Of the Safari Zone (Or, curse you Chansey!): A strategy guide

[Cubes]

Um, those probabilities ARE 'what happens 50%' of the time.

On a specific encounter, what has a 50% chance of occuring? When using baits, the sequence Bait>Ball>Ball>Ball>Bait>Ball>Ball has a 50/50 chance of being acted out fully. The probability at catching it at this point is 8.26%.

For balls, it's not exactly 50%, but there's a 55% chance of the sequence Ball>Ball being acted out fully. The probability at catching it at this point is 7.29%.

From here, it's already pretty clear that the bait method wins out over just balls, even with the slightly off percentages. Taking this fully, you get about 7.57% catch probability 50% of the time.

To continue on the source by heated:

125 * 100 / 1275 = 9,80, rounded down to 9.

9 / 4 = 2,25, rounded down to 2.

5 * 2 = 10.

So 100-(10+1)=89% chance of escaping. This is probably the source OP used. It seems realistic, but once again fails to mention a source =/

I had actually conjectured the escape rates based on the formula bulbapedia gave. When I checked these against that website, they ended up being correct. Whether both me and that site are actually true is something else.

[heated]

125 * 100 / 1275 = 9,80, rounded down to 9.

9 / 4 = 2,25, rounded down to 2.

5 * 2 = 10.

So 100-(10+1)=89% chance of escaping. This is probably the source OP used. It seems realistic, but once again fails to mention a source =/

Why is it 100-(10+1)? It should be 100-10, right? Random number from 0 to 99 less than 10 allows for 0-9.

The first chart has been modified to take in account the actual average amount of turns bait lasts. I would like to know how someone made it to 11%, could you also post your calculations? The middle chart is the classic balls-only method, a method where the only advantage is being easier to calculate the (lower) odds. The last chart reveals the odds when bait takes effect the turn after it's thrown, which would make bait absolutely pointless as you can see.

Again, due to being unable to predict accurately when the chansey will stop eating ahead of time, the trainer will throw balls until seeing the message "chansey is watching closely" and then throw bait again. This means a turn of the chansey not eating passes when throwing a ball, thus giving it a 55% chance of staying that turn.

The 11.62% is calculated with the following changes to your bait/ball/ball pattern: 90% stay on eating turns, pattern of bait/eating, ball/eating, ball/eating, ball/eating, ball/watching

Sum the series (up to 30 balls or less)

On a specific encounter, what has a 50% chance of occuring? When using baits, the sequence Bait>Ball>Ball>Ball>Bait>Ball>Ball has a 50/50 chance of being acted out fully. The probability at catching it at this point is 8.26%.

For balls, it's not exactly 50%, but there's a 55% chance of the sequence Ball>Ball being acted out fully. The probability at catching it at this point is 7.29%.

From here, it's already pretty clear that the bait method wins out over just balls, even with the slightly off percentages. Taking this fully, you get about 7.57% catch probability 50% of the time.

Are you serious? That's like saying if you flip coins until you get tails, the average number of heads you will get will be 1 at 50% is .5 heads on average.

However, it is 1 coin on average. via (chance of getting that many heads and then 1 tails for each number of heads, and then summing the series)

[Abblot]

God. My head.

[Cubes]

Again, due to being unable to predict accurately when the chansey will stop eating ahead of time, the trainer will throw balls until seeing the message "chansey is watching closely" and then throw bait again. This means a turn of the chansey not eating passes when throwing a ball, thus giving it a 55% chance of staying that turn.

We have no idea when the escape check is made. It's possible that on the turn eating would end, the escape check is calculated with the eating, and then the pokemon goes back to watching carefully.

Are you serious? That's like saying if you flip coins until you get tails, the average number of heads you will get will be 1 at 50% is .5 heads on average.

