Phase 2: Manager Underwriting & Decision

Pareto Technologies — Nomos (Diophan)

Review Date: 2026-03-23 Reviewer: CIO Agent Phase 1 Report: reports/pareto_diophan/phase1_analysis.md Qualitative Sources: Nomos ETH Breakout Scalping pitch deck (Oct 2025), IC Memo (2026-02-16), Data Room CSV/API files, Jan 28 DD call transcript, Feb 16 DD call transcript, independent Bybit API verification (9/9 accounts), Strategy Statistics (Kairos) xlsx, "Read this first" data room guide, sleeve-level correlation and metrics analysis

Recommendation

Decision: DEFER

• Proposed initial allocation: 0.50–1.00% NAV pilot — ONLY after remaining hard gates are closed (see Funding Conditions below)
• Conditions to scale:
  1. Close all 7 hard gates (Section 9 of IC Memo) including manager-signed account-to-sleeve mapping, reconciliation memo, DDQ/PPM/legal docs, trade blotter, capacity study, and incident logs
  2. Minimum 6 months of independently verified live performance in deployed sleeve
  3. Rolling 12m Sharpe remains ≥ 1.5
  4. No additional fraud-screening flags triggered
  5. Independent operational due diligence (ODD) by third party
• Key risks:
  1. Alpha decay — statistically significant declining trend (p < 0.0001) from ~300% (2021) to ~50–100% (2023–24)
  2. Track record provenance — much of 2021–2024 data may be backtest/simulated; live API verification limited to ~2 years (Bybit endpoint limit)
  3. Key-person risk — sole founder Junaid as PM and system architect; no succession plan
  4. Exchange counterparty risk — assets held on centralized exchanges (Binance/OKX/Bybit); no independent custodian (FTX precedent)
  5. Regulatory vacuum — no SEC registration, no Form ADV, no known regulatory oversight
  6. Capacity uncertainty — 250x monthly turnover creates severe market-impact scaling; manager's $200M capacity claim appears aggressive
• Review cadence: Monthly during pilot; quarterly thereafter

Step 1: Four Ps Scoring

Pass gate: All 4 dimensions ≥ 3, no dimension = 1. Result: FAIL — Process (2) and People (2) are below threshold. Two dimensions failing the ≥ 3 gate.

Philosophy Deep Dive

• Source of edge: Microstructure alpha from detecting price compression zones on ETH perpetual contracts and entering directional limit orders immediately after confirmed breakouts. Targets 0.4–0.7% moves with average 57-minute holding period. Win rate ~43% with avg win (+1.90%) exceeding avg loss (-1.41%) in high-risk mode.
• Edge persistence: Eroding. Phase 1 detected statistically significant alpha decay (slope p < 0.0001, R² = 0.47 on trend). Annual Sharpe declined from 29.67 (2021) → 6.60 (2022) → 2.40 (2023) → 4.99 (2024). Consistent with Adaptive Markets Hypothesis — crypto quant alpha is being competed away [B1]. The 2024 recovery provides some comfort that the team can adapt, but the long-term trend is unfavorable.
• Competitive moat: Proprietary "Alpha Shifter" execution middleware with <100ms signal-to-fill latency. Direct API integrations across 3 major exchanges. Smart order routing, adaptive cancel/replace logic, kill switches. Co-located in AWS Tokyo for Asia-dominant order flow. Not trivially replicable, but moat is execution speed (perishable) rather than structural/informational.
• Arbitrage risk: HIGH. Multiple well-capitalized crypto quant firms (Jump, Wintermute, Tower, Alameda successors) operate in the same microstructure space. The alpha decay trend confirms ongoing arbitrage of the edge. As crypto markets mature and institutional participation grows, microstructure inefficiencies will continue to diminish.

Process Deep Dive

• Systematic vs. discretionary: Fully systematic. All trade decisions automated. Risk parameters embedded at middleware level. Daily loss limits and rolling drawdown brakes auto-pause the system. However, cross-call inconsistency on intervention policy (Jan 28: "operator pause context described" vs Feb 16: "fully automated/no manual intervention emphasized") raises questions about actual discretionary override frequency.
• Documentation quality: POOR. No formal DDQ. No PPM. No governing documents. No SMA template. Data room consists of CSV files, API keys, and a brief pitch deck. This is severely below institutional standards for a fund claiming $64–84M AUM.
• Backtest discipline: DSR = 1.0 at 50 trials (Phase 1). However, this was computed on the full 2021–2024 Diophan series, much of which is likely backtest. The "Read this first" document explicitly states live data begins Jan 1, 2025 for some files. Earlier data is unverifiable beyond the Bybit 2-year API endpoint limit.
• Execution infrastructure: Credible. AWS Tokyo co-location, direct exchange API integrations, smart order routing, failover protocols, performance telemetry. The infrastructure description is consistent with a real low-latency crypto trading operation.

