Comparisons
Two approaches, side by side, in a decision matrix you can scan in one screen: a criterion-by-criterion table, pros and cons for each, who each is best for, and a verdict. For the long-form narrative treatment of the same questions, see the research articles.
Risk & portfolio construction 13
- Arithmetic vs Logarithmic Returns
Arithmetic vs log returns decision matrix: time aggregation, cross-asset aggregation, compounding, and which return definition to use when.
6 criteria
- Calmar vs MAR Ratio
Calmar vs MAR ratio decision matrix: lookback window, drawdown denominator, track-record length, and which return-to-drawdown metric to report.
6 criteria
- Dollar-Cost Averaging vs Lump-Sum Investing
Dollar-cost averaging vs lump-sum decision matrix: expected return, regret risk, time in market, and which cash-deployment strategy fits.
6 criteria
- Fixed vs Trailing Stops
Fixed vs trailing stop-loss decision matrix: profit protection, whipsaw risk, trend capture, and which exit rule fits trend or mean reversion.
6 criteria
- Full Revaluation vs Delta-Gamma VaR
Full revaluation vs delta-gamma decision matrix for option VaR: nonlinear-payoff accuracy, compute cost, and which approximation to use.
6 criteria
- Historical vs Parametric VaR
Historical vs parametric VaR decision matrix: distribution assumptions, fat-tail handling, speed, and which estimation method fits your book.
6 criteria
- Max Drawdown vs Ulcer Index
Max drawdown vs Ulcer Index decision matrix: single worst loss versus sustained pain, sensitivity to one event, and which drawdown metric to report.
6 criteria
- Mean-Variance vs Risk-Parity Allocation
Mean-variance vs risk-parity decision matrix: return-forecast dependence, estimation error, leverage, and which method to allocate with.
6 criteria
- Monte Carlo vs Historical VaR
Monte Carlo vs historical VaR decision matrix: model dependence, nonlinear payoffs, scenario coverage, and which simulation method to trust.
6 criteria
- Realized vs Implied Volatility
Realized vs implied volatility decision matrix: backward versus forward looking, the variance risk premium, what each measures, and which to use when.
6 criteria
- Sharpe Ratio vs Sortino Ratio
A decision matrix comparing the Sharpe and Sortino ratios: what each penalizes, when they diverge, their assumptions, and which to report for a given return profile.
6 criteria
- Value-Weighted vs Equal-Weighted Index
Value-weighted vs equal-weighted index decision matrix: concentration, factor tilts, rebalancing cost, and which weighting to benchmark with.
6 criteria
- VaR vs CVaR (Expected Shortfall)
VaR vs CVaR decision matrix: what each measures, tail sensitivity, coherence, regulatory status, and which risk number to report for a fat-tailed book.
6 criteria
Backtesting & validation 6
- Block Bootstrap vs IID Bootstrap
Block bootstrap vs IID bootstrap decision matrix: serial correlation, volatility clustering, block length, and which resampling fits returns.
6 criteria
- GBM vs Jump-Diffusion Price Models
GBM vs jump-diffusion decision matrix for synthetic prices: tail behavior, gap risk, calibration cost, and which process to simulate for backtests.
6 criteria
- In-Sample vs Out-of-Sample Testing
In-sample vs out-of-sample decision matrix: what each measures, optimism bias, overfitting detection, and how to split backtest data honestly.
6 criteria
- Paper vs Live Trading
Paper vs live trading decision matrix: fill realism, slippage, latency, psychology, and what a paper account proves before risking real capital.
6 criteria
- Simple vs Exponential Moving Average
SMA vs EMA decision matrix: lag, responsiveness, weighting of recent data, drop-off artifacts, and which moving average fits a signal or smoothing task.
6 criteria
- Walk-Forward vs K-Fold Cross-Validation
A decision matrix comparing walk-forward analysis and k-fold cross-validation for financial backtesting: leakage, regime handling, data efficiency, and when each fits.
6 criteria
AI in markets 11
- Brier Score vs Log Loss
Brier score vs log loss decision matrix for scoring probabilistic forecasts: penalty shape, sensitivity to confident errors, and which to use.
6 criteria
- Cohen's Kappa vs ICC for Agreement
Cohen's kappa vs intraclass correlation decision matrix: categorical versus continuous ratings, chance correction, rater count, and which to use.
6 criteria
- Cosine vs Euclidean Similarity
Cosine vs Euclidean distance decision matrix for embeddings: magnitude sensitivity, normalization, and which metric to use for vector search.
6 criteria
- Embedding vs BM25 Retrieval
Embedding vs BM25 retrieval decision matrix for financial RAG: semantic recall, exact-term matching, cost, and which retriever for filings.
6 criteria
- Greedy Decoding vs Beam Search: Which to Use
Greedy decoding vs beam search for LLM extraction: determinism, sequence quality, compute cost, and why beam search drifts bland. Greedy at temp 0 wins.
9 criteria
- JSON Mode vs Tool Calling for Extraction
JSON mode vs tool calling decision matrix for structured LLM extraction: schema enforcement, validity guarantees, multi-field control, and which to use.
6 criteria
- Platt vs Temperature Scaling
Platt vs temperature scaling decision matrix for confidence calibration: parameters fit, multiclass behavior, and when each one fits best.
6 criteria
- Point Estimate vs Prediction Interval
Point estimate vs prediction interval decision matrix: single number versus quantified uncertainty, calibration, and which to report when.
6 criteria
- Self-Consistency vs Single-Pass Inference
Self-consistency vs single-pass decision matrix for LLM tasks: reasoning accuracy, cost multiplier, and when sampling-and-voting pays off.
6 criteria
- Semantic vs Recursive Chunking
Semantic vs recursive chunking decision matrix for filing RAG: boundary quality, cost, determinism, and which splitting strategy to use.
6 criteria
- Zero-Shot vs Few-Shot Extraction
Zero-shot vs few-shot prompting decision matrix for filing extraction: accuracy, token cost, format control, and which approach to choose.
6 criteria