Momentum Cadence and Portfolio Design in Indian Equities

Price momentum — the tendency for stocks that performed well over the past year to keep outperforming over the next several months — is among the most replicated patterns in empirical asset pricing. Classic work documents the effect in US equities using portfolios sorted on past returns and rebalanced at regular intervals. What receives less attention in practitioner discussions is how implementation choices — how often to refresh the book, how wide to cast the investable net, how many names to hold, and how to weight them — jointly shape realised performance once trading frictions are included.

This report applies a controlled experimental grid to a liquid NSE equity panel drawn from our in-house universe (~160 listed tickers; roughly 130–140 names with usable monthly prices and market-cap data at any given month). We do not have a 500- or 750-name point-in-time database — universe tiers are 50, 100, and the full eligible panel so each row in the grid reflects a distinct investable breadth within our actual coverage.

We construct 144 long-only momentum portfolios varying four design levers: rebalance every 1, 2, 3, or 6 months; draw from the top 50, top 100, or full eligible names by market capitalisation; hold 15, 30, or 50 stocks; and apply one of four weighting rules — capitalisation-weighted, equal-weighted, rank-weighted, or capitalisation tilted by a normalised momentum score.

The ranking signal is 12-1 momentum: cumulative return from month $t-12$ through $t-2$, deliberately skipping the most recent month to reduce microstructure noise and short-term reversal effects documented in the global literature. Portfolios with rebalance intervals longer than one month use overlapping staggered tracks so that every calendar month reflects an active implementation. We report both gross and net-of-cost outcomes so the trade-off between signal freshness and trading frictions can be read directly from the data.

The exhibits below are grouped by theme: study metadata and headline patterns, return formation, overlapping implementation, risk-adjusted metrics, transaction costs, factor exposures, and portfolio turnover dynamics.

Cross-sectional momentum arises when relative winners continue to outperform relative losers over horizons of roughly three to twelve months. The effect has been documented across geographies, asset classes, and sample periods, though its magnitude varies with market structure, liquidity, and macro regime. A parallel literature emphasises that momentum is an active strategy: profits can erode once realistic turnover and transaction costs are applied, especially when portfolios are refreshed frequently.

Rebalance frequency sits at the centre of that trade-off. Refreshing often keeps the portfolio aligned with the latest ranking signal but raises turnover; refreshing rarely saves on trading costs but allows rankings to go stale. Commercial momentum indices and academic backtests often differ on this dimension — some products rebalance monthly, others quarterly or semiannually — yet few studies vary frequency jointly with universe breadth, portfolio size, and weighting scheme.

Our grid is designed to fill that gap for Indian equities specifically. India offers a deep but concentrated equity market with meaningful small-cap dispersion beyond the headline indices. By sweeping rebalance cadence and portfolio construction choices together, we can separate which design levers move gross alpha from which ones survive net of stylised trading costs.

The study is a full factorial over four rebalance intervals ($k\in \{1,2,3,6\}$ months), three data-supported universe tiers ($N\in \{50,100,N_{\max }\}$ where $N_{\max }$ is the count of names with valid market-cap each month, currently about 136), three portfolio sizes ($H\in \{15,30,50\}$), and four weighting schemes — yielding $4\times 3\times 3\times 4=144$ configurations. Each configuration is simulated on the same monthly return panel.

Universe membership is reconstituted monthly from market-cap ranks within our liquid ticker list. Momentum scores are computed cross-sectionally inside each universe. All portfolios are long-only. The headline sample window runs from September 2005 through February 2023 (six months trimmed from each end for six-month overlapping tracks). The study-summary panel reports exact ticker counts and universe caps from the latest pipeline run.

Let $r_{i,t}$ denote the total return of stock $i$ in month $t$. The formation-month score compounds lagged monthly returns, excluding the most recent month:

$${\text{Mom}}_{i,t}=\prod _{k=2}^{12}(1+r_{i,t-k})-1$$

At each month $t$, stocks are ranked by market capitalisation within our panel; the top $N\in \{50,100,N_{\max }\}$ names define universe ${\mathcal{U}}_t$. The portfolio ${\mathcal{P}}_t$ contains the $H$ stocks in ${\mathcal{U}}_t$ with the highest ${\text{Mom}}_{i,t}$.

