MA Arbitrage (Silver)
strategy_code | ma_arbitrage_ladder (Baseline ladder), ma_arbitrage_rule2 (Rule 2) |
| Scorecard | arbitrage (hedge income) |
| Horizon / family | swing / trend following |
| Instrument and timeframe | silver futures short against physical silver and Silver ETF; signals on the MCX spot daily close |
| Revision | 3 |
| Status | candidate |
| Tested | 2012-01-02 to 2026-10-23 (14.8 years), whole period, no out-of-sample split |
Revision history
| Rev | Date | What changed | Why | Result |
|---|---|---|---|---|
| 1 | 2026-11-01 | the two rule sets of the old Trend trade tab saved as strategies | the Strategy tester tab runs strategies from the strategy master | see below |
| 2 | 2026-11-01 | capital split 60 / 30 / 10 stored in the JSON (allocation); capital entered in whole kg | same capital model for every strategy | rules unchanged |
| 3 | 2026-11-01 | capital rules in the JSON (liquidity): physical silver sold first for margin, ETF capped at its target (excess of 1 kg or more becomes physical) | same capital rules as the Fib strategy; see its section 2.3 | results below |
Idea
Silver that trades below its longer moving averages is in a downtrend. Hedge a part of the holdings with futures while the trend is down, and take the hedge off when it turns. The physical silver is kept; only the short moves.
Rules
Signals use the MCX spot cash close of day t; the futures are traded at day t+1's settle (no look-ahead). Numbers are in backend/silverhedge/config/trend.yaml
and in each strategy's JSON in the strategy master.
Baseline ladder (Plan B Strategy A)
- Close below the 50-day average: hedge 25% of physical + ETF.
- Below the 100-day average and 50-DMA below 100-DMA: 50%.
- Below the 200-day average and 50-DMA below 100-DMA: 75%.
- The hedge steps up at once and steps down one level (25 points) after 3 closes in a row above the 50-DMA.
Rule 2 (Plan A)
- 3 closes in a row below the 20-day average: short 50%.
- Then a close below the 50-day average: short the other 50% (100% hedge).
- 2 closes in a row above the 20-day average: cover everything.
Short = hedge % of (physical + ETF) in whole kg, never typed. A cash shortage sells ETF first, then physical silver, and cuts the short to keep the same %; only a signal raises it again.
Parameters
| Parameter | Ladder | Rule 2 |
|---|---|---|
| Averages (days) | 20 / 50 / 100 / 200 | 20 / 50 / 100 / 200 |
| Ladder levels | 25 / 50 / 75 % | - |
| Step down after | 3 closes above the 50-DMA | - |
| First leg | - | 3 closes below the 20-DMA, 50% |
| Cover | - | 2 closes above the 20-DMA |
| Volatility overlay | off | off |
Data and costs
Signals: MCX spot close. Fills on the real dated SILVER contract at the settle. Costs from silverhedge/config/costs.yaml. Margin is a volatility estimate and the
ETF price is a Silver BeES proxy (stated on every report).
Results (physical 50 kg, ETF 30 kg-eq, cash 10 kg-eq; 2012-01-02 to 2026-10-23)
| Baseline ladder | Rule 2 | Buy and hold | |
|---|---|---|---|
| CAGR | 7.64 % | 8.26 % | 9.59 % |
| Max drawdown | -44.60 % | -52.05 % | -48.90 % |
| Calmar | 0.17 | 0.16 | 0.20 |
| Hedge changes | 238 (16.1 a year) | 308 (20.8 a year) | - |
| Short round trips | 188, 38.8% winners | 204, 38.7% winners | - |
| Net short P&L | INR -2.74 million | INR -4.60 million | - |
| Margin calls | 68 | 13 | - |
| Costs | INR 55,626 | INR 123,903 | - |
Both rules lose to buy and hold on return, and Rule 2 also has the deeper drawdown: silver falls faster than the averages react, so the hedge arrives late and whipsaws cost money. The hedge changes 16-21 times a year. Treat these as research, not a result to trade on.
With the capital rules (revision 3; 100 kg capital = 60 / 30 / 10)
| Baseline ladder | Rule 2 | Buy and hold | |
|---|---|---|---|
| CAGR | 7.58 % | 8.03 % | 9.59 % |
| Max drawdown | -44.62 % | -53.64 % | -48.90 % |
| Calmar | 0.17 | 0.15 | 0.20 |
| Short round trips | 195, 39.5% winners | 209, 38.3% winners | - |
| Net short P&L | INR -3.04 million | INR -5.56 million | - |
| Margin calls | 22 | 16 | - |
| Costs | INR 58,344 | INR 138,334 | - |
The conclusion is unchanged: both rules lose to buy and hold on return.
Verdict and deployment
Candidate, not deployed. Not done: out-of-sample split, cost and margin-shock runs, neighbouring average lengths (the old tab had a sensitivity check, no longer on the page). For comparison, Fib Swing Hedge gives 14.99 % CAGR and -31.26 % maximum drawdown on the same holdings and period.