NEAR Hackathon 2026 · Investment thesis track
The market values NEAR at $4.34B; Intents cash flow on the model's expected path supports $1.98B, a 2.2× gap. Closing it on Intents alone needs 133% a year of volume growth against a measured +27.9% a year; below are the measured strengths, the weaknesses and the triggers that would falsify the case.
Buybacks, gas burn and storage locks cover 21.6% of measured issuance, so the float grows by about 70,377 NEAR a day. It stops growing only at 100% coverage, which needs 4.8× today's Intents volume at the current inflation cap. As measured, this cuts against the bull case.
The counters are projected at the measured daily rates, not a live reading. Rates are from the snapshot at block 224,323,532.
See the supply modelThe bar is everything minted, to scale. The float stops growing when the removed share reaches 100%.
Reading on-chain total supply from NEAR RPC…
Market FDV $4.34B against an expected-path value of $1.98B over 36 months. Each lever is solved for the value that matches the market with every other input at base. Volume growth, public take, confidential share and year-3 holder revenue would each need 4.5× to 18.7× what the data shows today. The terminal multiple needed, 135×, is below where peers trade (2,498×, n = 3), but the model caps it at 60×, which is consistent with the market pricing NEAR as a layer 1 rather than as an Intents cash-flow stream. Even a 0% discount rate gives $3.82B, short of $4.34B: the gap is not a discount-rate question. The seeded Monte Carlo median is $644.7M (10th to 90th percentile $126.5M to $3.92B); the solves run on the expected path.
Peers: SOL, ETH, HYPE, with fees and revenue from DefiLlama and FDV from CoinGecko; NEAR counts L1 gas and Intents fees together ($72.26M a year). On revenue NEAR is at 186×, rank 2 of 4, lowest first. At the peers' FDV-to-fees multiples, NEAR's fees map to $11.20B to $43.73B: a cross-check on how the market prices layer 1s, not a fair value. None of the 3 peers trade at or below the model's 60× terminal multiple. The weak side: issuance pays stakers $108.5M a year, and holder revenue covers 22% of it.
Identical dry quotes at $500, $50K, $250K sent to 1Click, Relay, LI.FI and Across, compared on units delivered at the destination. Competitor friction on cross-VM routes caps a sustainable fee at 3.00 bps; 29 of 36 quotes returned.
Break-even needs $608.0M of daily Intents volume under the current 2.5% cap (4.8× today's $125.8M) or $386.0M under the SVRN glidepath (3.1×). The simulation puts the chance of a crossover within 36 months at 66.1% and 73.3% respectively.
The protocol charges 1 NEAR per 100 KB of contract state, released only when the state is deleted. Across 8 tracked contracts 283,182 NEAR is locked; the growth offsets 0.38% of daily issuance, a real but small sink.
Gross fees were 16.06 bps, of which 30.0% reached the buyback. About 16.7% of volume ran through confidential execution, estimated on-chain from Confidential Intents Treasury flows over the last 24 hours (band 13.5% to 16.7%); NearStats' own split failed the plausibility check. Volume is 22.0% below its 30-day record.
Over 122 fill participations the bootstrap 90% interval is 2,898 to 4,772, which straddles the 3,500 trigger. The fills span 4.5 h of settlement, under the 60 min minimum: one short window is serially correlated and its interval covers only within-window noise, so the trigger is reported as inconclusive, neither breached nor clear. On the deposit side, caller HHI is 3,056 (about 3.3 equal-sized callers) and AI agents account for 50.0% of 175 sampled receipts.
Headroom is execution_bench.headroom_by_archetype_bps, the median over comparable cells (cross_vm n=3, native_btc n=3, evm_l2 n=3) on basis dest_units. (out_1click - out_best_competitor) / out_best_competitor x 1e4 on destination units for the same raw input; no oracle
Coverage is supply_physics.sink_coverage_ratio: buyback 18,099 NEAR (DefiLlama holders revenue at spot) + gas burn 938 NEAR (within-epoch supply drift) + state lock 342 NEAR (archival growth), over measured issuance 89,757 NEAR a day. Break-even is supply_physics.breakeven_daily_volume_usd; crossover odds are sotp.crossover_month_distribution.*.p_crossover_within_36m.