People Deep Dive

• Key-person risk: CRITICAL. Junaid is the sole identifiable team member across all materials (pitch deck, data room, calls). He appears to serve as PM, system architect, and primary investor contact. No evidence of additional investment professionals, operations staff, or technology personnel.
• Team depth: Unknown. No org chart provided. All communications route through single contact.
• GP co-invest: Unknown — not disclosed in any available materials.
• Compensation alignment: Unknown — no fee or compensation structure documented.
• Succession plan: No evidence exists. If Junaid becomes incapacitated, there is no known backup to maintain or operate the system.

Portfolio Deep Dive (from Phase 1)

• % Idiosyncratic variance: 99.4% — far above 75% threshold [cc_B1_001]. R² < 2% across all specifications including 6-factor model (BTC, ETH, SPY, GLD, TLT, USD). This is definitively NOT a factor bet.
• Factor stability: Rolling BTC beta oscillates ±0.06 around zero with σ = 0.058. No persistent directional tilt. No material style drift detected in factor loadings [R1:rolling_factor_analysis_drift].
• Concentration: HIGH at asset class level. Primary exposure to ETH perpetual contracts on 3 centralized exchanges. SOL and DOGE sleeves provide partial diversification within crypto. All concentrated in a single strategy type (breakout scalping).
• Style drift: None detected in factor space. However, alpha decay trend (p < 0.0001) represents de facto strategy degradation — the return-generating capacity is declining even though factor exposures remain stable.

Step 2: DDQ Completeness Check

Sections complete: 0/13 (0 complete; 5 partial; 8 missing) Action items:

1. CRITICAL: Provide a completed DDQ covering all 13 sections
2. CRITICAL: Provide PPM, subscription documents, and governing agreements
3. CRITICAL: Provide fee schedule with management fee, performance fee, hurdle, HWM, and crystallization terms
4. CRITICAL: Provide organizational chart with all team members and roles
5. Provide Form ADV or equivalent regulatory filings (or formal explanation of regulatory status)
6. Provide audited financial statements or engage independent auditor
7. Provide business continuity and disaster recovery plan
8. Provide formal counterparty risk framework and exchange exposure limits
9. Provide NAV calculation and valuation methodology document
10. Provide at least 3 professional references for background verification
11. Provide manager-signed account-to-sleeve mapping for all marketed composites
12. Resolve cross-call inconsistencies (intervention policy, turnover, risk tiers, AUM)

Step 3: Fraud Risk Screening

Additional Dimmock-Gerken concerns:

• AUM inconsistency: $84M (Jan 28 call) vs $70M (Feb 16 call) vs $64M (pitch deck) — three different figures across three touchpoints
• Cross-call contradictions: Intervention policy, turnover definitions, risk-tier parameters all stated differently across calls
• No legal documentation: Despite multiple touchpoints, no PPM/DDQ/legal docs provided
• Unregistered manager: No SEC/CFTC/NFA registration eliminates entire regulatory oversight layer

Flags triggered: 3 (Self-custody + No auditor + Suspicious track record construction) Gate result: DEFER — 2+ flags require deferral pending investigation [cc_R3_005, R3]

Step 4: Fee Impact Analysis

Note: The Diophan track record is stated as "net-of-fee" per Phase 1 data quality notes. If these are truly net-of-fee returns, the implied gross returns are higher. Fee drag analysis below uses two scenarios.

Scenario A: If returns are GROSS (assumed 2/20 applies)

Scenario B: If returns are NET (implied gross with 2/20)

Gate: Fee drag < 30% of gross alpha. Result: PASS (conditional) — Fee drag is 25–27% of gross alpha across scenarios, below the 30% threshold. However, fee terms are entirely unconfirmed. This analysis uses assumed standard 2/20 terms. If actual fees are higher (e.g., 2/30 or 3/30 as some crypto funds charge), the result could flip. This gate cannot be formally closed until fee terms are disclosed. [R2, B3]

Step 5: Leverage Sustainability

Interpretation [B1 Ch10, cc_B1_004]: The feasible leverage band [0.89×, 2.69×] indicates the strategy's return/risk profile can sustain a standard 1× allocation and has headroom up to ~2.7× before the drawdown constraint binds. The strategy does not require excessive leverage to meet conservative return targets.