However, it is 1 coin on average. via (chance of getting that many heads and then 1 tails for each number of heads, and then summing the series)

You're right, I was wrong about that. But I'm not sure what you're saying is completely right. This whole thing had been confusing me because there are 2 probabilities at work here- the average amount of balls per Chansey encounter, and the average amount of balls per Chansey capture. After searching around the internet for a little while, I think I've found an elegant method to determining these 2.

For just balls:

How many balls on average before catching 1 Chansey?

x = (4.7/100) (1) + (95.3/100) (1 + x) = 21.28 balls on average to capture Chansey.

How many balls on average before Chansey flees?

x = (1) (45/100) + (55/100) (1 + x) = 2.22 balls per Chansey encounter.

Putting these 2 together gets us 21.28/2.22 = 9.59 Chanseys per 1 Chansey caught.

For bait balls:

How many bait balls on average before catching 1 Chansey?

x = (2.4/100) (1) + (97.6/100) (1 + x) = 41.67 ball on average.

How many bait balls on average before Chansey flees? Now here is where it gets a little tricky for me.

x= (11/100) (0) + (89/100) ((11/100) (1) + (89/100) (1+ x)) = 4.81 balls per encounter.

I put 0 in the first term because there's the chance Chansey will flee after the first bait, with no balls being used.

2 things- I'm not sure if I initially set up the equation correctly to take into account that first turn flee (I think I did). Second, I'm not sure how to account for bait ending after 4 balls on average.

[heated]

We have no idea when the escape check is made. It's possible that on the turn eating would end, the escape check is calculated with the eating, and then the pokemon goes back to watching carefully.

Unless you have any evidence to the contrary, Bubblepedia implies that the escape check is done based on what the pokemon is doing that turn.

Also, both Bubblepedia and the other site agree on "At the end of a turn, a random number from 0 to 99 is generated, and is compared to 5 times its "modified escape factor": 1275/100 of the escape rate (rounded down), which is doubled if the Pokémon is "angry" or quartered (rounded down) if the Pokémon is "eating". If the random value is less, the Pokémon escapes."

For Chansey, a random number from 0 to 99 is compared to 10 when Chansey is eating. If it is less than that, it escapes. (that's a .1 chance)

Why do you guys keep using .11 as your chance of escape?

[MrMasonMan]

Although I am enjoying reading the discussions you guys are having I just wanted to know what you all believe to be the best chance of catching a Chansey right now.

I used to just throw balls and I'd say I caught about a quarter of the Chanseys I encountered, while with bait ball, then rebait I've caught 1 out of at least 30. Obviously the sample sizes are way too small.

So there's 4 methods I know of, I would just like to know what both of you think is best:

Balls only

Bait first then balls the rest of the way

Bait, ball, then rebait when they start watching, then continue to ball

Same as before, but rebait every time the Chansey starts watching again.

The tables are really nice though, I'd hope in the future to be able to make something that you can input all the data in and each turn it will tell you the best move (taking into account when it starts watching, how many balls you have left, etc).

[heated]

^ Rebait when watching is the best we got so far, with an estimated 11.6% chance of catching one each encounter. (the exact % is currently being disputed but you can use that as an estimate proly)

[Cubes]

Unless you have any evidence to the contrary, Bubblepedia implies that the escape check is done based on what the pokemon is doing that turn.

I would say that Bulbapedia implies the opposite, actually. If we take the start of a new turn to be when your options come up, and the end to be seeing whether the pokemon is watchin/eating, or flees, then when it says "At the end of a turn ....which is doubled if the Pokémon is "angry" or quartered (rounded down) if the Pokémon is 'eating'", it factors in the state that was factored for the catch rate. Unless you mean to say that both the catch rate and escape rate are reverted to normal at the start of the first turn, then they both are affected by bait, and then it reverts to waiting. Honestly, it doesn't seem to imply either way.

As for the escape rates-

We know that the "modified escape factor" HAS to be less than 10; any pokemon at 10 would always flee when hit by a rock- since 5x2x10= 100, and every generated number would result in fleeing. This can't be true since it IS possible for Chansey to stay after throwing a rock.

