Case Study: Paragon Markets - Epoch 1
Paragon partnered with Monarch to launch market maker rewards through Marina on three Hyperliquid HIP-3 markets: BTC.D, TOTAL2, and OTHERS. The markets had baseline liquidity, but spreads frequently blew out and depth collapsed for hours at a time.
Result: tradeable in calm conditions, unreliable when it mattered.
Liquidity campaign
Marina went live with the following parameters:
| Parameter | Value |
|---|---|
| Markets | BTC.D, TOTAL2, OTHERS (Hyperliquid HIP-3 Perps) |
| Duration | 1 week (2026-04-21 to 2026-04-28) |
| Pool | $5,000 per market ($15,000 total) |
| Reward heuristics | 50% maker volume, 50% quoted depth at 1% of mid, spread multiplier |
| Eligibility | $500 minimum hourly depth, two-sided quoting required |
Epoch 1 results
BTC.D
| Metric | Pre-Marina | Epoch 1 | Change |
|---|---|---|---|
| Avg spread (bps) | 6.5 | 2.4 | -63% |
| Avg depth at 1% | $912K | $1.09M | +20% |
| Avg depth at 2% | $1.18M | $1.37M | +16% |
| Worst-hour spread (bps) | 30 | 8 | tighter |
| Worst-hour depth at 1% | $38K | $258K | 7x |
| Hours with depth at 1% under $200K | 11 | 0 | eliminated |
OTHERS
| Metric | Pre-Marina | Epoch 1 | Change |
|---|---|---|---|
| Avg spread (bps) | 33.2 | 3.5 | -90% |
| Avg depth at 1% | $411K | $491K | +19% |
| Avg depth at 2% | $513K | $584K | +14% |
| Worst-hour spread (bps) | 427 | 13 | tighter |
| Worst-hour depth at 1% | $1K | $82K | 80x |
| Hours with spread above 50 bps | 19 | 0 | eliminated |
TOTAL2
| Metric | Pre-Marina | Epoch 1 | Change |
|---|---|---|---|
| Avg spread (bps) | 3.6 | 3.2 | -13% |
| Avg depth at 1% | $512K | $551K | +7% |
| Avg depth at 2% | $601K | $662K | +9% |
| Worst-hour spread (bps) | 86 | 8 | tighter |
| Worst-hour depth at 1% | $36K | $161K | 4x |
| Active makers | 89 | 114 | +28% |
Reliability Is Everything
The headline is that the worst hour of epoch one looked better than the average hour before Marina rewards.
| Combined across all three markets | Pre-Marina | Epoch 1 |
|---|---|---|
| Hours with spread above 50 bps | 20 | 0 |
| Hours with depth at 1% below $200K | 49 | 5 |
Spread blowouts disappeared. Depth collapses were nearly eliminated. The same pattern held across BTC.D, TOTAL2, and OTHERS in one week.
Who showed up
The maker whitelist included 66 community market makers.
Across the three markets, whitelist makers filled 28% of BTC.D maker volume, 48% of OTHERS, and 24.5% of TOTAL2.
Spread multiplier
Every market's Quote Score is gated by a spread quality multiplier:
multiplier = max(0, (100 - avg_spread_bps) / 40) ^ 2For this campaign, C = 100 and D = 40. Future campaigns can tune either parameter around their market goals.
| Wallet's avg spread | Multiplier | Effect on Quote Score |
|---|---|---|
| 20 bps | 4.0x | Strong reward |
| 40 bps | 2.25x | Moderate reward |
| 60 bps | 1.0x | Weak reward |
| 80 bps | 0.25x | Minimal reward |
| 100 bps + | 0 | No Quote Score contribution under this campaign cutoff |
Combined with the $500 depth floor and a two-sided quoting requirement, the formula rewards continuous tight quoting, not stale or one-sided dust on the book.
Complementary effect
Existing market makers on Paragon were not displaced. The campaign added a layer of community quoting that filled the gaps where books used to thin out. Average spreads tightened, blowouts disappeared, books stayed deep around the clock, and on TOTAL2 the active maker count grew from 89 to 114.
- Existing MMs provide the deep base layer.
- Marina makers provide tight, continuous, two-sided quoting.
- Both groups benefit from the increased taker flow that reliable markets attract.
Key Takeaway
A $15,000 weekly campaign, spread across three markets, eliminated spread blowouts on all of them (20 hours to 0), near-eliminated depth collapses (49 hours to 5), tightened average spreads by up to 90%, and routed rewards to 66 community market makers.
Marina's playbook on Paragon works: programmatic incentives can coordinate community liquidity at scale, and the resulting market structure improves consistency more than any single average metric suggests.