What Actually Moves Your Capture Rate

Choose the task and finish the next step

Check the target’s capture eligibility, bring a suitable sphere, lower HP under control and confirm the world setting before planning a batch of throws. A field Alpha, a tower encounter and a raid are different encounter types; a positive numerical estimate does not override a scripted restriction.

Every percentage in this guide is an output of the declared reference-calibrated estimator, not a recovered native v1.0.3 probability. Its curve calibration comes from v0.7.3. Pocketpair’s official 1.0 announcement says the capture formula changed, so replacing input tables alone does not validate that older curve against 1.0.

Work from a reproducible starting point

Find the exact species and acquisition route, then record its observed level and whether the encounter is an Alpha. Do not use the ordinary species profile as permission to capture every scripted version.
Set the actual remaining HP, sphere, Capture Power rank, equipped module and PalCaptureRate. Keep critical off for an ordinary throw; it is a conditional event comparison.
Open the sphere comparison, change one input and compare the result. Save or export the complete plan so another attempt uses the same assumptions.
Recall an attacker before its next hit would defeat the target. Effects that preserve 1 HP can help control damage, but they do not establish that a target is capturable or account for every other source of damage.
For a status-based plan, read the current in-game prompt and the Pal’s exact condition. This estimator has no status/sleep or partner capture-power input, so do not silently fold those effects into sphere power.

Evidence boundary

The guide preserves every source comparison task while distinguishing input data from model behavior. It does not recover the native formula, status treatment, critical occurrence rate, scripted encounter permissions or the current HUD-to-success transformation. Those remain factual acceptance gaps rather than being hidden by precise-looking percentages.

Evidence layer	Recorded values	Confidence boundary
Pinned extracted inputs	10 sphere powers;288 species coefficients;16 effigy states; module item groups	Exact imported values, not independent current-game certification.
Reference setting literals	Back 0.3; critical 2; ignore-first-failed field 0.5; maximum-bounce field 2	Field names/values do not recover their native callsites. The 0.5 field is not asserted to mean half of failed throws are ignored.
Fitted model	HP/level/power curves,0.7 Alpha factor, final transform and 99.5%ceiling	Older v0.7.3 calibration; official 1.0 changed capture calculation. No claimed±5 point accuracy guarantee.
Cumulative throws	1−(1−p)^n for a fixedp	Exact probability arithmetic under independent identical attempts; unknown or changing game conditions invalidate that assumption.

Rank the six controllable levers within one model scenario

The comparison holds a level 30 target at 50% HP with species coefficient 1, an Ultra Sphere, rank 0, no module, world 1, no Alpha/back/critical flags. The low and high settings below change one lever only. The ordering describes this scenario and parameter range; it is not a universal ranking of native capture mechanics.

Lever	Modeled low	Modeled high	Swing, percentage points	Next action
Sphere power	<0.01%	99.50%	99.5	Bring a sphere tier suitable for the target; compare every tier before spending the rare one.
World setting	3.46%	70.29%	66.8	Use the actual PalCaptureRate. Changing it is a host/server decision, not a bonus earned by a throw.
Target HP	0.17%	63.32%	63.2	Reduce HP with controlled attacks, then stop damage before the finishing hit.
Effigy rank	22.56%	72.77%	50.2	Trade collected Lifmunk Effigies for permanent capture power; do not confuse rank with raw effigy count.
Module strength	22.56%	41.57%	19.0	Use one equipped module strength group. The model treats its label as added power; that interpretation is not native verification.
Back-hold flag	22.56%	38.30%	15.7	Only select it when the actual throw qualifies. The model adds 0.3 before its final transform.

Compare all 10 sphere tiers

Power is an exact value in the imported sphere table. The three chance columns are estimates using the same coefficient 1, world 1 and no statue/module/back/critical/Alpha bonuses. A displayed 99.50% is this model’s ceiling, not a guaranteed capture.

Sphere	Pinned power	Lv30 /50% HP, model	Lv30 /1% HP, model	Lv50 /20% HP, model
Pal Sphere	7	<0.01%	0.04%	<0.01%
Mega Sphere	14	0.08%	0.45%	<0.01%
Giga Sphere	20	0.53%	3.08%	<0.01%
Hyper Sphere	26	3.37%	17.13%	0.06%
Ultra Sphere	33	22.56%	63.32%	0.54%
Legendary Sphere	38	56.22%	88.38%	2.33%
Ultimate Sphere	44	88.08%	97.77%	12.07%
Exotic Sphere	48	95.89%	99.28%	30.27%
Sol Sphere	58	99.50%	99.50%	88.03%
Ancient Sphere	64	99.50%	99.50%	97.51%

Effigy ranks 0–15 and equal-power sphere alternatives

The pinned table has 16 states including unupgraded rank 0. Each rank adds 0.5 power, reaching+7.5 at rank 15. The raw requiredRelics column is cumulative: reaching rank 15 from zero requires 100 effigies. For a partial upgrade, subtract the current cumulative total from the target total; do not sum those cumulative rows. Equal total power gives identical outputs in this estimator because sphere and statue power enter the same sum. This is an identity of the declared model, not proof that two complete in-game situations are interchangeable.

Rank	Added power	Effigies for this step	Cumulative effigies
0	0	0	0
1	0.5	1	1
2	1	2	3
3	1.5	3	6
4	2	4	10
5	2.5	5	15
6	3	6	21
7	3.5	7	28
8	4	9	37
9	4.5	9	46
10	5	9	55
11	5.5	9	64
12	6	9	73
13	6.5	9	82
14	7	9	91
15	7.5	9	100

Eight exact power identities in the pinned inputs

For example,7+7 equals 14: Pal Sphere with rank 14 and Mega Sphere with rank 0 have equal input power. This statement does not apply to an unmodeled status, scripted encounter rule or changed native formula.