Weighting rules allocate capital across ${\mathcal{P}}_t$: capitalisation weight $w_i\propto {\text{MCAP}}_i$; equal weight $w_i=1/H$; rank weight $w_i\propto R_i$ where $R_i$ is the momentum rank; and score-tilted cap weight $w_i\propto z_i^{\ast }\cdot {\text{MCAP}}_i$ where $z_i^{\ast }$ is the cross-sectionally winsorised momentum z-score.

Holdings earn the next month's return: $R_{p,t}={\sum }_{i\in {\mathcal{P}}_{t-1}}{w}_{i,t-1}{r}_{i,t}$. The heatmap below maps annualised mean gross return across rebalance frequency and universe size for a representative 30-stock, capitalisation-weighted slice of the grid.

A single non-overlapping backtest that trades only every $k$ months produces sparse return series and confounds signal horizon with calendar timing. Following the overlapping-portfolio construction widely used in momentum research, we run $k$ independent tracks with rebalance months offset by $0,1,\dots ,k-1$. Each track maintains its own holdings and weights between rebalances.

The portfolio return attributed to month $t$ is the geometric average across all tracks active that month:

$$R_{p,t}^{(k)}={\left(\prod _{j=1}^k(1+R_{p,t}^{(j)})\right)}^{1/k}-1$$

This convention ensures that a six-month strategy still produces a monthly return series comparable to a monthly strategy. The cumulative wealth and drawdown charts plot gross outcomes for capitalisation-weighted, 30-stock portfolios at the mid-tier universe (top 100 in the latest run) across all four rebalance intervals.

We annualise mean return and standard deviation of monthly portfolio returns. The Sharpe ratio compares excess return to total volatility: $\text{SR}=({\bar{R}}_p-r_f)/{\sigma }_p\cdot \sqrt{12}$. The Sortino ratio replaces ${\sigma }_p$ with downside deviation, penalising harmful volatility only.

Maximum drawdown measures peak-to-trough loss on the cumulative wealth path: ${\min }_t(W_t/{\max }_{s\le t}{W}_s-1)$. Effective holdings $1/{\sum }_i{w}_i^2$ summarises concentration — equal-weight portfolios approach $H$; concentrated cap-weight books score lower.

When comparing monthly and semiannual implementations, we report Memmel (2003) test statistics for the difference between correlated Sharpe ratios. The heatmaps and bar chart below display gross Sharpe, Sortino, and drawdown across the rebalance–universe grid; the table at the end of this section tests whether monthly vs six-month Sharpe differences are statistically distinguishable for 15-stock cap-weight portfolios.

Published momentum studies often assume frictionless trading. We model a proportional cost applied on each rebalance date based on one-sided turnover ${\tau }_t$ — half the sum of absolute weight changes:

$$R_{p,t}^{\text{net}}=R_{p,t}^{\text{gross}}-{\tau }_t\cdot c$$

The baseline cost rate is $c=$ 20 basis points one-way (12 bps execution commission plus 8 bps slippage proxy). Sensitivity runs use 0, 15, 30, and 50 bps to bracket institutional and retail frictions.

The economic question is whether higher gross returns from frequent rebalancing survive after costs. The panels below show net Sharpe heatmaps, net cumulative wealth, and cost-scenario comparisons for a top-100, 30-stock, cap-weight configuration from our grid.

High realised returns may reflect genuine momentum exposure or passive loadings on market beta, size, value, and other style factors. We regress monthly portfolio excess returns on a six-factor model in the spirit of Fama and French (2015), extended with a momentum factor:

$$R_{p,t}-r_{f,t}=\alpha +{\beta }_{MKT}MK{T}_t+{\beta }_{SMB}SM{B}_t+{\beta }_{HML}HM{L}_t+{\beta }_{RMW}RM{W}_t+{\beta }_{CMA}CM{A}_t+{\beta }_{MOM}MO{M}_t+{\epsilon }_t$$

If portfolios are capturing cross-sectional momentum, **${\beta }_{MOM}$** should be positive and statistically meaningful. If $\alpha$ is insignificant after controlling for the six factors, performance is largely explained by known risk premia rather than an unexplained residual.

The bar chart and regression table below report momentum-factor loadings and full FF6 coefficients for 15-stock capitalisation-weighted portfolios across rebalance frequencies and universe sizes.

