A Beginner’s Guide to Prediction
Someone said to me that they felt like prediction is something impossible to learn. While I am certainly no fortune-teller of Pokemon, I can offer a few heuristics I use in my own battles to outpredict my opponent. These are probably more-or-less obvious, at least at an intuitive level, to veteran players. Still, I think it is good to try to explicate them a little and lay them out clearly for others to benefit.
The first principle is that prediction is no more than guessing. A skilled opponent does not have special knowledge of what move you are going to play, they simply give a guess based on the circumstances.
I want to define right now the concepts of early game, and late game, which are vital for prediction. The definitions will be intuitive rather than precise, as fits the flavor of this article. The early game is where nothing of consequence has fainted yet. It is not a time for incredible predictions, but where you should be in the scouting mindset (discussed in more detail below). This will allow you to predict accordingly later. It is a time where you start enacting your strategy. Each player sends out various counters to try to stop the other person. The late game is a time when five or more Pokemon on each team are known, or when the overarching strategies of each team have been played out. This is a time to think carefully about finishing the game and avoid blunders.
(1) THINK before you make a move. It’s so tempting to press that Use button right away, but how often do you press Cancel right away, hoping your opponent hasn’t moved yet? Oftentimes in late game, your opponent is essentially checkmated if you play the right sequence of moves. However, these moves are not always obvious. Just like in Go or Chess, think carefully before putting your hand on the piece. Remember: at each step there are only at most 9 things you can do: Use one of your four moves or switch to one of your five other Pokemon.
(2) Now, this may sound obvious, but ACT ON WHAT YOU THINK YOUR OPPONENT WILL DO. That is basically the definition of prediction, but so often people fail to do this. You should try to put yourself in the opponent’s shoes, and figure out what is his most reasonable option. For example:
a. When you have a super-effective move (especially a low power one) against your opponent that has already been revealed or is commonly used on your Pokemon, the opponent will likely switch. So it might be advisable to choose a move that screws over the counter (in late game) or a general coverage move like STAB Surf (in early game). The novice player often assumes or hopes that the opponent will stay in during that situation.
b. However, beware of exceptions. For example, and especially with offensive teams, the opponent will sometimes leave the Pokemon in as a late-game sacrifice. It will rarely happen if that Pokemon counters one of your threats or is necessary for the opponent to win. It will often happen if the opponent carries fast Pokemon that can safely “revenge kill” after the free switch afforded by the sacrifice. When there are entry hazards like Spikes and Stealth Rock the opponent is much more likely to sacrifice.
(3) USE THE IDIOSYNCRATIC TENDENCIES OF YOUR OPPONENT to your advantage. It may be difficult at first, so you can try to take careful note of how the opponent reacts in each situation. Pencil and paper may be helpful for this. If an enemy Swampert happens to face your Celebi, and the opponent switched to Heatran every time in the past to absorb the Grass Knot, then you might be able to predict that and switch out your favorite Heatran counter right away, leading you to yet another favorable matchup.
(4) On the other side of the coin, a good way to avoid being outpredicted is to RID YOURSELF OF PREDICTABLE TENDENCIES. In basketball players are trained to make their moves with changes of direction and speed. The same is true of Pokemon.
(5) KNOW THE METAGAME. You should know common sets and spreads, whether or not you can OHKO, etc.
a. It's important to know relative speeds of different Pokemon. These can be easily checked on the Smogon site, and using the usage statistics or moveset analysis one can try to guess what spreads and natures are commonly run. Of course, you must understand the huge difference EVs make, so don't just look at the base Speed stat. Your Clefable (base 60) is probably never going to outrun that Cacturne (base 55), which tends to run high speed.
b. For knowledge about other aspects of the metagame, I recommend Legacy Raider's monthly metagame analysis.
c. As Seven Deadly Sins suggested, a good crutch is to run calculators in battle. This will allow you to determine whether you would score a KO, for example. It can also be used to deduce information about enemy EV spreads/natures based on how hard the enemy hits against your (known) stat distribution and how hard you hit against the enemy. The standard calculator is Libelldra's. Steinhauser has apparently developed a calculator that circumvents some of the annoying or tedious aspects of Libelldra's; it can be found here. Although I have never used it personally due to compatibility issues, it seems promising.
