• Smogon Premier League is here and the team collection is now available. Support your team!

Prediction or Probabilities ?

Tarquin

Tarquin

Pokemon is a game of such diversity, such complexity and such excitement.

In order to counter the huge variety or threats, strategies and general kooky players, there is one defining tactic, a thread that runs through all competitive pokemon battles, which is predicition.

Almost everybody on Smogon is familiar with the concept. If I have a Salamence and my opponent has a Dragonite i play and a Scizor in the wings, the obvious move for me to use if a Dragon Type Move, to kill his Dragonite, who is weak to this attack. However, my opponent might well guess this and switch in his Scizor, which resists my Dragon attack. Therefore, I can use a Fire Attack instead of Dragon attack and kill his Scizor as it switches in.

This is all well and good and works well against inexperienced player. However, once you reach the higher levels of play, which I am familiar with at the moment, your opponent is predicting just as much as you are.

To go back to my previous scenario, as I decide to use a Fire Attack to hit his Scizor, my opponent might predict I will do so and leave Dragonite in, to take minimum damage. However, what if I 'predict' this move and use a Dragon Attack to hit his Dragonite.

And here we see we have moved full circle, now I am using a Dragon Type Move, not because I believe him to be inexperienced, but because I believe him to be so experienced and crafty that he will double bluff me. Now at this point, it becomes almost impossible to know what my opponent will do, because each player is now wrapped up in there own idea of what the other might be thinking. And even supposing my opponent isn't doing these mental Gymnastics, he might just be inexperienced enough to leave Dragonite in anyway.

If you are getting a little confused by this scenario, this will help my point. It is just a 50/50 toss up between each opponent. The player must weight up the odds, playing, not to prediction but to probability, a reason as presume that fragile offensive teams are declining in use, as this is so risky.

Situations such as this occur all the time in competitive battling and they completely undermine the most important of techniques in competitive battling.

So I leave this piece of my mind open to discussion.
Suggested topics are:

How do you handle these situations (such as the one I described in the examples) ?
Do you feel prediction is still as powerful as tool at the height of competitive battling?
What do you, in general, think about what I have raised here?


Thank you :)
Hope for some happy discussion!
 
Prediction is glorified guessing. "Prediction wars" come down to a coin toss for the most part. I usually go for the obvious move initially, then switch it up depending on my opponent's playstyle.
 
It is rather obvious that it's not glorified guessing. If that were the case, we would expect all players to have ~50% win rate, controlling for differences in teams. Put another way, if two players use the same team, their win rate should be identical.

This is not the case. You're using bad statistics here. Just because an event is not 100% doesn't mean it's 50%. Prediction is your ability to determine what they'll do. It's not an always right or always wrong thing, it's a gradient. Yes, they're trying to predict you as well, but the whole point is to be smarter than them (or else you'll be outpredicted).
 
To me, it is better to underpredict than to overpredict. Let's say my Kingdra is in Rain, and can therefore outspeed and destroy a Salamence with Draco Meteor. However, the opponent can shrug it off with a Scizor. I would go for the Draco Meteor anyway, since the Kingdra can live to fight another day if Scizor merely gets dented. If I use Hydro Pump and Salamence stays in, Kingdra is dead.

It might be better to go for a risky move if you're low on HP (i.e. Life Orb will kill you that turn or entry will kill you if you switch back in), and therefore have nothing to lose.
 
Originally, I play safe, and once I figure my opponent's style, I proceed to predict based on that.
 
My general strategy when it comes to prediction is to play conservately until I run into the same situation twice, then predict that my opponent will do the same thing and act accordingly.
 
So far the general conclusion appears to be to play it conservatively. :)

I will be interested to see any fragile offensive players coming in, like myself, as they cannot often afford to play it too safe.
 
In the Salamence vs. Dragonite/Scizor scenario, it does not just become a guessing game of 50% chance because there are other factors that the players should take into account. The two players must weigh the importance of the pokemon that are subjected to harm. Maybe the Scizor player must keep Scizor alive because otherwise Gengar would ruin his team, so he would not risk the Scizor switch and possibly get KO'd. The Salamence player should know that, and act accordingly. The players must know their best move, as well as their opponent's best move.
 
To Peachfuzz:
Although I appreciate your comments, there is no need to dive to deeply into this example, it is very simple and only focused on the matter in hand, Salamence versus Scizor and Dragonite.
 
I make decisions based primarily on what my team needs. If I predicted a Dragonite switch but Dragonite was a huge threat to my team, I'd go for Draco Meteor. This I'd choose over Outrage because I don't need to use it, and over EQ or Fire Blast because they don't hit Dragonite. A degree of randomness helps eliminate the opponent seeing this, but as before, you play randomly but safely.

On a similar token, if Skarmory is a big threat to my Salamence sweep, I might send it in and Fire Blast outright just in the hope of killing it early. This kind of prediction, I find, works best.
 