Critical caveat: The manager states the system can scale up to 15× leverage per trade on perpetual contracts. This is the POSITION-level leverage, not the PORTFOLIO allocation leverage. Because positions are short-lived (~57 min) and risk-per-trade is capped at 0.5–1.0%, the effective portfolio leverage at any given moment is much lower than 15×. However, simultaneous adverse moves across open positions at high leverage could generate losses well beyond the leverage bounds analysis.

Gate: L_lower ≤ L_upper AND current leverage within bounds. Result: PASS (with caveat) — Leverage band exists and a 1× allocation is feasible. Per-trade leverage of 15× is a risk to monitor but is partially mitigated by position sizing discipline and short holding periods. Effective portfolio-level leverage data not available.

Step 6: Type I/II Error Framing

Track record credibility assessment:

Track record length (credible live): ~1.0–2.0 years (borderline for conviction) Recommended action: Reduced initial size — pilot at 0.50–1.00% NAV, scale only after 6+ months of independently verified live performance [B3]

Step 7: Capacity Analysis

Model assumptions: 250× monthly turnover (normal risk mode, per pitch deck). Square-root market impact model (Almgren-Chriss): effective impact scales as √(AUM/AUM_base). Base impact estimated at 10 bps at $64M AUM.

CRITICAL FINDING: At the stated 250× monthly turnover with standard market-impact assumptions (10 bps base), the model implies the strategy cannot profitably operate at ANY AUM above ~$40M. This is clearly inconsistent with the observed profitable track record at $64M+ AUM.

Resolution of the paradox: The strategy already operates profitably at current AUM, meaning either:

1. Effective market impact is sub-2 bps — plausible for limit orders in ETH perps ($5–15B daily volume), which would imply impact costs of ~60% of AUM annually. This is still very high but potentially covered by the ~237% gross returns.
2. "250× turnover" is overstated or measured differently — e.g., notional turnover vs. risk-equivalent turnover, or includes cancelled/replaced orders.
3. Reported returns already embed friction costs — the pitch deck states results "reflect full fee, slippage, and latency considerations."

Sensitivity analysis at $200M (manager's claimed ETH capacity):

Estimated capacity limit: Highly sensitive to effective impact. At 1 bps base impact, capacity extends to ~$200M before net alpha declines meaningfully. At 2 bps, breakeven is ~$130M. At ≥5 bps, the strategy is capacity-constrained to <$50M. Current AUM as % of capacity: Unknown — depends on true market impact per trade, which requires trade blotter analysis.

Action required: Obtain full trade blotter with venue, timestamp, fill price, and theoretical fair price to compute realized market impact. Without this data, the $200M capacity claim cannot be validated. [R2]

Step 8: Liquidity Profile

Rationale: The strategy trades ETH/SOL/DOGE perpetual contracts on major centralized exchanges. Average holding period is ~57 minutes. All positions are intraday — no position remains open past the duration threshold. The portfolio is effectively 100% liquid on a daily basis (exchange withdrawal limits permitting).

Investor terms: NOT AVAILABLE — No PPM, no subscription documents, no lock-up/redemption terms disclosed. This is a critical missing piece.

Liquidity mismatch: CANNOT ASSESS — Without investor terms, we cannot determine if portfolio liquidity matches investor redemption rights. The underlying assets are extremely liquid, but if the fund structure imposes lock-ups (common in crypto funds), this could create synthetic illiquidity.