The state lock is near_rpc.tracked_sinks[intents.near].storage_locked_near = storage_usage × storage_amount_per_byte; growth is measured against archival state over 31.1 days.
Market-implied solves are market_implied.solves: Brent root-finding on the deterministic engine's expected path (deterministic_sotp expected path (mean-matching GBM); one input moved at a time, all others at the snapshot base), each required value compared with evidence.value. Relative value is relative_value.coverage (rank 1 = lowest multiple) and implied_fdv_band.fees. Peer 'revenue' follows each chain's own DefiLlama methodology (peers[].revenue_methodology), so revenue multiples compare different cash-flow definitions; with n = 3 peers the median is a reference point, not a distribution.
Monte Carlo percentiles are sotp.fdv_percentiles (model output, no single payload). Market FDV basis: spot price x on-chain total_supply at the snapshot block, price from nearstats price.currentUsd via intents_forensics.price.current_usd. Total supply 1.3116B NEAR at the pinned block.
Take rates are intents_forensics.fees.gross_take_bps and net_take_bps over aligned 30-day DefiLlama series. Solver HHI is intents_forensics.solver_concentration.hhi_solver over NEP-413 signers of execute_intents fills (122 fill participations; interval hhi_solver_ci90); the cartel trigger reads bear_case.scenarios[solver_cartelization].trigger_status = inconclusive. Caller HHI is hhi_external over 8 NearBlocks pages. Confidential share status: intents_forensics.volume.confidential_share_status = onchain_estimate.
Not on the evidence so far. 1Click delivers more than the best competitor on cross-VM (+0.98 bps), but trails on native BTC (−2.97 bps vs Relay) and EVM L2 (−1.02 bps vs Across). The advantage is corridor-specific. Carrying the market price on take rate alone would need a starting public take of 56 bps, against 15 bps today and a 3.00 bps competitive ceiling, so the price cannot rest on fee expansion.
Execution tabDeposit-side callers are highly concentrated: HHI 3,056, roughly 3.3 equal-sized callers, with AI agents at 50.0%. The solver side is at least as concentrated: HHI 3,197 (90% CI 2,898 to 4,772) across 6 solvers in 114 sampled fills, over 4.5 h: too short a window to grade the cartel trigger, so it is reported as inconclusive.
Solver auditStorage staking. intents.near alone holds 136,471 NEAR, growing +310.0 NEAR a day; 8 tracked contracts hold 283,182 NEAR.
Supply tabDaily Intents volume of $608.0M under the current issuance cap, or $386.0M under the SVRN glidepath, at today's take and capture. Trailing volume is $125.8M a day, and sinks cover 21.6% of issuance.
Crossover modelQuestion 1. Identical dry quotes go to 1Click, Relay, LI.FI and Across across three corridor archetypes and three clip sizes. Headroom = (1Click output − best competitor output) / best competitor output, on physical destination units, in basis points, with no price oracle in the comparison. Snapshot definition: (out_1click - out_best_competitor) / out_best_competitor x 1e4 on destination units for the same raw input; no oracle
Question 2. Predecessor accounts of recent intents.near receipts are classified (AI escrow agents, bridge relayers, implicit accounts, named users) for the deposit-side hhi_external and the AI-agent share. NEP-413 signers of execute_intents fills give hhi_solver, the index the cartel trigger reads. 175 receipts and 114 fills this run; both indices move between runs.
Question 3. storage_usage from view_account × storage_amount_per_byte from protocol_config; growth from archival state.
Question 4. Solve for the daily volume at which buybacks, gas burn and state-lock growth offset realized issuance under each regime (supply_physics.breakeven_daily_volume_usd), then run the same system stochastically.
Four collectors query public, zero-key endpoints. Every response, including error bodies such as “no liquidity” or “route not enabled”, is hashed with SHA-256 and timestamped.
