I said three, but then I actually remembered a fourth specific case. May be even more that I just forgot about. This is excluding accidental disconnects of course (which can also always happen).

I agree with Stathakis here.
After learning to think and weighing the probabilities, risk/reward factors, and judging the opponent's play style and tendancies as well as likelihood of them changing it (which is for the most part a guess), there isn't really any point above this one can reach.
One cannot deduce enough about any specific person to try and determine with accuracy beyond a specific point (in context of trying to win a battle; if the opponent wants to lose like in the quoted example that would only be benefiting you here). This is the practical limit, which isn't necessarilly that difficult to achieve.

Personally, I'm under the impression that building a team is much harder. The success I have when playing is based largely on the team I'm using, and the team I've been most successful with (and reached #20 on the OU ladder with a couple of months ago) is probably not really that spectacular of a team. I'm sure there is better (that teem keels over and dies when my opponent has a LO mamoswine doing anything other then leading).

Although I do have to recognize that I've used some teams (such as that of Stathakis and Reyscarface) and have been relatively unsuccesful using those teams compared to either of their owners, implying a difference in ability. I've already touched on my lack of thinking compared to them, although, this is thinking and planning based on logic. These players are however both in the "anti-prediction" camps; my point is that they are better due to logical planning and spending the time to think (or better experience, and thus being able to autopilot better). Not making unsafe choices based on the belief that the opponent will do something, as the definition of prediction often is (fireblasting a dragonite hoping they switch to scizor).

Thinking when battling is something I do not believe I do enough of, although it is a relief to see that ipl apparently doesn't do to much of it either.
As a comment to ipl, you've used the same team type, but when watching your games I've seen your team go through a lot of variations, so I wouldn't say it is exactly the same team.
You clearly are better then I, or many others, somehow. You have to give yourself some credit here. You are playing intelligently while playing safe.
You posted your RMT, and I've seen some relatively unknown people using it. These people weren't as good as either you or pana when using it, and met know where near the success. Simply "playing it safe" the entire game doesn't mean you don't do some specific things (even if its become habitat/obvious enough that no thinking is required).
My point here is that to reach great success, one musn't only play it safe and not predict. One must do these things, and somehow manage to still do some work towards a goal, and make key (safe) sacrafices, etc, in context of the situations.
Otherwise anyone could reach and hold the number one spot on the ladder using the team ipl posted, or any other such solid well built team-myself included. The point of course is: I haven't.

 

[Revolution.Z]

prediction is really the key to winning in any match. I like to go with obvious moves early on in the match to prevent me doing something stupid and like Ice sharding an ape predicting a flier or levitate to come in. If you then see ur opponent is also predicting nicely thats when i turn up my game. So really you predict based on your opponents skill level

 

[husk]

exactly. the "peak" of prediction is impossible to reach. however, there is no area between average predicting and perfect predicting.

I disagree...when I said that there was nearly no way to reach the peak I still believe different people are different lengths down the path. While it might be logically valid to say that on a road that begins at point A and ends at point B (an infinite distance away) everyone is equally far away from point B because whatever distance you have traveled is insignificant compared to infinity. However, that doesn't mean each person is the same distance away from starting point, point A. There are average players and there are amazing players...average players predict averagely whereas amazing players predict amazingly. Sure, each game is a fluctuation...sometimes amazing players are outpredicted, sometimes they outpredict themselves but usually they are far better than average. I find it hard to stop believing in this because there are obviously amazing pokemon players and there are average ones. Both can build teams, both can steal teams...the difference has to lie somewhere else. I believe it is prediction.

 

[Syberia]

A real perfect predictor could do it without knowing the teams or the other player's styles...I can't really explain how though.

Prediction with no knowledge of an opponent's team is really a game of probabilities and usage statistics. I can predict Scizor/Heatran coming in blind all I want, but whether or not my opponent even has those pokemon is strictly up to luck of the draw.

 

[husk]

Well I agree it is a bit of luck if you're guessing.

