Front vs back centering: which side limits your card?

Front and back centering are separate measurements with separate limits: the reverse can be more off-centre without violating the same named centering band. For PSA 10, the current reference is approximately 55/45 front and 75/25 back; the combined ceiling is limited by the lower face result, not an average. A strong front does not cancel a failing reverse, and an unmeasured reverse remains unknown.

Why is the back allowed to be more off-centre?

The different allowances come from grading policies, not from a different mathematical formula. The published PSA and CGC descriptions distinguish the two faces. A wider reverse allowance means a particular back ratio may remain within a named band even when that ratio would exclude the same band on the front. It does not mean the back is unimportant or that damage there is ignored.

Collectors can discuss how much a visible front imbalance affects presentation, but that does not establish a universal reason for every house's choice. Use the stated policy rather than assuming all services follow the same visual priorities. Some model back rules are interpreted, and SGC has no separate numerical back rule here. The comparison should retain those limits instead of inventing a common reverse allowance.

Gem Mint centering limits in the current model
House and gradeFront: worse axisFront: other axisBackStatus (front / back)
PSA 10 GEM MT55/4555/4575/25official / official
BGS 9.5 Gem Mint55/4552/4860/40interpreted / interpreted
CGC 10 Gem Mint55/4555/4575/25official / interpreted
SGC 10 Gem Mint55/4555/45—approximate / —

Limits apply to both axes separately. For BGS front, the worse axis must meet the loose limit and the other axis the tight limit. — means no separate numerical back rule in this model. A ceiling is not an overall grade.

The geometry remains identical: wider opposite border ÷ sum of that pair × 100. You still measure both axes on the reverse. A service with a permissive back centering limit can still reject a high grade because of reverse corner wear, scratches or other defects outside this checker’s scope.

How are front and back ceilings combined?

First evaluate the front horizontal and vertical ratios under the front rule. Then evaluate the two reverse ratios under the reverse rule. Compare the resulting bands by grade and designation, and retain the lower centering ceiling. If both land in the same band, the tool reports the face with less margin to its threshold as the limiting face.

Never average four ratios into a single number, average the two face grades, or apply the front tolerance to a reverse photograph. Those shortcuts can respectively hide a bad axis, invent an unmodelled grade or penalise the reverse using the wrong policy. The checker needs a confirmed measurement for each face before its combined result can answer this specific question.

For example, if one face reaches the top centering band and the other reaches a lower one, the lower band controls. That comparison is about centering alone. It does not establish that the card will receive that final grade, because the other condition criteria have not been evaluated.

What is an example in which the front limits the card?

Take a hypothetical front with 3.60 mm left and 2.40 mm right: 3.60 ÷ 6.00 × 100 = 60%, or 60/40. Suppose its top and bottom borders are equal. The PSA front ceiling is 9 MINT.

Now take a back with 5.60 mm on one side and 2.40 mm on the opposite side. Its ratio is 5.60 ÷ 8.00 × 100 = 70%, or 70/30, with an equal vertical pair. The PSA back ceiling is 10 GEM MT. The reverse is visibly more displaced in ratio terms but remains in a higher back centering band. The front therefore controls the combined PSA ceiling.

The different border sums are intentional: front and back frames need not have equal dimensions. Do not compare raw millimetre differences across faces without calculating each pair's ratio. The ratio guide explains why thickness and relative balance are distinct quantities.

What is an example in which the back limits the card?

Consider a front with equal 3.00 mm left/right borders and equal top/bottom borders, giving 50/50 on both axes. The PSA front ceiling is 10 GEM MT. Now give the back an opposing pair of 6.40 and 1.60 mm: 6.40 ÷ 8.00 × 100 = 80%, or 80/20. Its other pair remains equal.

The PSA back ceiling becomes 9 MINT, so the reverse controls this time. The generous back allowance has a limit; it is not an exemption. The table shows these cases and a third example where both faces fit the top model band.

PSA: which face sets the combined centering ceiling?
Front L/R; T/BBack L/R; T/BFront ceilingBack ceilingCombined ceiling / limiter
60/40; 50/5070/30; 50/509 MINT10 GEM MT9 MINT / front
50/50; 50/5080/20; 50/5010 GEM MT9 MINT9 MINT / back
55/45; 50/5075/25; 50/5010 GEM MT10 GEM MT10 GEM MT / front

Each face is evaluated first, then the lower ceiling controls. On equal bands the face with less threshold margin is reported. This does not average the two sides.

The model's reverse bands can be coarse. A back ratio outside one band is assigned the next available band, not an interpolated intermediate grade. Read the full standards when the detail matters and inspect the actual card before assuming the model output describes all grading outcomes.

What changes for BGS, CGC Pristine and SGC?

BGS adds the front pairing rule: the worse axis can meet a looser threshold while the better axis must meet a tighter one. The reverse uses its own band list. Retain both front ratios and both reverse ratios; a single worst number for the entire card is insufficient. The four-house comparison separates these interpreted BGS requirements from the approximate lower bands.

CGC's model distinguishes Pristine and Gem Mint on the front, but it does not have a separately established numerical Pristine reverse band. The combination logic can preserve the Pristine front designation when the reverse reaches its top model band. Treat that as a conditional model result, not proof that the reverse satisfies the official Pristine description. The practical perfect-centering tolerance is itself interpreted.

For SGC, the model can report a front ceiling without a numerical reverse constraint. State “front only; reverse rule not modelled” rather than saying both sides passed. A measured reverse ratio is still useful evidence even when this reference cannot assign it a separate SGC ceiling.

How should you record and share both faces?

Use separate labelled captures with complete physical cuts and a consistent printed boundary on each face. Check that the tool's face selection matches the photograph, then confirm the lines for both. A front and reverse need not use the same colours or inner artwork references; each requires its own defensible frame.

Keep the ratios, raw readings where available, uncertainty and any tilt or boundary warnings with the result. If one photograph is blurry or cropped, repeat it before reporting a combined ceiling. A confident front measurement cannot lend confidence to a weak reverse image. Follow the phone measurement workflow and the method, then use the submission checklist to inspect condition beyond the borders.

Frequently asked questions

Can a perfectly centred back rescue an off-centre front?

It cannot erase a lower front centering ceiling. The two faces are evaluated separately and the lower result controls the combined model outcome.

Should front and back have the same border thickness in millimetres?

No. Their printed frames can differ. Calculate each pair’s ratio from its own border sum; equal physical thickness across the two faces is not required.

Which face is called limiting if both have the same ceiling?

The model selects the face with less remaining threshold margin. That identifies the closer boundary within the same band, rather than changing the numerical ceiling.

Does a retained CGC Pristine front result prove the back is Pristine?

No. The current reverse model does not establish a separate numerical Pristine requirement. The designation remains conditional and should not be shared as full Pristine eligibility.

Which sources and limits does this guide use?

Tables are generated from the same data source as the tool. Threshold verification: 2026-10-01. Centering standards · How it works.

official = published numerical rule; interpreted = model interpretation; approximate = working estimate. retrieved = source retrieved; unavailable = not accessible; brief-only = supplied reference without direct retrieval. Individual band status is shown in the standards: the house-level status does not replace each row's status.