Issuance is read from supply jumps at epoch boundaries, not from the nominal formula. State-lock growth comes from archival state; take rates from aligned daily fee and volume series.
Supply physics under two issuance regimes, a three-engine sum-of-the-parts valuation with a seeded Monte Carlo, solves for what the market price requires of each input, and falsification scenarios on the same engine.
One command regenerates the snapshot, the manifest of payload hashes and the IC memo. The terminal and this page render only what those files contain.
| Source | What it provides | Payloads |
|---|---|---|
| NEAR mainnet RPC | Block, protocol config, validators, contract storage, epoch-boundary supply | 37 |
| NEAR Intents 1Click | Token universe and dry executable quotes | 36 |
| LI.FI | Competitor quotes from an aggregator | 33 |
| Relay | Competitor quotes | 33 |
| NearBlocks | intents.near receipts and caller classes | 31 |
| DefiLlama | Fees, revenue, holder revenue, volume, peers | 22 |
| NEAR archival RPC | Historical contract storage, for state-lock growth | 18 |
| Across | Competitor fees; non-EVM routes return unsupported | 15 |
| CoinGecko | Peer FDV and market cap | 2 |
| NearStats | Public and confidential volume, price history | 2 |
| Total | 15.61 MB; 26 error responses kept as evidence | 229 |
| Claim and caveat | Grade | Measured value and source |
|---|---|---|
| NEP storage staking is a measured on-chain supply sink (1 NEAR per 100 KB) growth 30,999,489 B/day = 310.0 NEAR/day over 31.1 d (archival); a 0.01 NEAR/KB lock is released if state is deleted, so it is a sink while state grows | Verified on-chain High confidence | intents.near storage_usage 13,647,147,373 bytes = 136,471 NEAR locked; 283,182 NEAR across 8 tracked contracts near_rpc/view_account:intents.near sha256 452dc535876a... @ block 224,323,532; storage_amount_per_byte from EXPERIMENTAL_protocol_config |
| Realized issuance is measured from epoch-boundary supply jumps, not assumed from the 2.5% cap second independent jump 25,422.56 NEAR; within-epoch drift -267.07 NEAR/epoch is the net gas burn | Verified on-chain High confidence | 25,544.21 NEAR minted at epoch start 224,308,337 x 1,283.4 epochs/yr = 32,783,583 NEAR/yr (2.4996%), -0.02% vs nominal 32,789,360 near_rpc/block:224308337 sha256 93bb2a2224aa... and its predecessor by prev_hash; previous epoch start 224265137 |
| Cross-VM execution headroom: 1Click out-delivers Relay/LI.FI net of fees on ETH USDC -> SOL USDC $500: +1.24 bps vs relay (rank 1/3); $50,000: +0.98 bps vs relay (rank 1/3); $250,000: +0.98 bps vs relay (rank 1/3); no 1Click capacity failures | Verified via live API Medium confidence | median headroom +0.98 bps over 3 comparable cells (dest_units); competitor ceiling 3.00 bps oneclick/relay/lifi/across quote records 2026-11-13T06:05:11.348Z |
| Native BTC corridor: 1Click delivers more BTC per USDC than LI.FI/Relay on ETH USDC -> native BTC clip-dependent: $500: -18.30 bps vs relay (rank 2/3); $50,000: -2.97 bps vs relay (rank 2/3); $250,000: +8.91 bps vs lifi (rank 1/2); LI.FI's best BTC route is often NEAR Intents itself (tool `near`); venue USD marks of BTC differ by tens of bps and are not used | Not supported Medium confidence | 1Click delivers LESS BTC per USDC than the best competitor (relay): median headroom -2.97 bps over 3 cells (dest_units); the claim is not supported by this run oneclick/lifi/relay quote records 2026-11-13T06:05:11.348Z |
| EVM-L2 stablecoin routing is commoditized: Across/Relay match or beat 1Click on Base USDC -> Arbitrum USDC (honest negative) $500: -0.99 bps vs across (rank 3/4); $50,000: -1.02 bps vs across (rank 3/4); $250,000: -1.02 bps vs across (rank 3/4); the moat, where it exists, is corridor-specific (cross-VM, native assets), not EVM-L2 stablecoins | Verified via live API Medium confidence | median headroom -1.02 bps over 3 cells (dest_units; 1Click trails) oneclick/across/relay/lifi quote records 2026-11-13T06:05:11.348Z |
