You've covered a lot of ground in Path 2. You know why rules beat gut feeling (2.1). You understand the five major factors (2.2). You can read risk metrics (2.3) and evaluate a backtest honestly (2.4). Now it all comes together.
Here is the complete process of building a systematic strategy from scratch: every decision, the trade offs, and what each one looks like in a real strategy on RupeeCase. The same five decisions sit under a simple momentum screen and under a multi factor portfolio run by an institution. Learn them once and you can read any strategy.
The five decisions in every systematic strategy
Every systematic equity strategy, no matter how simple or complex, requires five design decisions. Get these right and you have a strategy worth backtesting.
Real examples from the RupeeCase lineup: the Nifty 50 for Nifty 50 Equal Weight, and the Nifty Total Market, about 750 stocks, filtered to share prices under ₹5,000, for Allcap. A wider universe gives the signal more to choose from; a narrower one is more liquid and easier to hold in small accounts.
The signal must be objective (same calculation for every stock), historically documented (not invented for this backtest), and available at decision time (no look-ahead bias).
Equal weight (1/N each) is simplest and most robust. Volatility-weighted (inverse volatility) typically improves Sharpe. Market-cap weighting defeats the purpose of factor selection.
- UniverseWhich stocks are allowed in.
- SignalHow every stock is ranked.
- PortfolioHow many to hold and how to weight them.
- RebalanceWhen the ranking is refreshed.
- CostsWhat every trade really costs.
A complete strategy specification
Here's what a properly documented strategy specification looks like. Every decision is explicit. Nothing is ambiguous.
This specification is complete. Any programmer given this page would build exactly the same strategy. That's what systematic means: no ambiguity at execution time.
Now compare it with a real one. Here is the same template filled in for a strategy you can actually see on RupeeCase.
Simple Momentum, Nifty 500
- Universe: Nifty 500
- Signal: 12 month minus 1 month return
- Holdings: top 30, equal weight
- Rebalance: first trading day of each month
- Benchmark: Nifty 500 TRI
Momentum, Nifty Total Market
- Universe: Nifty Total Market, about 750 stocks, price under ₹5,000
- Signal: momentum ranking
- Holdings: top 50, equal weight, about 2% each
- Rebalance: every 2 weeks
- Benchmark shown: Nifty 50
Evaluating your backtest honestly
| What to check | What you want to see | Red flag |
|---|---|---|
| CAGR vs benchmark | A margin over the benchmark that holds across most of the period | Alpha concentrated in 1 to 2 exceptional years only |
| Max drawdown | A fall you have decided in advance you can hold through | Much deeper than the benchmark; multi year recovery |
| Sharpe ratio | Above the benchmark's, over a window that includes a fall | Sharpe above 2.5 on long only equity |
| Annual performance | Positive alpha in most years | Alpha only in 1 exceptional year |
| Parameter sensitivity | Similar results at 9M, 10M, 11M, 12M lookback | Only 12M works; 11M collapses |
| Cost impact | Net returns still attractive vs benchmark | Gross attractive; net mediocre |
The most important thing: stick to it
Most of Path 2 has been the thinking work of strategy design. But in my years of trading, the biggest destroyer of returns I have seen isn't bad strategy design. It's good strategy design abandoned at exactly the wrong moment.
Every momentum strategy has stretches, sometimes 6 to 12 months or more, where it lags the index. The portfolio is doing everything right: buying the highest momentum stocks, rebalancing on schedule. But momentum is out of favour and the strategy is down 20% while the market is flat.
Here is a real one. The Allcap model peaked on 20 December 2024, fell 22.7% and did not get back to that peak until 10 April 2026: 476 days underwater. It ended calendar 2025 down 4.2%. Then 2026 to July was up 26.6%. Every month of it is in the grid below.
Monthly model returns for Allcap from 2022, from the published nav file. The weak run starts in late 2024 and lasts into early 2026.
Drawdown dates and yearly returns from the RupeeCase strategy data, model, back tested, as of 8 Aug 2026. Not client returns and not a promise of future returns.
This is when most investors abandon the strategy. And it's almost always the worst time to do so, the subsequent recovery is often sharp. The investors who stick with the process through the bad periods capture the good ones.
The process is the product. A systematic strategy only delivers its stated returns if you follow it consistently. If you override it every time it underperforms, you don't have a systematic strategy, you have a discretionary strategy with systematic window dressing. Before deploying real money, answer honestly: what is the maximum drawdown I can actually hold through without overriding the system? Choose a strategy whose historical max drawdown is below that number. That is your real constraint.
What would you do?
It is February 2025. Your Allcap model portfolio is well below its December 2024 peak, and the index is falling too.
- Exit and wait for momentum to come back
You lock in the fall. The model went on to recover its December 2024 peak by 10 April 2026, and by July 2026 it was up 26.6% for the year.
- Check the rules are running, then stay
You sit through the rest of a 476 day drawdown, which was within the kind of falls the model had shown before, and you are there for the recovery.
Model, back tested, from the RupeeCase strategy data as of 8 Aug 2026. A past episode, not a forecast.
A worked example: your strategy on one page
Try writing your own before you invest in anyone else's. Say you want something calmer than pure momentum. A first draft could read:
That is close to how the Allcap Multi Asset strategy is built: the Allcap momentum sleeve plus a 10% debt and 10% gold sleeve, rebalanced every two weeks. Its model, back tested worst fall was 21.3%, against 22.7% for Allcap alone, as of 8 Aug 2026. The cushion did not stop the fall. It made it a little shorter and a little shallower, which is often the difference between holding on and giving up.
Common mistakes when building a first strategy
Starting from the back test. Write the rules first, then test. If you tune the rules to the back test, you are fitting the past.
Too few stocks. Ten stocks makes every single name a big bet. The factor needs breadth to show up.
No drawdown limit written down. If you have not decided the fall you will accept, the market will decide for you, at the worst moment.
Changing two things at once. When you tweak a rule, change one thing and test it. Otherwise you never know what helped.
MythA good strategy should beat the index every year.
FactNo real strategy does. Allcap's model lost 4.2% in 2025 after two very strong years. A strategy is judged over its full cycle, against the drawdown you agreed to before you began.
How this shows up in RupeeCase
On RupeeCase you do not have to code the five decisions yourself. Each strategy page lays them out, universe, ranking, holdings, weighting, cadence, and shows the model, back tested record, the worst drawdowns and the benchmark. You compare, you check the back test, and you invest; the stocks sit in your own demat, and at each rebalance you see the orders before they go to your broker. Path 3 goes factor by factor, so you know exactly what each ranking rule is betting on.
Can I run my own custom rules on RupeeCase?
Today you choose from the published strategies, each a fixed set of rules. The analysis and research module that is coming for qualified investors and institutions will add deeper tools. Until then, the template on this page is the way to think through a strategy of your own.
How much money do I need for a 50 stock strategy?
Enough that each position buys at least one share of the most expensive name at its target weight. That is why strategies publish a minimum: Allcap's is ₹2,25,000, with a price cap of ₹5,000 a share to keep it reachable.
Should I run two strategies at once?
Many investors do, so that one rule's bad stretch is cushioned by another's. Check that they really differ, in universe or factor, otherwise you own the same bet twice.