Performance Analytics

Daily: a quick P&L check (5 minutes). Weekly: summary metrics and notable patterns (30 minutes). Monthly: comprehensive analysis with charts (1-2 hours). Quarterly: a deep strategic review (2-4 hours). Don't over-analyze short periods - statistical significance requires sufficient data. But regular check-ins catch problems early.

Below 0.5: poor - rethink the strategy. 0.5-1.0: acceptable - room for improvement. 1.0-2.0: good - competitive with professional funds. 2.0+: excellent - rare and exceptional. For retail traders, a Sharpe above 1.0 is a reasonable goal. One Singapore note: because the local risk-free rate (~3%) is much lower than India's (~6.5-7%), the same returns produce a higher Sharpe here, so don't compare your Sharpe directly against figures computed on a higher risk-free rate.

Win rate alone doesn't determine profitability. You could win 80% of trades but if the average win is S$1,000 and the average loss is S$5,000, you lose money: (0.8 × S$1,000) - (0.2 × S$5,000) = S$800 - S$1,000 = -S$200 per trade. Payoff ratio (avg win / avg loss) matters as much as win rate.

Track both. Gross returns show your strategy's potential - the edge before costs. Net returns show actual results after brokerage, SGX fees, GST, and slippage. Compare them: if gross is good but net is poor, costs are eating your edge. Note that in Singapore there is no capital gains tax, so the gross-to-net gap is fees and slippage, not tax. Focus on net returns for reality, but monitor gross to understand if cost reduction could help.

Use the Singapore Government Securities (SGS) yield - for example the 6-month T-bill or 10-year SGS - or SORA, the Singapore Overnight Rate Average that replaced SIBOR/SOR. It is currently roughly 2.5-3.5% annually, much lower than India's because Singapore rates track US rates given its exchange-rate-centred monetary policy. For monthly calculations divide by 12, for daily divide by 365. Singapore Savings Bonds, T-bills, or money-market fund yields are practical risk-free alternatives.

Use statistical tests: a t-test on returns (t-stat > 2 suggests skill), Monte Carlo simulation (compare to random), bootstrap analysis (test robustness). Also: a longer track record = more confidence in skill. One year of data is suggestive, 3+ years is more conclusive. Consistent performance across different market conditions suggests skill.

Arithmetic: the simple average of periodic returns. Geometric: the compound return accounting for sequencing. Example: +50% then -50% = arithmetic average 0%, but geometric return -25% (S$100 -> S$150 -> S$75). Geometric is your actual growth rate. Always use geometric (CAGR) for true performance.

Use percentage returns, not absolute dollars. A S$10,000 profit on S$100,000 capital (10%) is very different from S$10,000 on S$1,000,000 (1%). For deposits/withdrawals during the period, use time-weighted return (TWR), which eliminates the impact of cash flows. Money-weighted return (IRR) includes cash-flow timing effects.

It depends on strategy and risk tolerance. Conservative: 10-15% max. Moderate: 15-25% max. Aggressive: 25-35% max. Professional funds typically target 15-20% max. Important: your max drawdown will likely be exceeded in the future - the historical max is not a ceiling. Plan for worse than you've experienced.

This is complex because no perfect benchmark exists. Options: the STI (broad market), your strategy's underlying, an options-selling index if you're selling premium, or a risk-adjusted comparison via Sharpe. For delta-neutral strategies, the risk-free rate is an appropriate benchmark. Match the benchmark to your strategy's risk profile. Note that on SGX single-stock options are thin, so much options activity uses index or overseas options.

Use point-in-time data (what was actually tradeable at that moment). Include delisted stocks in the universe. Don't use current index constituents historically. Recognize that live performance typically underperforms backtests by 30-50%. Apply haircuts to backtested metrics before making decisions.

It depends on return volatility and the effect size you're trying to detect. Rule of thumb: a minimum of 30 observations for basic statistics; 100+ trades for meaningful segmented analysis; 250+ for reliable factor analysis. For subtle effects (small alpha), you need more data. Use power analysis to determine the required sample size for your specific hypothesis.

Segment analysis by regime (bull/bear/sideways, high/low volatility). Use regime-switching models. Calculate regime-conditional metrics. Test if your edge persists across regimes. Be cautious of strategies that only work in one regime - they may fail when the regime changes. Regime-robust strategies are more reliable.

Normalize returns to equivalent leverage (e.g., 1x). Levered Sharpe = unlevered Sharpe (leverage doesn't change Sharpe if borrowing at the risk-free rate). But drawdowns scale with leverage. Report both levered and unlevered metrics. Be especially cautious of a high Sharpe achieved through very high leverage.

Use confidence intervals, not point estimates. Apply shrinkage (blend historical with the market average). Assume some mean reversion of exceptional returns. Account for volatility drag (geometric < arithmetic). Scenario analysis: bull, base, bear, tail. Monte Carlo for probability distributions. Be skeptical of projections from short histories.