My decision usually depends on the player. If I'm just on ladder autopilot, I just go for the obvious choice. Even if I make a mistake, against the run of the mill ladderer, I can still eke out a win.

However, if you are playing someone that you know is good, then you need to step it up, think, and watch for patterns of behavior that might help you make your decision.
 
Making large generalizations like this wouldn't work. Each time you make a prediction there are several highly variable things you need to account for. Your opponent's play style is a big one. If he has been playing it safe the whole match, you'll obviously want to take this into account. Another factor is the risk/reward. If you have a chance to KO somebody 100% of the time with a move, but his other pokes resist, you might be tempted to use a move that is SE but is not guaranteed to KO. If you predict right you might score an extra KO, if you predict wrong, you could end up getting KO'd back and possibly lose the match.

There are always factors beyond what is directly in front of you and better players can take advantage of them. Prediction is a skill that you need to have to win matches and is not as simple as "Okay this is a 50/50 chance because they're probably thinking what I'm thinking".
 
Thank you for so many responses!!

Many of you have commented on the opponents playing style, which is indeed a big factor in any prediction, so thank you all for that comment!

As a side note to this, I always find Balanced or Offensive high ranking opponents most unreadable, as they are more willing to take risks or use tactics like luring.

Thank you again!
 
It all depends on the opponent's playstyle. Without any knowledge of their strategy, I usually just go for the obvious prediction. If my guess is wrong, I switch up my strategy.

If I know enough of the opponents' team, I'll usually just use the most neutral move to make sure the switch-in still gets hurt, or read the situation and use a failsafe like, for example, pursuit.
 
I agree with whoever said playing conservatively until you know your opponents style works best. I also find that it pays off to play a little randomly when its usually to early to make predictions. An example from a few days ago was when I led with my SpecsMence and my opponent led with Scizor. The obvious move is to use a fire attack, as most people would. Instead I Draco Meteor'd his switch for a OHKO. (His switch was Swampert) I know that doesn't sound very impressive, but that's just one example.

Hi Tarquin ;)
 
Generally, if you have the advantage, there is no reason to overpredict your opponent. Even if your opponent switches to Scizor (in that example), he still takes a chunk from Dragon Claw / Draco Meteor as well as take Stealth Rock damage.
 
Obi said:
It is rather obvious that it's not glorified guessing. If that were the case, we would expect all players to have ~50% win rate, controlling for differences in teams. Put another way, if two players use the same team, their win rate should be identical.

This is not the case. You're using bad statistics here. Just because an event is not 100% doesn't mean it's 50%. Prediction is your ability to determine what they'll do. It's not an always right or always wrong thing, it's a gradient. Yes, they're trying to predict you as well, but the whole point is to be smarter than them (or else you'll be outpredicted).
Click to expand...

When you are in a true "prediction war," it is a toss-up on who will come out on top. The OP's situation is a great example: MixMence could Outrage against Nite or Fire Blast in anticipation of Scizor's switch-in. If you predict correctly, you are in a huge advantage, because you KOed 1/6 of your opponent's team. If you don't predict correctly, then your opponent has the big advantage. In this case, it IS a toss-up about who will come out on top. My opponent can outpredict my outprediction of his preceived prediction. If I'm in a "prediction war" of equal chances, I randomly pick a move then let it be. Can my opponent prediction my random choice that I don't even know?

I'm not saying that prediction isn't useful. If I'm up against an opponent that only makes the obvious moves (ie Thunderbolt vs Gyarados), then yes, I will predict his unchanging pattern and act accordingly. Most likely however, I will be pinned against an opponent that'll switch it up between obvious and predicted attacks. In the latter case, it is largely a toss-up about who will win that turn of prediction.

Absol is another great case of a toss up. Sucker Punch will bypass its moderate Speed and avoid getting itself hit first. Pursuit will be killer if the opponent switches out to dodge the Sucker Punch. Or maybe Night Slash straight up kill the opponent if it decides to use a status move to avoid Sucker Punch. Superpower could bash in the Steel-type switch-in. Absol is quite frail so it can't afford to mispredict, but there are many variables on what your opponent can do. Your opponent is as much puzzled as you are as to what to do. Yes, this would be a toss-up.

If prediction is that all-powerful, then why don't the top players use those awesome prediction skills for MixMagmortar? It can cause major damage if can anticipate everything right. The problem is that "prediction wars" turn into a glorified game of outsmarting the opponent.

Prediction is definitely useful, but it is certainly not all that it's cracked up to be. I don't feel too much pride after winning a "prediction war," because it could have easily went in favor of my opponent. When it is all said and done, this is just a game of chances.

I wonder how many people will be angered by this post...
 
Unlike some people here who play conservative 'till they can read their opponent, I actually just predict like however I feel. Lately I've not felt like making super complicated predictions(part of that is cause of the team I run but still) and so I just go for the obvious move. However, I do like to try and set up situations where my opponent must choose between switching and avoiding the obvious attack and staying in, as I can't count the number of times where my opponent avoids the obvious attack first time around, but second time tried to be sly and stay in.