But lets say you've seen three of your opponents pokemon: lead scarf meta/celebi/lo salamence. When you send out your gyarados you can anticipate he will send out celebi and so on...but what if you send out tinted lens specs yanmega on his celebi? Assuming you have HP: Ground would you use it because your opponent most likely has a Special Defensive heatran waiting simply because it would fit his team extremely well? Or would you bug buzz because it hits everything hard so far? Or would you even Air Slash because you believe your opponent would try to get mence in? Or maybe even predict that your opponent assumes you will hypnosis and just Bug Buzz anyway to hit whatever that pokemon is hard? Or maybe double-switch to your CBtar anticipating a Blissey? Or maybe even something else?

The perfect player would know exactly what to do...somehow. It's not something we can discern it's just theory, heh.

 

[Rurushu]

Well I agree it is a bit of luck if you're guessing.

But lets say you've seen three of your opponents pokemon: lead scarf meta/celebi/lo salamence. When you send out your gyarados you can anticipate he will send out celebi and so on...but what if you send out tinted lens specs yanmega on his celebi? Assuming you have HP: Ground would you use it because your opponent most likely has a Special Defensive heatran waiting simply because it would fit his team extremely well? Or would you bug buzz because it hits everything hard so far? Or would you even Air Slash because you believe your opponent would try to get mence in? Or maybe even predict that your opponent assumes you will hypnosis and just Bug Buzz anyway to hit whatever that pokemon is hard? Or maybe double-switch to your CBtar anticipating a Blissey? Or maybe even something else?

The perfect player would know exactly what to do...somehow. It's not something we can discern it's just theory, heh.

In that case, you should probably see wich option offers less risk and more benefit. Using HP ground offers a lot of risk, because your opponent could have no specs yanmega switch in, and let celebi to thunder wave you, or he could send metagross.
Air Slash hits everything hard (almost 2HKO the special defensive Heatran) and bug buzz does the same (it only doesn't hit heatran).
Switching to TTar could be risky to (if the opponent knows you have a specs, he could send Scarfgross to take a hit and then lock you in one move + getting you into some priority move).
Oh well, you could also use U Turn (Specs Yanmega doesn't really need more than Bug Buzz and Air Slash) because it would hit celebi for a nice damage and scout your opponent switch-in.

So, IMO the "better" prediction you can do is the one that can't turn your advantageous position into a bad situation (unless you know you opponent really well/need the prediction to win/the prediction puts you in a win/win situation if it works).

Another issue when predicting is guessing you opponent "level". There are a lot of times when you gets in a bad situation because you overestimated you opponent (bad situation, not a lose situation, because like i said before, risky predictions should be avoided imo).

Predictions involve probabilities, but you should weight the benefits and losses and act accordingly (probably scouting your opponent before, so you can know if he will do the same).

 

[Syberia]

Husk, if you're playing a good player who knows you can make all of those predictions and is capable of counter-predicting, it's really nothing more than a guessing game on both sides. Each move is most probably just as likely to occur.

 

[X-Act]

I agree with Syberia somewhat. Pokemon is a huge paper-rock-scissors game, really.

I'm gonna use paper against his rock, but he might counter me with scissors, so let's play rock instead, but what if he predicts me playing rock and counters with paper? I'll overpredict that and play scissors then... but what if he doesn't predict at all and plays the rock I was predicting that he'll play initially? ...

The only difference is that, in Pokemon, the outcome of playing paper, rock or scissors is not the same, but varies from situation to situation. Basically the best thing to do is to ensure that you don't get fucked up badly even if you mispredict. You might get fucked up, but not too badly.

 

[husk]

In that case, you should probably see wich option offers less risk and more benefit.

That's not prediction.

Though I admit I would probably just bug buzz as it doesn't have a chance of missing and it hits all the the known pokemon hard. If heatran does come out I'll deal with it later.

Husk, if you're playing a good player who knows you can make all of those predictions and is capable of counter-predicting, it's really nothing more than a guessing game on both sides. Each move is most probably just as likely to occur.

