Your Turn

The Pokémon's Turn

You always get the first turn in the Safari Zone, but on the Pokémon's turn, two things happen.

First, the game will check if either of the angry and eating counters is nonzero. If so, then a message saying "Wild [Pokémon] is angry!" or "Wild [Pokémon] is eating!" as appropriate is shown and the counter is decreased by one. If the angry counter is decreased to zero this way, the Pokémon's catch rate will also be reset to its initial catch rate, regardless of how it has been modified in the battle before this point; note that this last bit does not happen when a Pokémon stops eating, nor when the angry counter is reset to zero because you threw a bait.

After this, the game will perform a calculation to determine whether the Pokémon will run away on this turn. The run chance depends only on which state the Pokémon is in - angry, eating or neutral - but not on how many times you've thrown rocks/bait in any way: a Pokémon that you've thrown five rocks at followed by one bait will be exactly as happy to stick around as one that you threw a bait at on the first turn. Note that the Pokémon's actual current state does not necessarily correspond to the state indicated by the message that was just shown, since the message indicates only that the counter in question was nonzero before it was subtracted from. This also means that if you throw a rock or bait and the random number generated is 1, you will see an angry/eating message, but the Pokémon will in fact be back in its neutral state before even the run check is performed.

The run calculation itself goes as follows:

1. Make a variable X equal to the low byte (i.e. the remainder if you divide by 256) of the Pokémon's Speed (not the base Speed of the species, but the individual's actual Speed).
2. Double X.
3. If the outcome is greater than 255 (i.e. if the Pokémon's Speed was 128 or more), the Pokémon automatically runs. Skip the rest of the procedure.
4. Modify X again if the Pokémon is in a non-neutral state:
   • If the Pokémon is angry, double X again (if it becomes greater than 255, make it 255 instead).
   • If the Pokémon is eating, divide X by four.
5. Generate a random number R between 0 and 255 inclusive.
6. If R is less than X, the Pokémon runs away.

All in all, this means that so long as (the low byte of) the Pokémon's Speed is less than 128 (which it always will be in the actual game - the highest Speed any Pokémon actually found in the Safari Zone can have is 75), the chance that it will run is 2*Speed/256 if it's in a neutral state, min(255, 4*Speed)/256 if it's angry, or int(Speed/2)/256 if it's eating.

Crucially, since this is the actual individual Speed and not the base Speed of the species, lower-leveled individuals are less likely to run. While Scyther at level 25 or 28 have around or above a 50% chance of running every turn in a neutral state, for instance, Yellow's level 15 Scyther are considerably easier to catch, with only a 32% chance of running in a neutral state at the most. Thus, perhaps the best piece of strategic advice for the Safari Zone is to go for the lowest-leveled possible version of your desired Pokémon, given the lower-leveled version isn't unacceptably rare.

Wait, What?

Good question. If you don't care about getting an intuitive grasp on why this is true, feel free to skip to the Safari Zone calculator.

Here's the thing. The entire Safari Zone experience basically simplifies to a game where you and the Pokémon alternate turns, with each of you having a given chance of "winning" on each of your turns (you win if you catch the Pokémon, while the Pokémon wins if it runs). When you throw bait or a rock, however, you do that instead of throwing a ball on that turn, while the Pokémon will continue to have a chance of running on every single one of its turns; essentially, you are forgoing one of your turns (attempts to "win") in exchange for a later advantage.

What is that later advantage, then, and is it worth losing that turn? Well, in the case of a rock, you double your chances of winning (catching the Pokémon) for up to four subsequent turns - but you also double the Pokémon's chances of winning (running away), and because you used up your turn throwing the rock, it's the Pokémon that has the next move.

You can hopefully see how that's not really a recipe for success. However, it's not quite as bleak as it appears, thanks to the one place where the simplification breaks down: you have a limited number of Safari Balls. A rock, by doubling both yours and the Pokémon's chances of winning each turn, will shorten the average duration of the battle. Thus, if you have sufficiently few balls and the Pokémon has a sufficiently low catch rate and Speed, to the point that in an average battle against it you'd run out of balls before either catching it or it running, throwing a rock and shortening the battle so your balls will last can actually be worth it, even at the aforementioned cost. For instance, if you only have one Safari Ball left, then you can either throw that one ball with a regular catch rate or throw some rocks first, which will make your single ball much more likely to be effective once you do throw it; you'll only get one attempt to catch it either way. The risks will still outweigh the benefits if the Pokémon is pretty speedy, since then it will be likely to run before you can actually throw the ball at all, but for a sufficiently slow target (for a single Safari Ball, the highest Speed where a rock will be worth it is 25 or so), rocks can be a good idea when you don't have a lot of Safari Balls left.

Throwing multiple rocks can also help, at least in theory, since more rocks will continue to double your chances of catching the Pokémon without raising the running chance further. Primarily, in many of those situations where a lack of Safari Balls means one rock is a good idea, two (or possibly three) rocks improve your chances even further, though the range of situations where this works is even narrower than for one rock. Technically multiple rocks can also help in general for Pokémon with very low Speeds and low catch rates - however, that's low Speeds as in single digits, and no Pokémon that fit the bill are actually found in the Safari Zone, making that point kind of moot. Otherwise, if you have plenty of balls to spare, the free angry turns they usually get to run away before you even start trying to catch them just result in a disadvantage you can't make up for.

What about bait? Bait is immediately somewhat more promising than rocks, since it halves your chance of "winning" but quarters the Pokémon's. However, bait also differs from rocks in that the catch rate doesn't go back to normal after the Pokémon stops eating, and just like rocks shorten the duration of the battle, bait prolongs the battle - it makes both parties less likely to win on subsequent turns. And the longer the battle goes on, the more the up-to-four turns (remember, the counter is decreased before the run check) that the Pokémon is actually less likely to run diminish in significance compared to all the turns after the Pokémon stops eating, when it will still have a lowered catch rate but a regular chance of running. That's besides the fact that again you must forgo a turn to throw the bait in the first place. In fact, as it turns out this makes bait wholly useless: there is not even in theory a Speed/catch rate combination for which bait will do you any good.

Where does this leave that especially promising-looking "rocks, then bait" strategy? Ultimately, it's stuck in the same rut rocks are: it's normally only useful for Pokémon with such ludicrously low Speed that they don't actually exist in the Safari Zone, and unfortunately, while rocks at least have a niche when you're running low on balls, you're always going to be better off just throwing however many rocks you're going to throw and then throwing your ball than throwing the rocks and then wasting your time on bait if you only have a couple of balls left. This strategy requires wasting several turns without throwing any balls, during some of which the Pokémon will have an increased chance of running, and to make matters worse, if the number of angry turns generated is one, you're going to lose even the rock's advantage and end up with the bait's lowered catch rate after all that preparation. It just kills it.

So, again, in nearly every case the best strategy is to just throw balls and hope you get lucky. That is, however, assuming that what you want to maximize is your chance of success per encounter: since rocks shorten the battle and make for fewer Safari Balls required, rocks may actually save you time and money.

The Safari Zone calculator below includes a variety of strategies, despite their mostly limited usefulness; play around with it if you think you might go with a different one.