Lower tier plus statue rank	Total power	Same power as
Pal Sphere + rank 14	14	Mega Sphere
Mega Sphere + rank 12	20	Giga Sphere
Giga Sphere + rank 12	26	Hyper Sphere
Hyper Sphere + rank 14	33	Ultra Sphere
Ultra Sphere + rank 10	38	Legendary Sphere
Legendary Sphere + rank 12	44	Ultimate Sphere
Ultimate Sphere + rank 8	48	Exotic Sphere
Sol Sphere + rank 12	64	Ancient Sphere

Handle the level cliff before spending weaker spheres

This sweep uses a Pal Sphere,50% HP and coefficient 1. The rapid fall is a property of the fitted level curve. Use it to spot an underpowered modeled setup, then compare a stronger tier or a genuinely lower-level spawn of the species instead of assuming one more effigy rank fixes every gap.

Target level	Per-throw modeled chance
Level 1	41.19%
Level 5	11.31%
Level 10	2.28%
Level 15	0.49%
Level 20	0.11%
Level 25	0.03%
Level 30	<0.01%
Level 40	<0.01%
Level 50	<0.01%

Lower HP without losing the catch

The following Ultra Sphere/level 30 sweep changes only HP. The model gives most of the increase before the last sliver:25% HP is about 45.09%,10%about 57.16% and 1%about 63.32%. Stopping near a quarter of health is a source-informed risk tradeoff, not a native optimum for every Pal or weapon. Use controllable attacks and dismiss an attacker that is about to finish the target.

Mercy Hit’s recorded description prevents its bearer from reducing the target below 1 HP. Daedream’s partner projectiles also have a nonlethal description. These effects are damage-control aids, not a universal immunity to other damage. Official 1.0 changed some follow-up skill gear requirements, so item existence alone does not prove a necklace must be crafted first.

Remaining HP	Per-throw modeled chance
100% HP	0.17%
75% HP	5.61%
50% HP	22.56%
25% HP	45.09%
10% HP	57.16%
5% HP	60.69%
1% HP	63.32%

Back hold is an addition in this model

The imported reference setting is named CaptureRateAdd_ByLegHold with literal 0.3. This estimator adds it to the intermediate score before its display transform; it does not multiply the displayed percentage by 1.3 or add 30 percentage points to the final chance. The native callsite and the mapping between “leg hold”, a rear throw and current in-game behavior have not been independently recovered here.

Scenario	Ordinary model	Back-hold model
Lv30 /50% HP /Ultra	22.56%	38.30%
Lv60 /10% HP /Ultra	0.06%	2.01%
Source article’s Lv80 /fullHP /Pal Sphere	Published older-fit example:<0.01%	Published older-fit example:0.65%; outside this tool’s level 1–60 bounds

Compare the actual flag at reference inputs ↗

Four encounter factors are not free bonus switches

You may choose a different target or route, but cannot simply edit an encountered Pal’s level, species or Alpha status. The model’s critical switch is not a method for forcing the event and is not part of an unconditional average catch chance.

Factor	Pinned input or model treatment	Practical boundary
Target level	Fitted level curve	Read the target’s actual level. The tool accepts 1–60; this is not the game’s level cap.
Species	288 coefficients;26 differ from 1; range 0.8–1.5	Select the actual Pal. A 1.5 starter coefficient cannot be transferred to a harder species.
Alpha	This model uses 0.7 of the species rate	Set only for the intended Alpha scenario. Tower/raid eligibility is a separate rule.
Critical event	Literal bonus 2; this model interprets it as×2 of the intermediate score	Conditional what-if only. Trigger probability and current native application are unverified; do not budget every throw as critical.

Status, party support and server settings

Input or mechanic	What is recorded	What to do
PalCaptureRate	Official server documentation names it as the capture-rate multiplier; documentation version 1.0.4	Use the host’s actual value. The estimator’s 0.5–2 bounds are its control range, not an official universal server limit.
Module strengths	Groups+1,+2,+3; six item names	Choose the one equipped group. Treating a label as added power is a declared modeling assumption.
Freeze with Muffly	The partner description increases capture against Frozen targets; rank 5 numeric row 5	Keep the status and carrier conditions explicit. This estimator omits that row;5 is not established as+5 percentage points.
Sleep, poison and other statuses	No complete native status coefficient table is supplied to this estimator	Use current in-game status/aim feedback. No invented 1.3×/1.35× or stacking rule is applied here.
Capture eligibility	The numerical model contains no tower/raid/scripted encounter whitelist	Verify the actual encounter rule before spending a sphere.

A high-level Alpha comparison is not permission to capture a tower boss

The next table models a level 50 Alpha with coefficient 1 at 1% HP and world 1. The kitted columns use statue rank 15 and module+3; critical is off. Reptyro is a real level 50 field-Alpha record in the pinned map, so its acquisition guide provides a concrete starting route. Even here, the numerical result is only the older calibrated model.

The source groups tower bosses with Alphas in its heading, but this calculation only applies a numerical Alpha penalty. It does not establish that a tower boss, raid boss or other scripted variant can be captured. Use the boss preparation page for the fight and an actual acquisition record for the capture route. Near-ceiling output is still below certainty.

Sphere	Power	Bare model	Rank 15 +module 3 model	Kitted +back-hold model
Ultra Sphere	33	0.33%	6.69%	18.12%
Legendary Sphere	38	1.42%	23.20%	38.97%
Ultimate Sphere	44	7.65%	62.40%	72.21%
Exotic Sphere	48	20.75%	83.62%	87.50%
Sol Sphere	58	81.61%	98.79%	98.93%
Ancient Sphere	64	95.93%	99.50%	99.50%