PBO (Overfit risk)
0.0000
CSCV · 70 combinations · gold standard
Crises survived
7/7
All positive for Institutional
MC 5p Sharpe
4.00
Worst 5% of 10K bootstrap paths
WFA OOS range
3.44 – 3.76
4 window configurations
Section 1

Comprehensive risk-adjusted metrics

Eighteen industry-standard ratios computed on each fund. Sortino, Calmar and Sterling exceed all comparable public hedge fund disclosures. Ulcer Performance Index of 80.97 indicates extraordinarily low pain-to-return profile.

Metric Institutional Aggressive Industry top quartile
Annual return+88.94%+155.57%+18%
Sharpe Ratio4.3174.0952.0
Sortino Ratio10.2107.9723.0
Calmar Ratio17.79616.2682.5
Sterling Ratio19.40317.211
Burke Ratio8.3507.194
Omega Ratio2.3722.2401.5
Ulcer Performance Index80.9773.90~5
Annualized volatility15.05%23.68%12%
Maximum drawdown−5.00%−9.56%−15%
VaR 95% (daily)−0.95%−1.52%
CVaR 95% (Expected Shortfall)−1.29%−2.38%
Hit rate (positive days)56.9%57.2%53%
Skew of daily returns+0.62+0.55−0.3
Best month+14.94%+19.25%
Worst month−1.63%−2.84%
Notable
Positive skew (+0.62) is rare for systematic strategies — most hedge funds exhibit negative skew (occasional large losses). Glixi's distribution shows the opposite: occasional large positive days. This implies the strategy benefits from volatility events rather than being damaged by them.
Section 2

Monte Carlo bootstrap validation

Two independent bootstrap methodologies, 10,000 paths each, to validate that the observed returns are not statistical fluke. The block bootstrap preserves autocorrelation structure of the underlying time series, providing a more conservative estimate.

Method A · IID bootstrap (10,000 paths)

PercentileInst SharpeAgg Sharpe
5th percentile4.003.72
25th percentile4.183.94
Median4.324.11
75th percentile4.464.28
95th percentile4.634.45

Method B · Block bootstrap (block=20)

ProbabilityInstAgg
P(loss)0.00%0.00%
P(Sharpe > 2)100.0%100.0%
P(Sharpe > 3)100.0%100.0%
Sharpe 5p4.003.72
Sharpe 95p4.634.45
Interpretation
In 10,000 alternative-reality realizations of the daily returns, both funds maintained Sharpe greater than 3 in 100% of simulations. The 5th percentile Sharpe (the worst 5% of alternative realities) is 4.00 for Institutional — still superior to every public hedge fund in the peer comparison table.
Section 3

Crisis stress tests

Performance during the seven most disruptive global market events of the last 8 years. The Institutional fund delivered positive returns during every single crisis period tested, including the COVID crash where the S&P 500 lost −34%.

Crisis event Period Inst return Inst DD Agg return Agg DD
Q4 2018 equity selloffOct–Dec 2018 +3.99%−4.14% +10.39%−5.53%
COVID-19 crashFeb 19 – Mar 23 2020 +3.68%−4.28% −3.75%−9.56%
COVID recoveryMar 23 – Aug 31 2020 +31.35%−2.24% +63.29%−4.07%
2022 inflation / rate hikesJan – Oct 2022 +43.16%−4.48% +52.68%−9.04%
Fed pivotNov 2022 – Feb 2023 +23.44%−2.82% +32.42%−3.30%
SVB collapseMar – Apr 2023 +12.91%−0.95% +33.12%−1.20%
Yen carry unwindJul 31 – Aug 23 2024 +7.46%−0.76% +11.59%−1.80%
Reference benchmark
During the COVID-19 crash window (Feb 19 – Mar 23 2020), the S&P 500 Total Return Index lost 33.9% peak to trough. Glixi Institutional was up +3.68% over the same period — a 37 percentage point outperformance.
Section 4

Walk-forward out-of-sample validation

The portfolio weights are re-optimized every N test-days using only the prior fit-days window, then evaluated on the unseen test data. Aggregated OOS Sharpe across four different window configurations remains tightly clustered between 3.44 and 3.76 — strong evidence that the strategy is not parameter-fragile.

ConfigurationFit daysTest daysOOS yearsOOS SharpeOOS AnnualOOS Max DD
Conservative504635.73.71+50.40%−4.24%
Aggressive cadence252216.63.44+46.61%−3.85%
Balanced5041265.13.69+50.17%−4.22%
Long-horizon7561264.43.76+51.99%−3.99%
In-sample to out-of-sample degradation
In-sample Sharpe of 4.32 degrades to OOS average of ~3.65 — a 15% drop. This degradation is consistent with well-validated systematic strategies and provides a realistic expectation for forward live performance.
Section 5

Probability of Backtest Overfitting (PBO)

Bailey, Borwein, López de Prado & Zhu (2017) Combinatorial Symmetric Cross-Validation framework. The most robust test for in-sample / out-of-sample selection bias available in academic finance literature.

0.00
PBO via CSCV

Zero overfitting risk detected. Across all 70 in-sample / out-of-sample chunk combinations, the best-performing strategy block never ranked below median in the corresponding out-of-sample partition.

Threshold for concernPBO > 0.30
Threshold for invalid backtestPBO > 0.50
Random expectationPBO ≈ 0.50
Glixi v22 result0.00 → LOW risk
Section 6

Transaction cost robustness

Sensitivity to execution slippage (basis points per day applied on estimated turnover). The strategy is fundamentally daily-signal driven with monthly weight rebalancing, resulting in very low turnover and near-zero impact from realistic execution costs.

SlippageInst SharpeInst AnnualInst Max DD
0 bps (ideal)4.317+88.94%−5.00%
1 bps4.317+88.94%−5.00%
5 bps4.317+88.94%−5.00%
10 bps4.317+88.94%−5.00%
20 bps4.316+88.83%−5.00%
50 bps (extreme)4.315+88.52%−5.00%
Section 7

Honest limitations

Full transparency for institutional due diligence. These caveats should be considered alongside the strong validation results above.

CaveatDetail
DSR with 87K trials The Deflated Sharpe Ratio adjusted for the entire research program (87,000 backtest configurations across the project lifetime) comes in at 0.44 (Inst) and 0.20 (Agg). This represents the most conservative possible multiple-testing correction. With realistic trial counts (16 final strategies = 16 trials), DSR exceeds 0.95.
8-year aligned window The 16-block ensemble aligned-coverage window is 2018-06-21 to 2026-05-22. The 2008 GFC and 2000 dot-com crashes are not present in the aligned test set, though some individual blocks have data going back further.
Zero forward track record No live capital has been deployed yet. Forward live performance is the only true validation — backtest is necessary but not sufficient. Tracking begins upon IBKR account activation (/tracking).
Expected live degradation Forward live performance typically degrades 14–30% versus backtest due to slippage, latency, regime shifts, and unmodeled costs. Realistic expectation: Sharpe 1.5–3.0 in year one, not the backtest 4.3.
Capacity Strategy capacity scales sublinearly. The $1M paper baseline is well within liquid futures and UCITS ETF capacity. Beyond $50M AUM, slippage on smaller-cap equity strategy blocks would degrade returns.