Verdict
Research study: two of the three are redundant; RANK500 is the complement
Weekly returns 2012 to Sep 2025: W and M correlation 0.89; R against W 0.46, against M 0.42, about 0 against both in the index's worst weeks. 50/50 M and R: 22.5% a year, drawdown 19.8%, validation Sharpe 1.03. Kelly (no leverage) is fully invested in R and M; fitted on 2012-2017 it did worse out of sample than equal weights.
- Data: NSE daily file for every name that traded, delisted ones included; universe = 500 most traded names each month (not the index list). Earlier published figures for the three were too high; see each strategy page.
- Kelly: single-strategy full Kelly would be 4 to 5 times capital (not advice); with a no-leverage cap the weights are R and M only, bootstrap ranges are wide. Use at most a half to a quarter of Kelly and cap total exposure at 100%.
- Recommendation: run M (not W) with R, about 50/50, as a paper-tested pair; W stays as M's baseline. No orders, nothing here changes any frozen paper run.
Promotion gates
No gates published for this strategy.
Explain the strategy
Three strategies, one table. Two of them behave almost the same, so owning both is like owning one twice. The third behaves differently, so mixing it in smooths the ride. We also asked what the Kelly formula says about how much to put in each: it says be fully invested but the answer is too uncertain to trust, so we suggest only a simple equal split.
Report
Strategies and portfolios on the common window
| strategy / portfolio | CAGR 2012-Sep 2025 | max drawdown | Sharpe | CAGR 2012-17 | CAGR 2018-Sep 2025 |
|---|---|---|---|---|---|
| Longbow Weekly (W) | 17.5% | 32.5% | 0.71 | 17.7% | 17.3% |
| Longbow Learned (M) | 21.8% | 28.7% | 0.87 | 21.7% | 21.9% |
| RANK500 v2 (R) | 22.0% | 31.1% | 0.93 | 23.1% | 21.1% |
| Nifty 500 price index | 14.1% | 34.4% | 0.62 | 17.0% | 11.9% |
| Portfolio: EW | 21.0% | 21.5% | 0.98 | 21.2% | 20.8% |
| Portfolio: MR | 22.5% | 19.8% | 1.06 | 22.8% | 22.2% |
| Portfolio: IV | 21.0% | 20.6% | 0.99 | 21.3% | 20.8% |
| Portfolio: MinVar | 21.0% | 22.6% | 0.99 | 21.6% | 20.5% |
| Portfolio: MaxSharpe | 22.4% | 25.3% | 1.04 | 23.1% | 21.9% |
| Portfolio: Kelly | 22.2% | 28.2% | 0.99 | 23.1% | 21.6% |
| Portfolio: Half Kelly | 13.6% | 14.8% | 0.99 | 14.5% | 13.0% |
| Portfolio: Quarter Kelly | 9.3% | 7.5% | 0.99 | 10.2% | 8.6% |
Rules and parameters
Rules
- Strategies as frozen in their own pages; weekly net returns at 1x costs, 2012-01 to 2025-09; cash earns repo less 1.5%.
- Portfolios long-only, total weight at most 100%, weights fitted on 2012-2017 and applied to 2018-2025; sleeves rebalanced weekly; rebalancing costs between sleeves ignored.
- Kelly: maximise the mean log of weekly wealth; single-strategy f = (mean excess)/variance; fractional Kelly scales the weights and keeps the rest in cash.
Parameters
- window
- 2012-01 to 2025-09 (common)
- sealed
- not used
- code
- research/combo/
- documentation
- https://strategyreview.tradetechanalytics.com/docs/nifty500-trio-comparison.html
Instrument
Equal-weight sleeve of the three strategies (hypothetical); report compares the strategies one by one · capital ₹1,000,000
Known limitations
- Hypothetical combination of three back-tests that share one data pipeline.
- Kelly weights are in-sample estimates (14 years of weekly data) with wide bootstrap ranges.
- The universe is a turnover proxy for the index.
- Not investment advice.
Logs and raw data
Downloads appear once a data window is loaded.
Files are plain CSV so results can be re-computed independently. Trade dates are IST sessions; returns are net of modelled costs and slippage.