How EVE Online Attributes and Neural Remaps Work

For everyone. Numbers and details included — play with them yourself below.

Hey folks. With the Attribute Optimizer shipping in EveLens 1.4.1, this seems like the right time to lay out the numbers and details behind neural remaps — how training speed is actually calculated, what a remap really buys you, and how the optimizer decides where your remap points should go. I'm sharing the full math in the spirit of open discussion: if you spot a number that disagrees with the game, that's a bug report and I want it.

A remap locks your attributes for 365 days. Getting it wrong costs real months of training time. The good news: the math behind getting it right is small enough to fit in this post.

01

Training Speed

Every skill in EVE has a primary and a secondary attribute. Your training speed for a skill is:

SP per hour = primary × 60 + secondary × 30

That's the entire formula. No hidden multipliers, no rounding tricks. A fresh character has 17 in every attribute, so everything trains at 17×60 + 17×30 = 1,530 SP/hour. Omega clones train at this rate; alpha clones at exactly half. Implants add their bonus straight onto the attribute before the multiply — a +4 primary implant is worth exactly +240 SP/hour.

The one thing to internalize: a point of primary is worth exactly double a point of secondary. Everything else in this post follows from that.

02

What a Remap Gives You

Under the game's rules, a neural remap redistributes your base attributes as follows:

The greedy answer for a single skill falls straight out of the formula: 10 points into the primary (worth 60 each), the remaining 4 into the secondary (worth 30 each). That's a 27/21 spread, and it trains that skill at 2,250 SP/hour — 47% faster than flat 17s.

On a 200-day plan, 47% is around 64 days back. This is why remaps are the single biggest free acceleration in the game.

Here's the same math, live — pick what you're training, drag the sliders, and watch the days move.

03

The Catch: Your Plan Isn't One Skill

If your whole plan were Gunnery, you'd map 27 Perception / 21 Willpower and stop reading. Real plans mix Perception-heavy ship skills, Intelligence-heavy engineering, Willpower-heavy specializations. A spread that's perfect for one block is mediocre for another.

The right compromise depends on the time-weighted attribute mix of everything you're about to train — a 40-day skill pulls the answer a hundred times harder than a 4-hour one. Human intuition is genuinely bad at this. Fortunately, brute force is genuinely good at it:

With 14 points across 5 attributes, capped at +10 each, there are only about 14,641 legal combinations. EveLens simulates your entire plan under every single one and takes the shortest total. Under a second. Not a heuristic, not an approximation — the provably best single remap for your exact plan, by exhaustion.

04

When One Remap Isn't Enough

Long plans have a second lever: where you remap, not just what to.

Say your plan is 26 Intelligence-heavy skills (184 days) followed by 20 Willpower-heavy skills (214 days). One compromise remap serves both segments badly. Two remaps — Int/Mem now, Per/Wil at the boundary — serve each perfectly. Veterans remember this from old EVEMon as mid-plan remap points.

The optimizer's auto-placement treats this as a segmentation problem:

1. Scan the plan in training order for points where the dominant primary attribute shifts
2. Split there — but only if the resulting segment is at least 30 days long, because a remap you burn on a 3-week block is a remap you don't have when your doctrine changes next year
3. Run the full 14,641-combination search on each segment independently
4. Each segment's simulation starts from the skills the previous segments already trained — prerequisites you've paid for don't cost twice

On a real 46-skill test plan: one whole-plan remap saved 18 days. Auto-placed remaps at the two attribute boundaries saved 61 days. Same plan, same character. The only difference was admitting the plan changes personality halfway through.

05

The Design Intent

Two decisions in the optimizer worth explaining, since they're deliberate:

Apply is disabled when the proposal doesn't beat your current setup. If you've already applied three well-placed remaps, a one-remap strategy will honestly report "your current setup already matches or beats this" rather than offering a downgrade with a friendly green button. An optimizer should be incapable of making your plan slower.

The optimizer's numbers always match the plan editor's. Same training order, same implants, same applied remap points. If two parts of the app ever show you different durations for the same plan, that's a bug — file it. This is backed by the EVE Accuracy Suite, a set of automated checks pinning EveLens's math to the game's published values (the 1,530 base rate, the 0.5× alpha rate, the SP-per-level table, and so on) that run on every build.

06

TL;DR

• Primary is worth double secondary. 27/21 into your focus is +47% training speed.
• The best remap for a whole plan is a time-weighted compromise — let exhaustive search find it.
• Long mixed plans want remaps at attribute boundaries, if the segments justify spending a yearly remap.
• Open your plan in EveLens → Optimize Plan → preview every placement and spread → Apply. There's an Omega/Alpha what-if toggle for planning your clone state alongside.

o7 and train fast,

Alia

EveLens is a free, open-source character planner for EVE Online — Windows, macOS, and Linux. Numbers disagreeing with the game are bugs: file an issue.