Now, we also know that this "modified escape factor" is a whole number, since bulbapedia says it's rounded down. So the highest possible "modified escape factors" would be either 8 or 9.

If we go by 9, then 9x5 = 45, meaning there's a 45% chance it will flee. Bait quarters this number, bringing it down to 11.25. So the escape rate is now either 12% or 11%, depending on whether 11.25 is rounded down or not.

If we go with 8 as our "modified escape factor", we get 8x5 = 40, which, when under the influence of bait, gives us a value of 10, or a 10% chance of fleeing.

Now, we don't know for sure which is the proper "modified escape factor" for chansey, tauros, etc. I just went with 9 because it makes them the hardest to catch. I hope this clears it up a bit.

Although I am enjoying reading the discussions you guys are having I just wanted to know what you all believe to be the best chance of catching a Chansey right now.

I used to just throw balls and I'd say I caught about a quarter of the Chanseys I encountered, while with bait ball, then rebait I've caught 1 out of at least 30. Obviously the sample sizes are way too small.

So there's 4 methods I know of, I would just like to know what both of you think is best:

Balls only

Bait first then balls the rest of the way

Bait, ball, then rebait when they start watching, then continue to ball

Same as before, but rebait every time the Chansey starts watching again.

The tables are really nice though, I'd hope in the future to be able to make something that you can input all the data in and each turn it will tell you the best move (taking into account when it starts watching, how many balls you have left, etc).

Figuring out the numbers has been a sticky work in progress, but in essence, bait>ball should net you a chansey about 1 encounter fewer than just balls.

[heated]

I would say that Bulbapedia implies the opposite, actually. If we take the start of a new turn to be when your options come up, and the end to be seeing whether the pokemon is watchin/eating, or flees, then when it says "At the end of a turn ....which is doubled if the Pokémon is "angry" or quartered (rounded down) if the Pokémon is 'eating'", it factors in the state that was factored for the catch rate. Unless you mean to say that both the catch rate and escape rate are reverted to normal at the start of the first turn, then they both are affected by bait, and then it reverts to waiting. Honestly, it doesn't seem to imply either way.

It freaking says it is compared based on whether it is eating or not. How are you confused by this? Yes, both rates are reset at the beginning of each turn. (if applicable)

We know that the "modified escape factor" HAS to be less than 10; any pokemon at 10 would always flee when hit by a rock- since 5x2x10= 100, and every generated number would result in fleeing. This can't be true since it IS possible for Chansey to stay after throwing a rock.

I was talking about the escape chance. It should have been clear if you had read the math.

Now, we also know that this "modified escape factor" is a whole number, since bulbapedia says it's rounded down. So the highest possible "modified escape factors" would be either 8 or 9.

If we go by 9, then 9x5 = 45, meaning there's a 45% chance it will flee. Bait quarters this number, bringing it down to 11.25. So the escape rate is now either 12% or 11%, depending on whether 11.25 is rounded down or not.

If we go with 8 as our "modified escape factor", we get 8x5 = 40, which, when under the influence of bait, gives us a value of 10, or a 10% chance of fleeing.

Now, we don't know for sure which is the proper "modified escape factor" for chansey, tauros, etc. I just went with 9 because it makes them the hardest to catch. I hope this clears it up a bit.

Upokecenter says Chansey's escape rate is 125 but we'll ignore that. The highest modified escape factor before factoring in pokemon state is 9. Multiplying by 5 is the last operation before comparing. 9 becomes 2 (not 2.25) after eating because it's rounded down. Then it is multiplied by 5 to compare.

Or, floor(9/4)*5=10% escape chance

[Cubes]

It freaking says it is compared based on whether it is eating or not. How are you confused by this? Yes, both rates are reset at the beginning of each turn. (if applicable)

How are YOU confused by what I am saying? My question is this- when a pokemon would revert to watching at the end of the turn, does the escape check factor in eating (like the catch check did) or not? You say it doesn't, but as far as I can tell, this hasn't been addressed anywhere.