| Deposit/caller concentration on intents.near is measurable and AI-agent-labeled flow is material 175 unique receipts over 22 s; a point sample of the live receipt stream, not a market-share estimate. AI-agent is a name-pattern classification (*.ai.near, *.agent.near); escrow.ai.near has no deployed contract, so the label is a heuristic. This HHI is over deposit sources (bridge callbacks, direct callers), not solvers | Derived Low confidence | deposit/caller HHI_external 3,056 [90% CI 1,944-4,625] (N_eff 3.27), ex-AI-agent 2,222; AI-agent share 50.0% (>= 10% materiality line), bridge share 25.0% nearblocks/txns:intents.near:p1 sha256 0c2f1a3c09f3... ... p8 |
| Solver concentration on intents.near is measured from execute_intents fills, not inferred from deposit callers 462 execute_intents receipts decoded; settlement hub 51e8f94d77b5... (146 counterparties) routes most swaps via transfer legs; see intents_forensics.solver_concentration.method. The CI is within-sample (one window of minutes), so it understates long-run uncertainty | Verified via live API Medium confidence | HHI_solver 3,197 [90% CI 2,898-4,772] (N_eff 3.13) over 122 fill participations in 114 fills (37 atomic token_diff fills, 77 hub-settled legs), 6 distinct solvers; CI straddles the 3,500 cartel threshold (inconclusive) nearblocks/txns:intents.near:p1 sha256 0c2f1a3c09f3... ... p8 + 8 method-filtered execute_intents page(s); args.signed[].payload.message (NEP-413 signer_id) |
| Confidential (dark-pool) execution is a material share of 30d NEAR Intents volume nearstats' split was rejected by the plausibility guard; the share is estimated from the public treasury boundary instead. Biases: same-hour shield/unshield netting and in-shard turnover push it down, solver inventory pushes it up; 24h window | Derived Low confidence | 16.7% (band 13.5%-16.7%) estimated on-chain: Confidential Intents Treasury 24h inflow $29.4M / outflow $27.6M against 24h volume $176.4M; >= 10% materiality line near_rpc/confidential:balances:head sha256 198c8b7f149c... treasury balances in intents.near (NEAR mainnet RPC) |
| Confidential TVL backs the dark-pool engine (Engine 2) Engine 2 is flagged assumption_heavy and its terminal multiple carries the confidential_multiple_haircut (0.7); the confidential take (20 bps) is an assumption | Verified on-chain High confidence | $132.6M TVL read on-chain from the Confidential Intents treasury in intents.near (NEAR-reported via nearstats: $137.9M) near_rpc/confidential:balances:head sha256 198c8b7f149c... intents.near mt_batch_balance_of (treasury) |
| Buyback capture alpha = holders revenue / gross fees is measured, so V x tau x alpha / P is a live sink DefiLlama revenue methodology: NEAR's net captured revenue, measured as NEAR/wNEAR swept into its revenue wallets (reconciles with revenue.near.org and the on-chain buyback). Split by destination wallet into front-end affiliate revenue and Other. Transfers between those wallets are excluded. | Derived Medium confidence | alpha 30.0%; gross take 16.06 bps, net take 4.82 bps; buyback sink 18,099 NEAR/day = 21.6% coverage incl. burn and lock defillama/fees:near-intents:dailyHoldersRevenue sha256 826a018a3a3c... / defillama/fees:near-intents sha256 562751003b45... |