If you say so...I never think that my opponent has just made a lucky move if they predict me correctly. Similarly, I never feel that when I predict I'm getting lucky. I admit, there is no hard evidence that assisted my reasoning for my choices and my opponent could've made whatever move he made for reasons that I can not even begin to understand. I'll also accept that prediction is very inconsistant, in one battle you can completely be in someones head and in a battle 5 minutes later with the same player, he can be in yours. You can call it guessing or whatever you'd like but prediction is a skill, maybe to you it's the skill of guessing? When I predict correctly I am making a move based off what I know my opponent will do. When I predict incorrectly what I believe to know is not true. I failed to understand the situation.

 

[Stathakis]

Husk, if you're playing a good player who knows you can make all of those predictions and is capable of counter-predicting, it's really nothing more than a guessing game on both sides. Each move is most probably just as likely to occur.

thanks for wording it better for me

 

[Tarquin]

This last page is really a very on-topic discussion about prediction versus probabilities. Good job =D

So far the bias appears to be pretty neutral. I wonder which will work better in battle conditions...

Thank everybody who has posted so far :)

 

[husk]

Husk, if you're playing a good player who knows you can make all of those predictions and is capable of counter-predicting, it's really nothing more than a guessing game on both sides. Each move is most probably just as likely to occur.

So you're saying a bad or even average player could not see all of those possibilities (or is not capable of counter-predicting...which would probably define a bad or average player)...what does that mean if they can't tell I could air slash if i don't air slash?

You can accept that there are people who are bad at prediction and people who are good. So then why is it unacceptable that even at a high level there is a difference amongst players?

 

[Stathakis]

what he is proposing (I think) is that there are two levels of prediction: bad and good. if you go through the process of thinking what he might bring in / what he might do, you are good. if you select moves randomly, you are bad.

edit: once again my thoughts have been worded better for me (look right below this post)

 

[Chou Toshio]

So you're saying a bad or even average player could not see all of those possibilities (or is not capable of counter-predicting...which would probably define a bad or average player)...what does that mean if they can't tell I could air slash if i don't air slash?

You can accept that there are people who are bad at prediction and people who are good. So then why is it unacceptable that even at a high level there is a difference amongst players?

Because the "bad" players are bad for very specific reasons, usually having to do with lack of knowledge. "I didn't know Salamence's Draco Meteor wipes the floor with Porygon 2!" "I didn't know Infernape's Grass Knot does so much to suicune!" "I didn't know Infernape's HP Ice does so little to Latias!" or even stupid things like not knowing latias outspeeds infernape or that Gyarados can get taunt.

The difference between "good" and "bad" players at prediction basically comes down whether or not the players have an understanding of the capabilities of all the pokemon involved. Notice the phrase "whether or not," it is one or the other-- there is no varying degrees between the two. Of course it's not completely black and white, but I would have to agree with Syb that between 2 players who have an understanding of all the possible outcomes, it is almost a dice roll in terms of just predicting correctly. The key is almost. Of course that's not 100% true, but Syb is right that it is true to an extent.

If my last pokemon is Infernape against your Celebi and Blissey, with 30% HP left and sand blowing, I can can Close Combat or Fire Blast. You can leave celebi in or switch blissey in. If I guess right and make the kill, I won the game. If I guess wrong and miss the kill, LO will kill me before I can kill you, so I lost. That's a pure 50/50. There's no skill there.

Of course the game isn't like that from start to finish, but this "luck" involved in predicting plays a huge part on the game.

 

[muna]

I see deciding which move to make as a process.

1. Assess the situation. What do you know about your opponent's team and the Pokemon that they currently have out?

2. Approximate their skill level. Are they ranked high on the ladder or are they at the bottom?

3. Taking these two aspects into account, approximate what the chances are of them making their most likely move. Make a statement like "There is a 60% chance of them switching to Scizor and a 40% chance of Dragonite staying in".

4. After weighing these odds, compare the risks and rewards that a correct prediction or misprediction can bring to your team.

5. Pick the move based on this and hope for the best as no prediction is 100% accurate (unless the opponent is down to the last Pokemon and it has a choice item).