This knowledge is key to making good predictions, and takes a great deal of time and experience.
(6) UNDERSTAND THE OPPONENT’S STRATEGY. Obviously you should try to understand the opponent’s overall goal. This will help greatly in predicting how he will behave. However, there are also more subtle strategies. For example, if an Infernape switches in to your Swampert on a predicted Ice Beam, then you can assume it carries Grass Knot. It’s also good to recognize Pokemon holding Choice Scarf in this way, if the opponent switches in a naturally slower Pokemon that is vulnerable to one of your attacks. Another useful thing to do is scout for counters. If you play Infernape, the Infernape counter will likely show itself. Now you know who will likely switch in if you try to send out Infernape again, and can play accordingly. That is, of course, the reason U-Turn is so popular. While thinking about this, you should realize what information the opponent knows about your team and individual Pokemon. It matters whether or not he knows that Blissey has Counter, or whether you pack a Ghost.
(7) CONSIDER RISK. When you figure out what you think your opponent will do, it’s sometimes not advisable to just act assuming that your guess is right. You have to always remember the risk if you are wrong and measure it against the reward if you are right, and the likelihood you are right. Try not to take too many big risks, especially unnecessary ones. Of course, don't always choose the move with lowest risk, as that violates principle (4). (Note: More concrete examples are needed here. It's not clear from the exposition what are "good risks" and "bad risks").
This post is not so much meant as a definitive guide, but rather to 1) serve as a reference for beginners, and 2) encourage discussions about the subject of prediction, outside of outlets like warstories. I encourage all of you to share your prediction experiences, personal principles and observations. I hope that you all leave with a better sense of that mysterious and important term, PREDICTION, and look forward to hearing your wisdom on this matter!
EDITS:
I changed "Reduce Risk" to "Consider Risk." That makes more sense as one should not always seek to reduce risk.
Added a suggestion of SDS to mention calculators.
Removed references to the "mid-game," an artifact of an earlier idea I decided not to implement.
Someone said to me that they felt like prediction is something impossible to learn. While I am certainly no fortune-teller of Pokemon, I can offer a few heuristics I use in my own battles to outpredict my opponent. These are probably more-or-less obvious, at least at an intuitive level, to veteran players. Still, I think it is good to try to explicate them a little and lay them out clearly for others to benefit.
The first principle is that prediction is no more than guessing. A skilled opponent does not have special knowledge of what move you are going to play, they simply give a guess based on the circumstances.
I want to define right now the concepts of early game, and late game, which are vital for prediction. The definitions will be intuitive rather than precise, as fits the flavor of this article. The early game is where nothing of consequence has fainted yet. It is not a time for incredible predictions, but where you should be in the scouting mindset (discussed in more detail below). This will allow you to predict accordingly later. It is a time where you start enacting your strategy. Each player sends out various counters to try to stop the other person. The late game is a time when five or more Pokemon on each team are known, or when the overarching strategies of each team have been played out. This is a time to think carefully about finishing the game and avoid blunders.
(1) THINK before you make a move. It’s so tempting to press that Use button right away, but how often do you press Cancel right away, hoping your opponent hasn’t moved yet? Oftentimes in late game, your opponent is essentially checkmated if you play the right sequence of moves. However, these moves are not always obvious. Just like in Go or Chess, think carefully before putting your hand on the piece. Remember: at each step there are only at most 9 things you can do: Use one of your four moves or switch to one of your five other Pokemon.
(2) Now, this may sound obvious, but ACT ON WHAT YOU THINK YOUR OPPONENT WILL DO. That is basically the definition of prediction, but so often people fail to do this. You should try to put yourself in the opponent’s shoes, and figure out what is his most reasonable option. For example:
a. When you have a super-effective move (especially a low power one) against your opponent that has already been revealed or is commonly used on your Pokemon, the opponent will likely switch. So it might be advisable to choose a move that screws over the counter (in late game) or a general coverage move like STAB Surf (in early game). The novice player often assumes or hopes that the opponent will stay in during that situation.
b. However, beware of exceptions. For example, and especially with offensive teams, the opponent will sometimes leave the Pokemon in as a late-game sacrifice. It will rarely happen if that Pokemon counters one of your threats or is necessary for the opponent to win. It will often happen if the opponent carries fast Pokemon that can safely “revenge kill” after the free switch afforded by the sacrifice. When there are entry hazards like Spikes and Stealth Rock the opponent is much more likely to sacrifice.