| Chain Signatures (v1.signer) is a growing second usage engine beyond Intents sampled estimate: each window covers minutes; bridge callers serve cross-chain transfers that overlap Intents settlement, so the engine is not fully independent of Intents; storage lock 11.6 NEAR is small. sign() attaches 1 yoctoNEAR (median of sampled requests), so the contract takes no material per-request fee; the requester's cost is gas. | Not supported Low confidence | 15,597 sign requests/day (7d) vs 12,918/day (30d); ln-trend +39% per 30 d [90% CI -9% to +113%], t = 1.28; top caller omni.bridge.near 42% of 382 sampled sign receipts, *.bridge.near callers 66%; gas 0.0021 NEAR per request (inconclusive: growth not shown at t >= 2) nearblocks/txns:v1.signer:before_block:* (12 windows of 100 receipts, 1,200 receipts) + nearblocks/txns:v1.signer:count |
| The volume engine (GBM with measured sigma) is calibrated out of sample: P10-P90 covers ~80% of realized 30-day volume Tests the volume distribution only; the take rate, holder capture, confidential share, multiple and discount rate are not backtested. 3m and 6m forecasts from monthly origins overlap, so their effective sample is smaller than n. | Not supported Low confidence | 1m 11/20 = 55% inside P10-P90 (binomial p 0.01); 3m 9/17 = 53% inside P10-P90 (binomial p 0.01); 6m 7/14 = 50% inside P10-P90 (binomial p 0.01); 1m median realized / forecast median 1.37x (under-prediction); 1m coverage 20% for origins before 2026-01-16, 90% from it; 20 monthly origins 2025-03-22 to 2026-10-13 defillama:dexs/near-intents totalDataChart (raw/defillama__dexs_near-intents.json) sha256 3a66e2dab201... |
| A block-bootstrap volume engine (fat tails, autocorrelation) is calibrated out of sample where GBM is not Inside P10-P90 against an 80% target, exact two-sided binomial p. Low power: at n = 20 the test accepts 14 to 19 hits and rejects a 70%-coverage engine only 39% of the time; 3m and 6m forecasts overlap. Block length (Politis-White) and path count were fixed before coverage was computed; nothing is tuned. The valuation still uses the GBM; this row grades a parallel variant and tests the volume distribution only. | Not supported Low confidence | block bootstrap 1m 18/20 (90%, p 0.40), 3m 17/17 (100%, p 0.03), 6m 13/14 (93%, p 0.33); GBM 1m 11/20 (55%, p 0.01), 3m 9/17 (53%, p 0.01), 6m 7/14 (50%, p 0.01); bootstrap at the GBM's expected path (shape only) 1m 12/20 (60%, p 0.04), 3m 6/17 (35%, p <0.01), 6m 0/14 (0%, p <0.01); bootstrap band 3.3x the GBM's log width at 1m; 20 monthly origins 2025-03-22 to 2026-10-13 defillama:dexs/near-intents totalDataChart (raw/defillama__dexs_near-intents.json) sha256 3a66e2dab201...; valuo.models.volume_bootstrap (seed 20261114, 10000 paths per origin) |
| NEAR Intents' realized gross take rate is not trending to zero as volume grows Graded on the trailing-365d trend only: a decline is counted when the Newey-West 90% interval and the non-overlapping-window Student-t interval both sit below zero; DERIVED needs no such decline and a latest rolling take above the model floor. Gross take includes affiliate fees kept by frontends; NEAR's captured share is the net / holder take. Rolling points overlap, so an interval that includes zero is not proof of a floor. The full-history trend is not used (fee ramp from zero in 2025). | Not supported Low confidence | rolling 30d gross take 16.06 bps on 2026-11-11 vs 16.45 bps on 2025-11-12; trend -30.3%/yr, 90% CI -42.3% to -15.8% (excludes zero; Newey-West 60d lag, 365 overlapping points); floor 0.5 bps, 0 of 365 trailing points below it; non-overlapping 30d windows (n=13) -26.6%/yr, Student-t 90% CI -43.1% to -5.3%; trailing 180d trend +22.4%/yr, CI -11.8% to +69.9% defillama/fees:near-intents:dailyFees sha256 562751003b45...; defillama/fees:near-intents:dailyRevenue sha256 1241564a028c...; defillama/fees:near-intents:dailyHoldersRevenue sha256 826a018a3a3c...; defillama/dexs:near-intents sha256 3a66e2dab201... |
