Backtest View

Source: Cohen, Aiche & Eichel (2025), Entropy 27, 550 · Jan 2020 – Jan 2025 · NASDAQ-100

Backtested results, Jan 2020 – Jan 2025. Not a representation of live or future performance. NASDAQ-100 universe only. Paper benchmarks used ChatGPT-4o; live system uses Claude.
⚠ Backtest benchmarks were generated with ChatGPT-4o as the LLM provider. Live system uses Claude. Historical LLM scores are not regenerated. Paper-published results shown as reference only.

Performance Statistics

Source fact · Locked values from Table 1, Cohen et al. (2025)

Source Fact
Metric
technical
fundamental
entropy
Cumulative Return1977.71%700.52%578.40%
Sharpe Ratio0.69340.42070.5001
Average Return7.50%5.23%4.32%
Volatility10.82%12.44%8.63%
Optimal ML Weight1.000.700.15
Optimal LLM Weight0.000.300.85
● Technical monthly: highest cumulative return (1977.71%)● Technical quarterly: highest Sharpe ratio (1.2967) — distinct winner● Fundamental monthly: lowest volatility (8.63%)

Cumulative Return — All Configurations

Source fact · Exact paper values

Cumulative Return Time Series

Full monthly return series — Figure 1 equivalent

Time-series data required

This chart requires the full monthly return series from the paper's underlying data. Qualitatively: technical diverges upward sharply after ~30 months; fundamental and entropy grow steadily.

Required Fields

period_indexstrategy_typerebalance_frequencycumulative_return

Weight Sensitivity — Cumulative Return vs. ML Weight w

Figure 3 equivalent · Full sweep w ∈ [0.00, 1.00] step 0.05

Qualitative description (source: Section 4, Cohen et al.): Technical: monotonically increasing → peak at w=1.00 (monthly), w=0.45 (quarterly). Entropy: concave relationship with peak at w=0.70 (monthly) — balanced blend essential. Fundamental: peaks sharply at low ML weights → LLM semantic context dominates.

Weight sensitivity series required

This chart requires the full sweep data (w ∈ [0, 1] in steps of 0.05) per strategy-frequency pair from the paper's underlying dataset.

Required Fields

strategy_typerebalance_frequencyml_weightcumulative_return
Feb–May 2020

COVID-19 Market Crash — Stress Test

Section 4.1, Cohen et al. (2025): All three strategies encountered extreme volatility but none collapsed. Technical outperformed early and rebounded fastest by May 2020. Fundamental and Entropy showed more muted, conservative profiles. This validates the framework under high-risk, high-uncertainty conditions.

Stress-test time series required

This chart requires daily cumulative return data for Feb–May 2020. Per the paper, no strategy collapsed during this period. Technical was quickest to recover.

Required Fields

datestrategy_typecumulative_return