Public memo · cryptographically signed

Risk-Parity Hedge Falls Short of Bank-Crisis Thesis

Run 25/5/2026
risk_parity
6 assets
monthly rebalance
signed · wRKTmyEG59lkKWt3
Marking position to market…
Sharpe
0.46
Risk-adjusted
Sortino
0.50
Downside-adj.
Ann. return
7.09%
CAGR
Max DD
-9.1%
Peak-to-trough
Vol (ann)
6.8%
Standard dev
CVaR 95%
-0.64%
Tail loss
Win rate
56.2%
Daily positive
Total return
60.1%
Since inception
Equity curve · INR
Portfolio vs. NIFTYBEES.NS
Portfolio NIFTYBEES.NS
Selected window
2019-05-27 2026-05-25 · 2555d
Portfolio+60.09%
NIFTYBEES.NS+116.86%
Alpha-56.78%
Ann. alpha-11.29%
2019-05-27drag a handle · drag the band to slide the window2026-05-25
2019-052021-022022-112024-082026-05
Drag the slider handles to zoom the chart. Click-drag on the chart itself to mark a sub-window and read its gain / alpha.
Multi-model review
Disagreement · 38%

A second frontier model re-reads the same evidence. Thesis stance is what gets signed; the independent row is for audit and risk review — track accuracy over time on Model scorecard.

Primary · Signed
Mixed evidence
Evidence 8/10
Independent · OpenAI (gpt-5.5)
Not supported
Evidence 7/10
Evidence gap · Δ -1 vs primary · Independent model is more cautious about evidence strength.
Independent rationale: I would be harsher than the primary analyst: this is not a hedge or clean implementation of the bank-crisis thesis, because it remains structurally long the exact assets expected to suffer and only reduces them after volatility has already risen. The backtest's modest Sharpe and shallow drawdown are reassuring for a diversified low-volatility allocation, but they do not prove resilience in the specific forward scenario of rupee stress, bank equity repricing, and correlation breakdown.
Top independent risk: Correlation and liquidity regime shift: HDFCBANK, ICICIBANK, BANKBEES, Indian equities, and even defensive offsets could move together before the monthly inverse-volatility process has time to cut exposure.
Read in:

Research simulation only · Not investment advice · Not a trade recommendation · Past backtests do not predict future results.

Before investing, consult a SEBI-registered research analyst or investment adviser. Attevia is not registered and does not provide personalised advice.

Thesis stance · Mixed evidence
Evidence 8/10

Evidence is mixed or benchmark-relative performance is inconclusive — no clear edge. Independent model review is shown separately — it does not change what is signed unless the evidence gate applies.

Risk-Parity Hedge Falls Short of Bank-Crisis Thesis

Indian bank equities are expected to underperform INR-hedged defensive and USD-linked assets over the next 12 months as rupee depreciation tightens funding conditions and weakens risk appetite.
Executive summary

The tested full-universe risk-parity strategy produced a positive but modest 7.09% annualised return from 2019-05-27 to 2026-05-25, with a 0.46 Sharpe. Its -9.06% maximum drawdown and 6.76% annualised volatility indicate useful risk control, but not a compelling crisis hedge. The structure remains long Indian banks through BANKBEES, HDFCBANK and ICICIBANK, so it does not directly express the underperformance thesis. Given mid-range risk-adjusted returns and clear regime fragility, the appropriate stance is hold.

Key findings
  • The backtest generated a 60.09% total return and 7.09% annualised return over 2019-05-27 to 2026-05-25, so the strategy is not structurally broken.
  • Risk-adjusted performance was modest, with a 0.46 Sharpe and 0.50 Sortino, placing it in the unconvincing but viable range.
  • Realised risk was contained, with 6.76% annualised volatility, -9.06% maximum drawdown, and -0.64% 95% CVaR.
  • The 56.2% daily win rate suggests some consistency, but not enough to offset the middling Sharpe profile.
  • The allocation method naturally cuts exposure as realised volatility rises, but monthly rebalancing and a 63-day volatility signal may react after bank-equity losses have already occurred.
  • The universe includes MON100, GOLDBEES and LIQUIDBEES as defensive or USD-linked offsets, but also retains overlapping exposure to the Indian bank credit cycle.
Risks
  • The strategy is long BANKBEES, HDFCBANK and ICICIBANK, so a bank-led crisis would likely hurt performance before volatility-based sizing can reduce exposure.
  • Inverse-volatility sizing ignores correlation, allowing HDFC Bank, ICICI Bank and BANKBEES to appear diversified until a common macro shock reprices them together.
  • Monthly rebalancing and a 3% minimum weight per asset create stale-risk and forced-long exposure during gap-down events.
  • MON100 may not hedge a global risk-off regime, as USD/Nasdaq exposure can be pressured alongside Indian banks during FII outflows or rate-hike shocks.
Method / sources
Backtest metrics, Full Universe Risk Parity Baseline, 2019-05-27 to 2026-05-25Strategy specification: monthly rebalance, inverse-volatility risk_parity allocation across BANKBEES.NS, HDFCBANK.NS, ICICIBANK.NS, MON100.NS, GOLDBEES.NS and LIQUIDBEES.NSCritic concerns and stress scenarios provided for Mar-2020 COVID shock, May-2013 taper tantrum and 2022 rate-hike cyclearXiv:1403.0481v1 — Support Vector Machine Model for Currency Crisis Discrimination
Strategy
Full Universe Risk Parity Baseline
risk_parity
monthly

Allocates inversely to volatility, naturally cutting stressed bank exposure if realized risk rises.

Current allocation
BANKBEES.NS
Nippon India ETF Bank BeES
10.9%
HDFCBANK.NS
HDFC Bank
10.9%
ICICIBANK.NS
ICICI Bank
10.9%
MON100.NS
Motilal Oswal Nasdaq 100 ETF
11.1%
GOLDBEES.NS
Nippon India ETF Gold BeES
8.2%
LIQUIDBEES.NS
Nippon India ETF Liquid BeES (overnight)
48.0%
Why this is tamper-proof

Viewing this page, you're trusting Attevia's server. The signature is what matters once this memo leaves it — in a PDF, a screenshot, a pitch deck, or after Attevia is gone. It does three things no screenshot can: it proves these exact numbers haven't been altered since signing (even by us), it proves when the call was made (you can't backdate a track record), and it's verifiable by anyone offline against /api/pubkey without trusting Attevia at all. Change a single digit and the verification breaks. Software, not investment advice.