Two microstructure statistics help interpret implementation feasibility. Annualised one-sided turnover scales the average weight change at each rebalance to a yearly rate. Median new-entrant share measures what fraction of holdings at a rebalance were not held in the previous portfolio — a proxy for how disruptive each refresh is, independent of weight shifts among continuing names.

These metrics often move in opposite directions as rebalance frequency falls: slower refresh schedules trade less often, but each trade tends to replace a larger fraction of the book. That pattern matters for liquidity planning and operational workload even when per-trade costs are fixed. The turnover heatmap, new-entrant table, and full summary statistics below cover the 144-configuration grid.

Cadence and gross returns. In this sample, shorter rebalance intervals (one and two months) generally produced higher mean gross returns and Sharpe ratios than quarterly or semiannual schedules, holding universe size, holdings count, and weighting fixed. Rank-weighted and score-tilted schemes showed the largest spread across cadences.

Universe breadth (within our panel). Moving from top-50 to the full eligible set raised average returns and volatility where the tiers differ materially — the effect depends on how many names our data actually support each month.

Costs matter monotonically. Applying 20 bps one-way costs narrowed the gross advantage of monthly rebalancing. At higher friction assumptions, semiannual portfolios approached or exceeded monthly net Sharpe for several configurations.

Factor structure. Six-factor regressions typically showed positive momentum-factor loadings with insignificant intercepts — consistent with returns being driven by systematic style exposure rather than unexplained alpha.

Turnover dynamics. Annualised turnover declined as rebalance intervals lengthened, while the median share of new entrants per rebalance rose — implying that less frequent trading does not always mean simpler implementation.

This document is quantitative research, not investment advice. Historical simulations describe past relationships that may not persist under different liquidity conditions, regulatory regimes, or market structure.

Transaction costs are modelled as a flat rate per unit of turnover. We do not separately estimate bid–ask spreads by cap tier, market impact for large orders, or jurisdiction-specific levies. Net performance should be read as illustrative, not broker-accurate.

Our investable panel contains on the order of 130–140 liquid NSE names with price and cap data — not a full exchange-wide database. Universe tiers (50 / 100 / full panel) are chosen to stay within that coverage; results do not speak to momentum on 500+ stock universes.

Universe and signal construction use month-end prices and a cap snapshot scaled by price each month. Delisting adjustments and corporate-action timing can shift absolute return levels. Cross-configuration rankings of rebalance cadence are more robust than point estimates of annualised return.

The six-factor model is estimated on a broad domestic equity panel and may not perfectly align with every portfolio's investable set. Attribution coefficients describe average factor co-movement, not a formal fund-accounting decomposition.

Asness, C. S., Moskowitz, T. J., & Pedersen, L. H. (2013). Value and momentum everywhere. *The Journal of Finance*, 68(3), 929–985.

Carhart, M. M. (1997). On persistence in mutual fund performance. *The Journal of Finance*, 52(1), 57–82.

Daniel, K., & Moskowitz, T. J. (2016). Momentum crashes. *Journal of Financial Economics*, 122(2), 221–247.

Fama, E. F., & French, K. R. (2015). A five-factor asset pricing model. *Journal of Financial Economics*, 116(1), 1–22.

Grinblatt, M., & Moskowitz, T. J. (2004). Predicting stock price movements from past returns: The role of consistency and tax-loss selling. *Journal of Financial Economics*, 71(3), 541–579.

Jegadeesh, N., & Titman, S. (1993). Returns to buying winners and selling losers: Implications for stock market efficiency. *The Journal of Finance*, 48(1), 65–91.

Jegadeesh, N., & Titman, S. (2001). Profitability of momentum strategies: An evaluation of alternative explanations. *The Journal of Finance*, 56(2), 699–720.

Lesmond, D. A., Schill, M. J., & Zhou, C. (2004). The illusory nature of momentum profits. *Journal of Financial Economics*, 71(2), 349–380.

Memmel, C. (2003). Performance hypothesis testing with the Sharpe ratio. *Finance Letters*, 1(1), 21–23.

Moskowitz, T. J., & Grinblatt, M. (1999). Do industries explain momentum? *The Journal of Finance*, 54(4), 1249–1290.

Novy-Marx, R., & Velikov, M. (2016). A taxonomy of anomalies and their trading costs. *The Review of Financial Studies*, 29(1), 104–147.