Vaal Orb corruption: when it pays, when it loses

A Vaal Orb on a skill gem is a coin flip with four sides. The expected value depends entirely on what each of those four outcomes is worth in the current market, minus what you paid for the gem. For most gems most of the time, the math doesn't clear — you'd lose chaos on average. For a specific subset of gems it absolutely clears, and the corruption table is built to find them.

The four outcomes

The official Vaal Orb mechanic on skill gems produces exactly one of:

• No change. The gem becomes corrupted but keeps its level and quality. Functionally identical to the uncorrupted version for everything other than further modification.
• Level changes by ±1. Half the time +1 (which can push the gem above its normal max — Awakened Empower L4 from L3 input, regular gem L21 from L20 input), half the time −1. The +1 outcome is the prize the strategy is chasing; the −1 outcome is the painful one.
• Quality is rerolled. Quality is replaced with a uniformly-random value between 0 and 23%. Most of the time the new quality is lower than what you started with (especially if you start at Q20). Occasionally it lands at 23, which is the maximum corrupted quality and trades at a real premium.
• Vaal-version transformation. If the gem has a Vaal variant (Vaal Spark exists; Vaal Greater Volley Support does not), this outcome replaces the gem with that variant at the same level and quality. If no Vaal variant exists, the slot collapses to "no change" — it's just a missed roll.

Each outcome is weighted at 25%, with the level branch split evenly between +1 and −1, giving 12.5% each. There is no chance of a random "corruption implicit mod" — that mechanic only applies to equipment, not to skill gems. A Vaal Orb on a gem never adds bonus properties; it just selects one of the four outcomes above.

Expected value, in concrete terms

The EV of a Vaal Orb on a specific gem is a probability-weighted sum:

EV = 0.25 × price(no_change) + 0.125 × price(L+1) + 0.125 × price(L−1) + 0.25 × price(Q_rerolled) + 0.25 × price(Vaal_version)

Profit per attempt is EV − input_cost − Vaal_Orb_cost. The Vaal Orb itself is essentially free relative to even cheap gems (around 1 chaos on most leagues, fractions of a chaos on new leagues), so the decision is really about whether the probability-weighted outcome value exceeds your input cost.

When the math clears

Three categories of gems consistently produce positive EV per Vaal Orb:

• Awakened supports with a valuable +1 outcome. An Awakened Empower at L5 sells for many divines; the L+1 corruption outcome — Awakened Empower L6 corrupted — is rarer and often worth more again. If your input is a finished L5 Awakened gem, the L+1 branch alone can carry the EV above input cost. The Awakened Multistrike Support, Awakened Brutality Support, and similar high-value Awakened gems show up reliably at the top of the corruption ranking when their L+1 markets are active.
• Gems with valuable Vaal variants. Some skill gems have Vaal versions that are far more popular than the base. Vaal Reave, Vaal Spark, Vaal Power Siphon, Vaal Summon Skeletons — whenever a Vaal version is in a top build, transforming into it via a Vaal Orb is a real outcome with real price upside. The 25% transformation chance can dominate the EV calculation.
• Cheap inputs where any positive outcome covers cost. For a gem you can buy at 3c, even a 5c "no change corrupted" sale is a positive trade. Most of the gems with extreme ROI percentages (the ×-suffixed numbers in the table) come from this category — small input, modestly positive expected outcome, multiplier looks enormous in percentage terms.

When the math doesn't clear

Most gems are bad Vaal Orb targets. The common pitfalls:

• No Vaal variant exists. Most support gems have no Vaal version, so 25% of your rolls collapse to "no change" — you essentially have a 75% probability tree instead of 100%, which puts pressure on the other branches to make up the difference.
• High-quality input. If you start at Q20, the 25% "quality rerolled" outcome is on average worse than your input (the reroll distribution skews lower than 20). Even on the lucky 1/24 Q23 roll, the upside is bounded; on the 19/24 Q ≤ 19 rolls, the gem is worth less than your input.
• Lvl 20 cap with no upside. Some regular skill gems simply don't trade meaningfully at level 21 — there's no liquid market for the corrupted L21 outcome. The +1 branch becomes effectively worthless, and the strategy collapses to "you might get a Vaal version" — which is fine if the Vaal version is valuable but offers nothing if it isn't.

Double corruption

Doryani's Institute Tier 3 (a specific room you can build into the Temple of Atzoatl) applies two sequential Vaal Orb rolls to a single gem. This expands the outcome space dramatically: 25 distinct paths through the probability tree, ranging from "no change twice" (still the modal result) to "L+2 with a Vaal transformation and Q23" (vanishingly rare, but extreme upside).

The cost basis is different: you don't supply two Vaal Orbs; you supply a single Temple of Atzoatl with the Doryani's Institute room at tier 3. Temples are traded as items on poe.ninja and run from ~50 chaos in new leagues to over 1000 chaos on Standard. The corruption table subtracts the current Temple price from the double-corrupt EV automatically.

Double corruption is generally only worth attempting on gems where the single-corrupt EV is already positive and the variance reduction of two rolls (versus one) lets you target a specific high-value combination like L+1 Q23 corrupted. The math rewards gems with thick markets on those compound outcomes.

Practical workflow

• Open the corruption table sorted by single- or double-profit (whichever strategy you're running).
• Filter to your tier of interest. Awakened tier surfaces L+1 plays; Regular tier surfaces Vaal-version plays.
• Confirm the high-confidence toggle is on — corruption math is especially sensitive to thin-market price noise.
• Pick gems where the profit number is positive and meaningful relative to the input price. A 50c profit on a 30c gem is excellent. A 50c profit on a 5000c gem is fine but slow to compound.
• Buy a stack, Vaal Orb each individually, sort by outcome, list the keepers. Reuse the duds as map-leveling fodder.

The site's posture is conservative

The corruption EV numbers on this site are deliberately a lower bound rather than a best-case estimate — the methodology page details exactly why, but the short version is that the listed prices for "L20 Q20 corrupted" outcomes appear inflated in ways that don't reflect actual sale prices, and we cap those listings at the uncorrupted input price to avoid double-counting the upside. In practice, a careful trader should slightly outperform our numbers rather than underperform.