(3) USE THE IDIOSYNCRATIC TENDENCIES OF YOUR OPPONENT to your advantage. It may be difficult at first, so you can try to take careful note of how the opponent reacts in each situation. Pencil and paper may be helpful for this. If an enemy Swampert happens to face your Celebi, and the opponent switched to Heatran every time in the past to absorb the Grass Knot, then you might be able to predict that and switch out your favorite Heatran counter right away, leading you to yet another favorable matchup.
(4) On the other side of the coin, a good way to avoid being outpredicted is to RID YOURSELF OF PREDICTABLE TENDENCIES. In basketball players are trained to make their moves with changes of direction and speed. The same is true of Pokemon.
(5) KNOW THE METAGAME. You should know common sets and spreads, whether or not you can OHKO, etc.
a. It's important to know relative speeds of different Pokemon. These can be easily checked on the Smogon site, and using the usage statistics or moveset analysis one can try to guess what spreads and natures are commonly run. Of course, you must understand the huge difference EVs make, so don't just look at the base Speed stat. Your Clefable (base 60) is probably never going to outrun that Cacturne (base 55), which tends to run high speed.
b. For knowledge about other aspects of the metagame, I recommend Legacy Raider's monthly metagame analysis.
c. As Seven Deadly Sins suggested, a good crutch is to run calculators in battle. This will allow you to determine whether you would score a KO, for example. It can also be used to deduce information about enemy EV spreads/natures based on how hard the enemy hits against your (known) stat distribution and how hard you hit against the enemy. The standard calculator is Libelldra's. Steinhauser has apparently developed a calculator that circumvents some of the annoying or tedious aspects of Libelldra's; it can be found here. Although I have never used it personally due to compatibility issues, it seems promising.
This knowledge is key to making good predictions, and takes a great deal of time and experience.
(6) UNDERSTAND THE OPPONENT’S STRATEGY. Obviously you should try to understand the opponent’s overall goal. This will help greatly in predicting how he will behave. However, there are also more subtle strategies. For example, if an Infernape switches in to your Swampert on a predicted Ice Beam, then you can assume it carries Grass Knot. It’s also good to recognize Pokemon holding Choice Scarf in this way, if the opponent switches in a naturally slower Pokemon that is vulnerable to one of your attacks. Another useful thing to do is scout for counters. If you play Infernape, the Infernape counter will likely show itself. Now you know who will likely switch in if you try to send out Infernape again, and can play accordingly. That is, of course, the reason U-Turn is so popular. While thinking about this, you should realize what information the opponent knows about your team and individual Pokemon. It matters whether or not he knows that Blissey has Counter, or whether you pack a Ghost.
(7) CONSIDER RISK. When you figure out what you think your opponent will do, it’s sometimes not advisable to just act assuming that your guess is right. You have to always remember the risk if you are wrong and measure it against the reward if you are right, and the likelihood you are right. Try not to take too many big risks, especially unnecessary ones. Of course, don't always choose the move with lowest risk, as that violates principle (4). (Note: More concrete examples are needed here. It's not clear from the exposition what are "good risks" and "bad risks").
This post is not so much meant as a definitive guide, but rather to 1) serve as a reference for beginners, and 2) encourage discussions about the subject of prediction, outside of outlets like warstories. I encourage all of you to share your prediction experiences, personal principles and observations. I hope that you all leave with a better sense of that mysterious and important term, PREDICTION, and look forward to hearing your wisdom on this matter!
EDITS:
I changed "Reduce Risk" to "Consider Risk." That makes more sense as one should not always seek to reduce risk.
Added a suggestion of SDS to mention calculators.
Removed references to the "mid-game," an artifact of an earlier idea I decided not to implement.