Upokecenter says Chansey's escape rate is 125 but we'll ignore that. The highest modified escape factor before factoring in pokemon state is 9. Multiplying by 5 is the last operation before comparing. 9 becomes 2 (not 2.25) after eating because it's rounded down. Then it is multiplied by 5 to compare.

Or, floor(9/4)*5=10% escape chance

Ah I see. It factors in eating/angry into calculating the "modified escape factor", rather than applying it afterwards. A little hard to see on bulbapedia at first.

[Bartolomew]

Very very usefull guide. Thank u a lot. I ve catch all Pokemon in Saffari Zone... Now going for lucky egg :o

[heated]

How are YOU confused by what I am saying? My question is this- when a pokemon would revert to watching at the end of the turn, does the escape check factor in eating (like the catch check did) or not? You say it doesn't, but as far as I can tell, this hasn't been addressed anywhere.

Please explain your question. The way you have phrased it seems pretty confusing to me.

Also, correction for higher catch chance on watching turns for the bait/ball/ball method.

Edit: adjusted for catch probabilities assuming all balls have failed up to that turn. (ball/ball/ball adjusted as well)

So bait/ball/ball is now about 26% better than ball/ball/ball.

[Cubes]

Please explain your question. The way you have phrased it seems pretty confusing to me.

The turn when bait ends, it says the pokemon is watching. Was the escape rate 45% or 10%? You keep saying it's 45%, but I haven't seen anything definitive either way. I had always though it was 10%, because the catch chance is affected by bait that turn.

I'm asking about the sequence that happens like this.

End of turn 3- "The pokemon is eating"

Throw Ball in turn 4- "Oh, the pokemon was almost caught.

Possibility 1: Bait check - Bait runs out, revert to watching.

Escape check- The pokemon stays.

Possibility 2: Bait check - Bait runs out, revert to watching.

End of turn 4- "The pokemon is watching carefully."

I'm talking about that transition. I've always thought it was possibility 2, which means the escape check is at 10%. You talk definitively as if it's possibility 1, which means the escape check is at 45%. I haven't seen anything definitive for either possibility.

[heated]

The turn when bait ends, it says the pokemon is watching. Was the escape rate 45% or 10%? You keep saying it's 45%, but I haven't seen anything definitive either way. I had always though it was 10%, because the catch chance is affected by bait that turn.

I'm asking about the sequence that happens like this.

End of turn 3- "The pokemon is eating"

Throw Ball in turn 4- "Oh, the pokemon was almost caught.

Possibility 1: Bait check - Bait runs out, revert to watching.

Escape check- The pokemon stays.

Possibility 2: Bait check - Bait runs out, revert to watching.

End of turn 4- "The pokemon is watching carefully."

I'm talking about that transition. I've always thought it was possibility 2, which means the escape check is at 10%. You talk definitively as if it's possibility 1, which means the escape check is at 45%. I haven't seen anything definitive for either possibility.

End of turn 3- "The pokemon is eating"

Possibility 1: Bait check - Bait runs out, revert to watching.

Throw Ball in turn 4- "Oh, the pokemon was almost caught.

Possibility 2: Bait check - Bait runs out, revert to watching.

Escape check- The pokemon stays.

Possibility 3: Bait check - Bait runs out, revert to watching.

End of turn 4- "The pokemon is watching carefully."

It says that the pokemon will be eating for some number of turns, so I assume that eating is checked between turns.

[Cubes]

End of turn 3- "The pokemon is eating"

Possibility 1: Bait check - Bait runs out, revert to watching.

Throw Ball in turn 4- "Oh, the pokemon was almost caught.

Possibility 2: Bait check - Bait runs out, revert to watching.

Escape check- The pokemon stays.

Possibility 3: Bait check - Bait runs out, revert to watching.

End of turn 4- "The pokemon is watching carefully."

It says that the pokemon will be eating for some number of turns, so I assume that eating is checked between turns.

