Research note: This material is for educational and strategy-design purposes. It is not investment advice or a promise of future returns.
Executive summary
Short-term mean reversion buys recent losers and sells or avoids recent winners over very short horizons, typically one day to one month. Jegadeesh and related work document short-horizon reversal effects, although implementation is transaction-intensive and therefore highly cost-sensitive.
In India, short-term reversal is best run in liquid names, often market-neutral or near-neutral, because the edge is small and can be overwhelmed by taxes, impact and borrow friction. A long-only version can still help as a timing overlay for entries into fundamentally selected names.
Description
This strategy exploits temporary overreaction, liquidity demand and microstructure bounce. It works best when yesterday’s or last week’s extreme move is not driven by true information but by liquidity pressure or transitory flow. That means the strategy should avoid obvious information events like earnings days, promoter actions or major regulation shocks.
For a PM, the most useful Indian design is a two- to five-day reversal in liquid large caps, either market-neutral or benchmark-aware, with event exclusions, spread filters and very tight turnover discipline. Longer reversion windows usually blur into value or post-event rebalancing rather than pure microstructure alpha.
Key attribute table
The range below is a highly cost-sensitive research target, not a broad promise.
| Time horizon | Turnover | Typical Sharpe or return expectation | Data needs | Complexity | |---|---|---|---|---| | Very short term | Very high | Sharpe roughly 0.6–1.2 gross before heavy costs; annualised return target roughly 8–15% net only if execution is strong | Daily or intraday prices, spreads, event flags | High |
Details
Universe: top 50–100 liquid NSE names. Signal: prior one-day or five-day idiosyncratic return relative to sector or market return. Buy the worst decile and sell or underweight the best decile, excluding names with earnings, corporate actions or abnormal news volume. Hold for one to five sessions. Position sizing should be inverse-volatility or equal-risk, targeting low single-name idiosyncratic risk. Risk controls: strict gross and net exposure limits, intraday spread filters and a rule that blocks names if the reversal is likely information-driven. NSE corporate-action calendars and result filings are essential event filters.
Backtests must include bid-ask spread and taxes. That is the only way to know if the signal survives. Delivery-based cash implementation is rarely sensible because delivery STT is too heavy for ultra-short horizons; if the strategy is run actively, derivatives or broker-enabled non-delivery structures are usually operationally superior, subject to strategy mandate and current regulations.
Edge cases include lower-circuit “falling knives”, upper-circuit squeezes and sudden macro headlines that make a nominal “reversal candidate” actually the start of a new information regime. The signal should therefore be small, tactical and heavily filtered.
Implementation guide
- Restrict the universe to the most liquid NSE names.
- Compute a one- or five-day residual return against sector or market.
- Drop names with fresh information events.
- Buy short-term overreacted losers and cut after one to five sessions.
- Use very small risk units and tight exposure controls.
- Measure edge only net of costs.
India-specific example
Suppose ICICIBANK underperforms Bank Nifty by 4% over three sessions on no scheduled result, while HDFCBANK outperforms by 3% on no major filing. A short-term reversion model may buy ICICIBANK and reduce or short HDFCBANK for a two-day holding period, expecting the relative move to partially mean-revert. If a fresh RBI policy or result release explains the move, the trade should be blocked. That event filter is not optional.
On a ₹10 crore market-neutral sleeve, the PM may allocate ₹50 lakh long and ₹50 lakh short per pair, then rebalance daily for beta neutrality. Because turnover is extreme, the strategy’s viability depends more on execution quality than on signal elegance. Any backtest that ignores Indian taxes and spread impact is liable to overstate the edge.