| 1Click quotes institutional size: at least $1M on every benchmarked corridor Largest size quoted on an eight-rung ladder of dry quotes, one run. A dry 'no liquidity' is solver quoting policy at that moment, not proven depth, and a dry quote is not a fill. | Not supported Low confidence | ETH USDC -> SOL USDC: 1Click $500K, best competitor LI.FI $5M; ETH USDC -> native BTC: 1Click $1M, best competitor LI.FI $1M (routed through NEAR Intents); Base USDC -> Arbitrum USDC: 1Click $500K, best competitor LI.FI $5M (1 of 3 corridors with an attempted 1Click rung at or above $1M) 1Click, Relay, LI.FI, Across dry quotes; ledger records ladder:<venue>:<corridor_id>:<size_usd> (81 responses), run 2026-11-13T06:05:14.905Z |
| NEAR Intents is gaining share of cross-chain routing fees Fee share, not volume share; adapter fee definitions differ; the pool is the protocols DefiLlama covers. A share trend is DERIVED from live fee series, never VERIFIED. | Not supported Low confidence | 20.4% of the six-protocol fee pool (30 days to 2026-11-11), rank 2 of 6 behind Relay at 55.4%; share at window start 35.0% (202-day common window); trend -6.59 pp per 30 days, 90% interval -9.50 to -3.67; ex-Relay 45.7%, trend -0.89 pp per 30 days, 90% interval -1.65 to -0.14 (share is falling) defillama summary/fees dailyFees: near-intents, relay, across, debridge, chainflip, thorchain-dex |
| The model's terminal multiple cap sits inside the observed peer range of FDV/revenue Rule: DERIVED when the cap is at or below the 75th percentile of the wide set and n_peers >= 8. Here the cap is inside the observed range and at or below its upper quartile. FDV/revenue on 30 days annualised, one snapshot; revenue definitions differ by chain adapter; peers ranked by revenue and fixed before any multiple was computed (inclusion_rules, excluded). | Derived Medium confidence | cap 60x vs 15 chain tokens: 2 of 15 trade at or below it (13th percentile); P25 249x, median 531x [bootstrap 90% 274x to 999x], P75 1,748x, range 3.1x to 6,508x; NEAR L1-only 3,907x (87th percentile); without flagged rows: median 600x, n 10 defillama/overview:fees:dailyRevenue sha256 78e5a4b19413... + coingecko/coins_markets_wide sha256 99dcae049583... |
Net take rate, 30 days is 4.82 bps; the trigger is < 3.00 bps. Monitor
Protocol inflation cap is 2.50%; the trigger is > 1.60%. Monitor
30-day volume vs the 30-day record is −22.0%; the trigger is ≤ −40.0%. Monitor
Solver concentration (HHI) is 3,197; the trigger is > 3,500. Inconclusive: sample: 122 solver fill participations over 16,269 s. Monitor
Base value: bear_case.base_fdv_p50 = $1.98B (deterministic at median inputs; not a Monte Carlo percentile). It differs from the Monte Carlo median by construction, and equals the expected-path value the market-implied solves in section 1 start from.
Early-warning metrics, by key: Aggregator bypass: public take races to 1 pip reads net_take_bps_30d; Governance stalemate: 2.5% issuance forever reads max_inflation_rate; Post-incentive volume decay (NEAR@3.33 unwind) reads volume_30d_vs_record_30d; Solver cartelization: HHI_solver > 3,500, 1Click loses the spread reads hhi_solver. Not graded this run: solver_cartelization. The count is bear_case.triggers_breached of triggers_evaluated; an inconclusive trigger (trigger_status = inconclusive, trigger_breached null) is counted as neither breached nor clear.
# from the repository root $ make reproduce $ make test $ make dev # equivalent to make reproduce $ cd quant && uv run python -m valuo.reproduce
To verify offline, shasum -a 256 artifacts/latest_snapshot.json must equal snapshot_sha256 in the manifest. The terminal recomputes it on load.
15 warnings raised by the pipeline, printed as recorded.