Investment Modes
Understand how each mode allocates capital and constructs your portfolio before you create your strategy.
Ranks all stocks in the universe by your alpha score on each day and allocates as much capital as possible into the top-ranked name. A single stock can receive 100% of the portfolio if it scores highest. This is the most aggressive, concentration-maximizing mode.
RELIANCE scores 0.95 (highest).
The engine allocates your entire portfolio into RELIANCE. Tomorrow, if
INFY scores highest, it switches fully into INFY.
Best used for high-conviction, single-name momentum signals.
Selects the top-N highest scoring stocks each day and divides the portfolio equally among them. Each stock gets the same weight, regardless of how much higher its score is relative to others. This provides diversification while still using the alpha ranking to filter names.
Top Stocks = 5. On a given day, the engine picks the 5 highest-scoring stocks
and allocates 20% of the portfolio to each of them, irrespective of their individual
scores. Ideal for factor strategies (momentum, quality) where you believe in the group, not a single pick.
Like Equal-weight Top-N, but allocates capital proportionally to each stock's alpha score. A stock with twice the score of another gets approximately twice the capital. Combines the filtering benefit of Top-N with a signal-strength-based weighting.
HDFC=0.8, TCS=0.5, WIPRO=0.3. Total = 1.6.
Weights: HDFC = 50%, TCS = 31%, WIPRO = 19%.
Higher-conviction picks get more capital — useful when your alpha has strong ordinal predictive power.
Rebalances the portfolio only at fixed intervals (daily, weekly, monthly, or quarterly) instead of every trading day. Between rebalance dates, the portfolio is held static. Includes optional stop-loss and stop-gain guardrails on individual positions.
Frequency = Monthly, Stop Loss = 5%, Stop Gain = 20%.
On the 1st of each month, the engine picks the top-N stocks and equal-weights them.
If any position drops 5% before month-end, it's exited early and cash is held until
the next rebalance. Great for lower-turnover, swing-trading style strategies.
A market-neutral long/short strategy. Goes long the top-N highest-scoring stocks and simultaneously short the bottom-N lowest-scoring stocks. The total dollar value of longs equals the total dollar value of shorts — hedging market beta. P&L is driven purely by the spread between good and bad stocks.
₹1L each, and short the bottom 5 at
₹1L each. If the market falls 5% uniformly, both legs lose/gain equally —
your net exposure is ~zero. You profit only if longs outperform shorts.
Best for mean-reversion or cross-sectional momentum alphas.
Similar to Capital-neutral but adjusts long and short sizes so that the portfolio's net market beta is zero. High-beta shorts are sized smaller, and high-beta longs get less capital. This removes residual market exposure even when individual position betas differ.
HDFC (β=0.8) and short ZOMATO (β=1.5).
Capital-neutral would give each equal capital, leaving you net short beta.
Beta-neutral scales ZOMATO's short down so that 0.8 × long_₹ = 1.5 × short_₹,
eliminating the residual beta. Best when high-beta stocks dominate either side.
Runs a long/short portfolio where long and short exposure is balanced within each sector. For each sector, the engine longs the top-scored stocks and shorts the bottom-scored in equal measure. Removes sector-level macro risk — your alpha is tested purely within-sector.
HDFC, ICICI and short BANDHAN, IDFC.
In the IT sector: long TCS, INFY and short MPHASIS, HEXAWARE.
If IT sector rallies 10%, both the long and short IT legs gain and lose equally —
you're only exposed to the relative stock rankings within IT.
An all-weather mode that reads the market's current regime every trading day and adapts its execution style automatically. Instead of always running long-only or always hedging, it switches between three behaviours depending on what Nifty is doing:
Choose how the engine reads the market regime. Each method has different speed and sensitivity trade-offs.
ema_crossoverBest for: sustained multi-month bear markets (e.g. 2022 global selloff).
Weakness: slow — can lag a sudden crash by 4–8 weeks since EMA crossovers take time to form.
drawdown_from_highBest for: catching sudden crash events fast (e.g. COVID Feb 2020).
Weakness: triggers on short corrections too — may whipsaw in choppy markets.
hybridBest for: balancing speed and reliability.