That said, a few months ago there was a period where I simply loved going for those super predictions, it was really great especially when you pulled off some amazing hits.

So yea, I actually predict like how I feel on the given day lol.
 
Honestly this game is a basis of probabilities (and even playing completely on auto-pilot and ignoring your opponent) and you could be a completely amazing player if you just played according to those. However no one will take the effort to calculate them every turn and most players would lack the mathematical ability. The thing about pokemon is the factor of the other player and the fact that there is incomplete information. Perhaps you know what his best move from your calculation however he is completely oblivious or he would never make that move. You would have to take that into account and act accordingly. Ah fuck it I'm not qualified enough to explain game theory when I can barely describe what it is. Just guess people it's probably the best choice. =P
 
I agree with obi and JLei2K. To say that it's 50-50 is a real crude reductionist approach. If you're truly good at prediction, you have to calculate your opponent's skill level as well as his/her play style. Do they assume the obvious attack? Do they like to keep pokes at low health for death fodder later?

Furthermore, there are risk and rewards with each move. It could be that your only chance of killing that salamence is on the switch, but you have plenty left to deal with that scizor. Obviously, the wise choice (in this sense) is to go for the mence kill. What are the awards for getting it right? What will happen if you get it wrong? How does killing that threat affect the game in the larger sense? If you don't get the kill, is the damage on the other poke still useful? Are you risking a poke that you absolutely need to win?

My approach is simple, unless you have a good read on your opponent (knowing his/her thought process well, knowing his/her team and play style) go for the move that helps you out the most. Go for the move that helps you out the most in the long run. Take the time to weigh up the consequences and rewards. Note that this does not equate to playing it safe. What is regarded as safe play can often be predictable. When play is predictable you can be easily beaten so in hindsight, it may not be the safe play at all.
 
Predictions important, but i normally play depending on their play style and other pokemon (provided i know them). It may in the end be 50/50 but by understanding the opponents play style and team co-operation, one can determine what the opponent will do easier.
 
It all depends on the stage of the game for me, if all 6 of my opponents' pokemon are revealed, then prediction comes quick and easy. If not, I'd tend to play it safe.
 
To go back to my previous scenario, as I decide to use a Fire Attack to hit his Scizor, my opponent might predict I will do so and leave Dragonite in, to take minimum damage. However, what if I 'predict' this move and use a Dragon Attack to hit his Dragonite.
Click to expand...

I think this is an excellent scenario, I remember playing against a good quality player, who had his Salamence out against my Scizor. I run a specially defensive Scizor. I expected him to expect me to be a good player and either stay in or switch to another Dragon resist i.e. He would expect me to expect a Draco Meteor and Fire Blast instead. So, I switched to a Fire resist (Latias) and got killed by the incoming Draco Meteor. The problem with this scenario was (embarrasingly my team didn't really have anything solid to switch into MixMence early game,) but mostly that it was too early in the game to make rash predictions like that, as I did not have a good read on my opponent. Overprediction early-game can cost you, I find it is best to play conservative early game, this allows you to get a good read on your opponents playstyle and it also allows you to reveal a lot of your opponents team. If you know all of your opponents options, and his mindset, you almost already have the match in the bag.

I feel a lot of people (myself included) guiltily tend to "ladder on autopilot." You tend to revert to your natural playing style, and that can be easily exploited by a good player. If you are facing a very conservative Stall player, use Starmie as a pivot to go to CBTyranitar when they switch in Blissey, you can Pursuit as they switch. If the player is taking more risks, you can opt to Crunch instead, it might even be harder to force their Blissey in in the first place. The number of people that leave their Blissey in on Scizor, expecting a U-turn, to get hit by SuperPower is incredible. You have to engineer a situation into which they feel uncomfortable, it is at this point that people tend to make the descision that feels the most comfortable to them.

If every player played "perfect Pokemon" with regards to prediction, then everyone would mix-up their playstyles to confuse their opponent, reducing prediction to almost a 50-50 level. It is true that the nearer two players skill levels, the nearer prediction is to "glorified guessing." If a player has a significant skill advantage over his opponent, then predictions will not be needed so much, taking control of a battle lessens the need for prediction.
 
Wow, I woke up this morning.. (feelin' fine :P) and my little thread has loads of discussion on it :O

It's really interesting seeing how everybody feels about this and everybodies different views and playing styles. I also get the impression we have a huge range of players now, from stall to offensive.

In particular I was charmed here...
Luphrous said:
So yea, I actually predict like how I feel on the given day lol.
Click to expand...

This is I think the reason I have stopped OU laddering (that and the pressure Pressure PRESSURE). It seems so bizarrely free to me to be able to predict how you want and accept the consequences, without risking vital points :)

Thank you to everybody who has posted!
 
You must log in or register to reply here.
Back
Top