Everything an engine can hold — and the guard rails it runs inside. Every number on this page is taken from the engine source itself.
| Block | What it does | Limit |
|---|---|---|
| Contenders | Who competes — drivers, fighters, teams, animals. They are what people bet on. You pick the name and colour. | 2–64, plus 16 groups |
| Attributes | Abilities as numbers, for example speed 0–100. You set the scale once and fill in each contender in a table. | 32 |
| Variables | Conditions of the world. Four kinds: weighted choice (rolled every round), fixed choice, number, yes/no. | 32, up to 20 options each |
| Effects | “If it rains, then −14 speed.” Six arithmetic modes. Applies to all, to a group, or to individuals — with exceptions per contender. | 100, up to 50 cases each |
| Conditions | Comparisons with AND, OR, NOT. They act as switches for effects, as triggers for events and as flags in the outcome. | 100, nesting 8 |
| Formulas | Plus, minus, times, divided, min, max, average, clamp, absolute. Per contender or once globally. Cycles are ruled out. | 100 |
| Randomness | Integer, percentage or weighted choice. Drawn per round, per step, or per contender and step. | 50 |
| States | Values that build up during the round: tyres, fuel, score. They may read themselves — feedback across the steps. | 50 |
| Events | Named moments like OVERTAKE or CRASH. Front ends animate on them, outcomes can count them. Nine names are reserved for the lifecycle. | 50, 2,000 per run |
| Simulation | How many steps a round has, what the final sort goes by, and in which direction. A tie is broken by order, never by chance. | 1–100 steps |
| Outcomes | What stands at the end: winner, position, top N, score, best step, event count, flag. | 50, seven kinds |
| Markets | The questions asked of the outcome. You only pick the question and the outcome — odds and settlement are Sol-Core’s job. | 100, eight kinds |
There are no other market kinds — deliberately. The settlement rule lives in the template, not in the creator’s draft. That is why no market can exist that cannot be settled cleanly.
Who wins?
The selection is a contender.
Will Alpha finish second?
Yes or no, on a fixed place.
Does Alpha make the podium?
Yes or no, on the best N.
Who does better, A or B?
A duel taken from the ranking.
Who takes the best value?
The fastest lap, for example.
Which band does the value fall into?
2 to 12 bands to choose from.
Is the value above 1,700?
Yes or no, against a threshold.
Does Beta retire?
Yes or no, on a flag or a counter.
Draft
Every change is saved — always the whole document, up to 512 KB. The draft lives on its own; nobody sees it but you.
Validate
19 rules, from the unknown reference to the impossible schedule. Every finding jumps to the block that caused it.
Test lab
Up to 10,000 runs on the server. The same seed gives bit-identical results — randomness here is reproducible, not arbitrary.
Publish
The version is frozen and gets a checksum. There is no code path to change it afterwards — only a new version.
Odds
Sol-Core measures every selection until the band is tight enough. A duel takes about two minutes; rare outcomes take far longer.
Active
Rounds run to schedule. One active engine per wallet. A new version only takes over once you activate it — running rounds keep theirs.
Deleting is only possible while nothing was ever published. After that only archiving remains — the version records are meant to survive.
These limits are not penny-pinching. They are the promise that every round finishes inside its scheduled window.
Sol-Core does not know an outcome’s true chance — it measures it. Every measurement has a spread, and that spread decides the price: the maths uses the upper edge of the band. The payout therefore always stays below the 96% target, however precise the measurement is. With the average it would not — at only 200 hits, the chance that the true payout slips above 100% is around 29%.
Never landed
Not once in any run. Without a hit there is no price.
Too fuzzy
The measurement band is too wide. A price from it would be a guess.
Above the cap
The price would exceed the maximum multiplier of your creator tier.
No margin
The payout would fall under the floor — that floor protects the player.
It is always the single selection that is rejected, never the whole market: every bet settles on its own, so one unpriceable outsider does not kill the race.
Public
Name, description, status, version, checksum, schedule, contenders and groups, the kinds of outcome, and the markets.
Never public
Formulas, effects, conditions, random blocks, states, base values, weights, the canvas and the draft. That is your build — and at the same time the surface someone could use to price the markets sharper than you.
Shown: the randomness was fixed before the betting window closed. The checksum of the seed is published beforehand, the seed itself afterwards.
Not shown: that this randomness produces exactly this result. That would take the rules — and they stay secret. Which is why this says “confirmed by the server” and not “provably fair”.
And to be straight about it: the secrecy is not encryption. Across many rounds, weights and thresholds can be estimated statistically. For the money that does not matter — the odds are priced cautiously. For ownership of your build it does.
Done: builder, validation, test lab, immutable versions, the full lifecycle of an engine.
Running: rounds on schedule, odds per version, bets in demo mode — through the very same chain the real thing will use, but with no real money.
Planned: the money path with fees and limits, engines inside engines, and a starter kit for publishing your own front end.
19 · 10 000 · 96 % · 4 σ