LZLZL/Prediction markets/Rebates
FREEJUDGE IT B · RebatesSubsidy

LP rewards
the one that pays without a fill

2026-08-21 · with the scoring formula and depth measurements

The odd one out among the four: no fill required — resting size scores. Two things to establish first though: not every board has it, and the scoring formula specifically pays for tightness. "Quote deeper, stay safe, still collect" is structurally impossible.

1How it differs from the maker rebate

Maker rebateLP rewards
Triggermust be filledpays unfilled
You wanttight quotes, get hitqualifying quotes, ideally not hit
Main riskadverse selectionbad fills when swept
Availableany board with a taker feeonly boards in the programme

Both pay posters, with exactly opposite triggers. That is why "market making" is two businesses here — and why lifting someone's approach so often fails: they are collecting this one while you assume you are collecting the other.

2The scoring formula

S = ((v − s) / v)² × b s = distance of your quote from mid · v = the board's maximum spread · b = your size
PropertyConsequence
s measures distance from midnot queue level. Closer to mid scores higher
The bracket is squareddecay is fast — stepping back costs more than you would guess
Beyond v it is zeroquote too wide and you get nothing, not less
⚠ "One level deeper, same reward, safer" is false

It appears in a lot of guides and it contradicts the formula directly: the programme is built to buy tight quotes.

I ran a full replay across every recorded reward board, comparing quoting at the best price against stepping back one and two cents:

Quoting atRewardFill countBad-fill $
join best×1×1×1
−1¢×0.59×0.33×0.53
−2¢×0.26×0.16×0.29

Two cents back leaves a quarter of the reward. "Safer" is real (bad fills fall to 0.29) but it is not free — you are paying three-quarters of the reward for it.

3⚠ How to tell whether a board has a pool

clobRewards: [] does not mean no rewards

An empty array from the events endpoint does not imply "no pool". The authoritative source is the CLOB's sampling-markets endpoint — pulling it in full returned 8,631 boards carrying a real daily rate, with a nominal pool of about $172,177/day. Not remotely the same set as filtering on clobRewards.

Same family: a board's metadata carries fields like minimum size for rewards and maximum spread. Their presence does not mean money is paid — they are part of a config template. More of the family in eight API traps.

4⚠ On crypto up/down this is essentially zero

TenorWhat the rewards endpoint returns
1 hour / 15 minutesempty
5 minutesa placeholder config (daily rate 0.001)

So market making up/down earns essentially none of this subsidy; that lane's only subsidy is the fill-triggered one. Anyone telling you "up/down market making has LP rewards" has conflated the two.

5When the money arrives

ItemDetail
Frequencydaily
Timeabout 00:45 UTC
Formbatched transfers, roughly 400 per batch
Recorded innot trading P&L — it is a transfer, not a trade

That last row makes you wrong in both directions: underrating players who live on this (their leaderboard number can look poor or negative) and overrating your own strategy if you counted subsidies as strategy income. Real net income is trading P&L + total received, queried separately and summed.

6I ran this once; tuition about $210

I ran a line on weather doing nothing but collecting this subsidy. It is archived at about −$210. Rather than the playbook, here are the transferable conclusions:

LessonNote
The pool is realthe money genuinely pays out daily, verifiable on-chain
You are not the only one quotingit is split by score; more participants, less each
Bad fills are the main costscoring forces you tight, and tight gets swept
Subsidy removal is policy riskrates, splits and eligible categories can change — and then it is negative

The last one deserves a thought: a strategy that is 100% dependent on a subsidy does not live or die by your judgement — it lives or dies by a product decision. Size it as though the subsidy could be zero tomorrow.

7One line to keep

This one does not require a fill, but it does require tightness — and tight quotes get swept. So it is not free money; it is money bought with sweep risk. The only way to know whether that trade is good is to net the bad fills against it.

EvidenceCheck it yourself

Scoring S = ((v−s)/v)² × b, s measured from mid, v the board's maximum spread; quadratic decay, zero beyond v.
Depth replay every recorded day × every reward board, naive round-the-clock quoting: reward ×1 / ×0.59 / ×0.26, fills ×1 / ×0.33 / ×0.16, bad-fill $ ×1 / ×0.53 / ×0.29 (join best / −1¢ / −2¢).
Pool size full pull of the sampling-markets endpoint: 8,631 boards with a live daily rate, nominal pool about $172,177/day.
Zero on up/down measured returns for live 1h / 15m / 5m boards.
Payout form about 00:45Z, batched transfers of roughly 400, enumerable on-chain.
Weather tuition my own line, about −$210, archived.
Checked 2026-07 to 2026-08. Reward rules change frequently — verify before use.

NextWhere to go

B · REBATES
Four rulebooks, four pools
B · REBATES
How much is paid out each day
B · REBATES
Why tight quotes get swept
E · UP/DOWN
Weather LP: the −$210 in full
This is an educational and research record. It is not investment advice, promises no returns, and offers no personalised trading recommendations. Rules and API behaviour are per the official documentation; this page states when it was checked and both can change without notice. Prediction markets are restricted or unavailable in some jurisdictions — confirm your own before taking part.