Exchange-level liquidity risks:

• Withdrawal limits: Centralized exchanges impose daily withdrawal limits that could constrain large redemptions
• Exchange downtime: Major exchanges have experienced extended outages during high-volatility events
• Regulatory freeze: Exchanges are subject to regulatory action that could freeze assets (Binance SEC action, 2023)
• Smart contract risk: If settlement involves on-chain transfers, smart contract vulnerabilities apply

Quantitative Summary (from Phase 1)

Sizing Logic

Red Flags Summary

Monitoring Plan

Stop-loss thresholds (per IC Memo pilot limits):

• Soft drawdown trigger: -6% (increase monitoring frequency to daily)
• Hard drawdown trigger: -10% (halve risk / redeem 50%)
• Stop trigger: -12% (full redemption unless IC override)
• Governance trigger: any undocumented policy deviation

Review cadence: Monthly during pilot phase; quarterly after scale-up Escalation triggers:

1. Rolling 12m Sharpe falls below 1.0
2. Drawdown exceeds -10%
3. Any unresolved API reconciliation discrepancy > 300 bps
4. Exchange regulatory action affecting any deployment venue
5. Loss of contact with manager for > 48 hours
6. Any undisclosed strategy or parameter change

Funding Conditions (Hard Gates — from IC Memo, updated)

The following gates must ALL be closed before any capital is deployed, even at pilot level:

1. Manager-signed account-to-sleeve mapping and weighting policy for all marketed 9M variants
2. Reconciliation memo covering Jan 28 vs Feb 16 inconsistencies (intervention policy, turnover definitions, risk-tier definitions, AUM point estimates)
3. DDQ package and legal docs: PPM, fee schedule, governing docs, SMA template
4. Full order/trade blotter with venue, fee tier, slippage fields for 2025 and 2026 YTD
5. Capacity study with explicit depth/stress assumptions per venue
6. Incident/control logs for stop events and discretionary overrides (if any)
7. Updated data room exports for 2026 YTD, reconciled to API results already pulled

Questions for Manager Follow-Up

1. Fee structure: What are the management fee, performance fee, hurdle rate, high-water mark, and crystallization terms? Please provide a formal fee schedule.
2. Legal entity: What is the legal structure (LP, LLC, SMA)? Where is the fund domiciled? Who is the legal counsel?
3. Team composition: Who else works at Pareto Technologies? Please provide an org chart with roles, tenures, and backgrounds.
4. Succession plan: If you (Junaid) were unable to manage the strategy, what is the contingency plan? Who would operate the systems?
5. Track record boundary: Which portions of the Diophan track record are live trading vs. backtest/simulation? Please provide a clear date for the transition to live trading for each sleeve.
6. AUM reconciliation: Please explain the variation in stated AUM ($84M → $70M → $64M) with a dated AUM bridge showing flows and PnL.
7. Intervention policy: Please clarify: is the system fully automated with zero manual intervention, or are there circumstances under which you manually override the system? Please provide incident logs for any manual interventions in 2025–2026.
8. Turnover definition: The pitch deck states 250× monthly turnover. Please clarify: is this notional turnover, risk-equivalent turnover, or something else? How many individual round-trip trades does the system execute per day on average?
9. Effective market impact: What is your measured average market impact per trade? Please provide fill price vs. mid-price analysis from the trade blotter.
10. Capacity: Your pitch deck claims $200M ETH capacity. What analysis supports this? At what AUM level do you expect meaningful alpha erosion?
11. Custody: Would you be willing to work with an independent third-party custodian? If not, what exchange-level protections exist for investor capital?
12. Audit: Would you engage an independent auditor to audit the fund's financial statements? If an audit already exists, please provide the most recent report.
13. Regulatory status: Are you registered with any financial regulator (SEC, CFTC, NFA, FCA, MAS, etc.)? If not, do you plan to register?
14. Scalar risk tiers: Please confirm that the High/Medium/Low risk variants are scalar multiples of the same underlying signal, not independent strategies. What is the exact scaling factor?
15. January 2026 drawdown: API data shows -7% to -14% drawdowns across accounts in Jan 2026. What caused this? Has the strategy recovered? How does this compare to your risk model's expectations?

Files Generated

Appendix: Sleeve-Level Performance Summary (Data Room Analysis)

Note: ETH 9M RM+ATR High/Medium/Low have pairwise correlation = 1.000, confirming exact scalar relationship. These are NOT independent strategies.

Underwriting conducted using Phase 1 quantitative profile plus available qualitative materials. Multiple sections marked with information gaps require follow-up before final IC decision. The DEFER recommendation is driven by: (1) Four Ps gate failure (Process=2, People=2); (2) 3 fraud screening flags triggered; (3) 0 of 13 DDQ sections complete; (4) track record provenance concerns; and (5) unresolved cross-call inconsistencies. This is consistent with the IC Memo recommendation of 2026-02-16.