I see what you mean. It doesn't make a lot of sense, to me at least, that they would force a turn of regular catch and escape chances. To me, it makes more sense for them to let players know that the bait has reset. Essentially, possibility 3 makes more sense than possibility 1 to me, but I can only guess.

[KnightofChaos]

This ought to help my Lucky Egg hunting... thanks!

The Repel tip seems worth trying out. Gets rid of all the Paras and low level Nidos.

[LightningChaser]

This will help me get my lucky egg, and dragon scales, alot faster! And to get rid of those crazy Nidos.

[TheMaster42]

I've been sampling some numbers regarding some of the questions in the thread. The sample size could be larger, but it's pretty safe to make a few assumptions already.

First turn, I throw bait. I measure how many Chanseys flee on the first turn. So far I'm at:

• Number of Chanseys stay: 24
  
• Number of Chanseys fled: 2
  

Small sample size, but it's already clear that the chance of a Chansey fleeing with bait is much more like 10%, and astronomically unlikely to be 45%. So we can figure that the stuff about "fleeing decided before commands are chosen" is incorrect/misleading.

The next set of data I've collected is the chance for a Chansey to stay after I've used bait. On the second-plus turn, I throw balls until the "watching carefully" message reappears. These are the turns up to, but NOT including after, a "watching carefully message". To put another way, on the second turn, I throw a ball and I measure whether it stays or flees - once the "watching carefully" message appears, I stop measuring this category.

• Rounds Chansey stayed: 84
  
• Rounds Chansey fled: 20
  

Even with this small sample, you can see the chance of bait rounds being a global 10%-ish is also astronomically unlikely. We can assume one of two things:

1. The chance for Chansey to stay is not actually around 10%, and is more like 20% - 25%. OR
   
2. The chance for Chansey to stay when the bait runs out does not stay at 10%.
   

I believe the second option is extremely more likely the case. First, based on the first set of data, the first bait round is probably closer to 10% than 25% (a larger sample size would be desirable, however). Second, we've seen nothing anywhere that says the chance of fleeing while under bait increases/changes while the bait effect lasts in gen 3.

It seems plausible that what's happening is, the second set of data includes the one turn where the bait wears off - and Chansey has something like a 45% chance to flee. This would add up:

10% + 10% + 10% + 10% + 45% / 5 turns* = ~17% chance of staying overall. 20 flees / 84 stays = ~23% - well within deviation given the small sample size.

(*: Assuming bait lasts an average of 4 turns.)

This is slightly corroborated by the third set of data I've been collecting, which is how many Chanseys stayed around long enough for me to throw a second bait: 8 out of my original 24 seen. The sample size is even smaller so we can't really draw conclusions from it, but it appears to be statistically significantly below the 56% stay chance we would expect by the second bait. This is anecdotally supported by the player experience that a good portion less than half of Chanseys stay beyond the first bait.

TL;DR: IzzetGuildmage's table on page 3 of the thread is unfortunately lacking full data. It assumes a continued 10% stay chance when the bait runs out. Instead, it is extremely likely Chansey gets a turn of "unbaited" 45%-ish escape chance on the turn that it would return to saying, "watches carefully." This significantly changes the balance between "bait-ball-ball-until you see watches carefully" and other methods.

[heated]

20 flees / 84 stays = ~23% - well within deviation given the small sample size.

You mean 20 flees / (20 + 84 sample size) = ~19% right?

Also, if you read further down the thread, I recalculated the chances given 55% stay and 4.8% catch every 5th turn and removing double-counting and got around 13%.

Furthermore, how do we know the Pokemmo team followed the original code to a tee at all? Maybe they wrote their own code. If that's the case, statistical data is all we got as we don't have access to the server code.

[Cubes]

I've been sampling some numbers regarding some of the questions in the thread. The sample size could be larger, but it's pretty safe to make a few assumptions already.