Weakness: the asymmetry (easy to enter bear, hard to exit) can keep the strategy in hedge mode too long after a short correction recovers.
confirmed_hybrid RECOMMENDED
Best for: choppy markets with frequent short corrections that recover quickly (e.g. 2024–2025 Indian market).
Weakness: introduces a small lag (~N days) when entering a real bear — catches ~90% of a sustained crash, misses the first week.
| Setting | What it does | Recommended | Applies to |
|---|---|---|---|
| Short EMA Window | Fast EMA period used to detect Golden/Death Cross | 50 | EMA Crossover, Hybrid, Confirmed Hybrid |
| Long EMA Window | Slow EMA period for the crossover signal | 200 | EMA Crossover, Hybrid, Confirmed Hybrid |
| Bear Drawdown Threshold % | % drop from recent high that triggers Bear mode | 12 (choppy) / 9 (sensitive) | Drawdown, Hybrid, Confirmed Hybrid |
| High Lookback Days | How many days back to measure the recent high from | 40 (choppy) / 25 (sensitive) | Drawdown, Hybrid, Confirmed Hybrid |
| Confirmation Days | How many consecutive days the Bear/Bull signal must hold before switching | 5 | Confirmed Hybrid only |
Hybrid / Drawdown detector: Bear triggered on day ~1–2 of the crash (8% drop breached quickly). Strategy switched to beta-neutral hedge immediately — captured most of the downside protection.
Confirmed Hybrid (5 days): Bear confirmed by day ~9 (all 5 days agreed). Missed the first ~10% of the crash but protected against the remaining 30%. V-shaped recovery meant the exit to Bull was also clean.
Hybrid (8% / 20 days): Bear triggered on every correction → strategy went short → market recovered → short positions lost money. Repeated 4–6 times = significant whipsaw losses.
Confirmed Hybrid (12% / 40 days / 5 confirmation days): Short corrections never held the bear signal for 5 consecutive days → no false switches → strategy stayed long-only or sideways and participated in the recovery. Outperformed by ~9% in 2025 alone.
Hybrid (8% threshold): Bear triggered on day 2 → went short → market bounced back → losses on short positions.
Confirmed Hybrid (12% threshold): 9% drop did not cross the 12% threshold → no regime switch → stayed long-only and participated in the recovery.
Key lesson: raising the threshold from 8% to 12% is the single most impactful setting change for filtering event-driven noise.
All detectors: Bull regime held throughout → equal-weight long-only the entire year → full participation in the rally. No regime switches = no unnecessary hedges dragging performance.
In a clean bull market, all detectors behave identically — the differences only show up at turning points.
| Your market expectation | Recommended detector | Threshold | Lookback | Confirmation |
|---|---|---|---|---|
| Choppy / frequent corrections (2024–2025 style) | Confirmed Hybrid | 12% | 40 days | 5 days |
| Sudden crash risk (COVID-style) | Confirmed Hybrid | 9% | 25 days | 3 days |
| Slow sustained bear (2022 style) | EMA Crossover | — | — | — |
| Unknown / general all-weather | Confirmed Hybrid | 12% | 40 days | 5 days |
| Mode | Direction | Rebalances Daily | Stop Controls | Best For |
|---|---|---|---|---|
| Max Position | Long only | ✓ | ✗ | High-conviction, concentrated bets |
| Equal-weight Top-N | Long only | ✓ | ✗ | Diversified factor strategies |
| Score-weighted Top-N | Long only | ✓ | ✗ | Alphas with strong ordinal power |
| Scheduled Rotation | Long only | ✗ | ✓ | Low-turnover, swing trading |
| Capital-neutral L/S | Long + Short | ✓ | ✗ | Market-neutral, stat-arb |
| Beta-neutral L/S | Long + Short | ✓ | ✗ | Beta-hedged long/short books |
| Sector-neutral | Long + Short | ✓ | ✗ | Within-sector ranking strategies |
| Regime-Switching | Dynamic | ✓ | ✗ | All-weather strategies, crash protection + choppy market survival |
All modes use simulated backtests. Past simulated performance does not guarantee future returns. Long/short modes require margin and may not reflect real brokerage constraints. Start with Equal-weight Top-N if you are new to quantitative strategy design.