First turn, I throw bait. I measure how many Chanseys flee on the first turn. So far I'm at:

• Number of Chanseys stay: 24
  
• Number of Chanseys fled: 2
  

Small sample size, but it's already clear that the chance of a Chansey fleeing with bait is much more like 10%, and astronomically unlikely to be 45%. So we can figure that the stuff about "fleeing decided before commands are chosen" is incorrect/misleading.

No, this is very much possible if when bait/rock is thrown instead of ball, it will affect the escape rate for that same turn. So first turn could go

Bait Check- Watching carefully

Bait Chosen/Thrown- Recalculate bait check.

Escape check- Chansey stays (using bait chance of fleeing)

The next set of data I've collected is the chance for a Chansey to stay after I've used bait. On the second-plus turn, I throw balls until the "watching carefully" message reappears. These are the turns up to, but NOT including after, a "watching carefully message". To put another way, on the second turn, I throw a ball and I measure whether it stays or flees - once the "watching carefully" message appears, I stop measuring this category.

• Rounds Chansey stayed: 84
  
• Rounds Chansey fled: 20
  

Even with this small sample, you can see the chance of bait rounds being a global 10%-ish is also astronomically unlikely. We can assume one of two things:

1. The chance for Chansey to stay is not actually around 10%, and is more like 20% - 25%. OR
   
2. The chance for Chansey to stay when the bait runs out does not stay at 10%.
   

I believe the second option is extremely more likely the case. First, based on the first set of data, the first bait round is probably closer to 10% than 25% (a larger sample size would be desirable, however). Second, we've seen nothing anywhere that says the chance of fleeing while under bait increases/changes while the bait effect lasts in gen 3.

Talk about small sample-size theater.

Honestly, these are just projections based on what we know about the real games. We won't get a definitive answer for how it works until we see the code or a dev comes in and tells us it's one way or the other. Either way though, we've calculated it and shown bait>ball is better in all cases. It's just a matter of how much better.

[TheMaster42]

You mean 20 flees / (20 + 84 sample size) = ~19% right?

Good catch, that makes more sense with the ~17% too.

Furthermore, how do we know the Pokemmo team followed the original code to a tee at all? Maybe they wrote their own code. If that's the case, statistical data is all we got as we don't have access to the server code.

For sure.

No, this is very much possible if when bait/rock is thrown instead of ball, it will affect the escape rate for that same turn. So first turn could go

Bait Check- Watching carefully

Bait Chosen/Thrown- Recalculate bait check.

Escape check- Chansey stays (using bait chance of fleeing)

I'm not sure what you're saying here. I was referring to that second site you discovered that had language which seemed to imply the escape value was determined that turn, before the player had thrown bait/rock. Which almost certainly isn't the case.

Talk about small sample-size theater.

It is, however, much more than enough, statistically, to determine that something is almost certainly more like 10% over 50% (or even 25%).

[TheMaster42]

Also, if you read further down the thread, I recalculated the chances given 55% stay and 4.8% catch every 5th turn and removing double-counting and got around 13%.

Is this the chance of catching an individual Chansey with either method? My statistics knowledge isn't deep enough to figure that on my own, but figuring out to 30 balls always seemed incorrect. Or did you expand beyond that already...

If you have spare time/are bored and want to walk me through it, I'd be delighted. :) Mainly how this is the right formula to use for this situation, etc.

[Cubes]

I'm not sure what you're saying here. I was referring to that second site you discovered that had language which seemed to imply the escape value was determined that turn, before the player had thrown bait/rock. Which almost certainly isn't the case.

What I said was that when you throw a bait/rock, then the escape value is recalculated for that same turn. It fits the first turn bait taking affect (which we all agreed it did), and allows for a whole turn of regular catch rate and 45% escape rate before we can even see the message.

I'm basically saying that we need to see the code before we know anything for sure.

[Hermes]

What's the % rate for the old method? bait -> ball -> ball -> bait ->ball all the way until it flees/caught.

[fathuntress]

Hi, Why not just

BALL-BALL-BAIT-BALL-BALL-BALL

MATH ME.