Skip to main content

Fib Swing Hedge Ladder (Silver)

Status: Candidate - validation gate not passed Backtest: 2012-01-02 to 2026-10-23 (14.8 years) Signal: SILVERM-I daily bars Strategy code: fib_swing_hedge

1. Executive summary​

The strategy holds physical silver and Silver ETF while using short silver futures to hedge downside movements. A short hedge is opened when the daily close breaks below the 23.6% Fibonacci retracement. The hedge is adjusted as price touches deeper or shallower Fibonacci levels. The Fibonacci anchors are frozen at breakdown and remain fixed until a daily close above the frozen ATH closes the hedge.

The whole-sample backtest is promising versus buy-and-hold, but the strategy is not ready for deployment. Out-of-sample validation, higher-cost tests and margin-stress tests are still required. The revised 60 / 30 / 10 capital allocation and the capital rules of section 2.3 (physical silver sold first for margin, ETF capped at its target) have been re-run (sections 9.2 and 9.3); the headline result held.

2. Initial capital allocation​

2.1 Capital input​

The user enters total strategy capital as a silver-equivalent quantity in whole kilograms, not as a rupee amount.

Examples: 48 kg, 67 kg.

  • Minimum input: 1 kg.
  • Decimal quantities are not accepted.
  • The entered quantity represents the total capital budget in silver-equivalent terms.
  • The application converts this quantity to rupees using the selected silver reference price when it needs to calculate monetary values.

2.2 Allocation percentages and whole-kilogram rule​

The target allocation is:

Capital componentTarget allocationPurpose
Physical silver60%Core silver holding
Silver ETF30%Pledged / collateral support for futures margin
Trading-account cash10%MTM and margin-call liquidity
Total100%

Allocation quantities must be whole kilograms. Silver Micro is the smallest futures contract at 1 kg, so the strategy must not create fractional-kg allocation quantities.

Use the following deterministic allocation algorithm:

  1. physical_kg = floor(total_capital_kg x 0.60)
  2. etf_kg = floor(total_capital_kg x 0.30)
  3. cash_kg_equivalent = total_capital_kg - physical_kg - etf_kg

The cash allocation is a silver-equivalent accounting amount. Actual cash is held in rupees in the trading account; it is not a physical 1 kg unit and is not a futures contract.

Examples:

Total capital inputPhysical silver (kg)Silver ETF (kg-equivalent)Cash (kg-equivalent)Total
48 kg28 kg14 kg6 kg48 kg
67 kg40 kg20 kg7 kg67 kg
1 kg0 kg0 kg1 kg1 kg

The examples use the stated floor-and-remainder method. As a result, the cash remainder can differ slightly from the nominal 10% target. The total must always reconcile exactly to the entered whole-kilogram capital.

The percentages are stored in the strategy's JSON (allocation: physical_pct, etf_pct, cash_pct) and the split is done by strategy_defs.allocate().

2.3 Capital rules: cash, physical liquidation and the ETF cap​

Two rules keep the holdings at their agreed shape (revision 4). They apply after every day's MTM, margin check and trades.

Rule A - margin shortage: physical silver is sold first. If cash falls below zero or the margin is not covered (cash + the pledge credit of the ETF, 75% of its value by default), physical silver is sold in whole kg, the least that clears the shortage. The short is then cut so it stays the same % of what is still held (physical + ETF). The ETF is sold only if no physical silver is left. The ETF therefore stays in place as pledged collateral.

Rule B - the ETF is capped at its target; the excess becomes physical silver. The ETF's target is its starting size (30% of the capital, in kg of silver). When the ETF, valued in kg of silver, is 1 kg or more above that target, the excess is sold and bought as physical silver in whole kg at that day's spot (a remainder below 1 kg stays as cash), provided the margin is still covered after the pledge shrinks. The ETF is not topped up when it falls below the target.

How the two rules work together:

  • Gains on the short first lift cash. Cash above the margin required plus the 10% reserve is bought into ETF in whole 1 kg steps (section 7); Rule B then turns whatever lifts the ETF above its target into physical silver. The ETF therefore stays at about 30% and the physical silver grows with the profits; a margin shortage takes it back down (Rule A).
  • The 10% cash is a starting buffer. No rule restores it to 10%: cash is rebuilt from MTM gains and held at margin + 10% reserve, anything above goes to the ETF / physical as above.
  • Sizing and costs: physical silver moves in whole 1 kg (Silver Micro is the smallest contract); the model books no brokerage on physical silver, as in the original spec. ETF purchases and sales carry the ETF costs.

The rules are parameters in the strategy's JSON (liquidity: liquidate_first = physical, etf_cap_kg = 1), so they are part of the revision. Revision 3 and earlier sold the ETF first and had no cap; section 9.2 shows that run.

2.4 Futures hedge ratio (not a capital bucket)​

The short-futures hedge is an exposure overlay measured against combined eligible physical silver and ETF-equivalent holdings. It is not a fourth capital-allocation slice and is therefore excluded from the pie chart.

Fibonacci levelTarget short hedge
23.6%75.00%
38.2%55.09%
50.0%38.86%
61.8%22.91%
78.6%0.00%

Calculate the target hedge quantity from the eligible silver holdings, then round down to whole 1 kg Silver Micro contracts so the actual short does not exceed the target hedge ratio. Never create a fractional futures contract.

2.5 In the Strategy tester: as-of date, capital and the signal​

Silver Arbitrage > Strategy tester has one panel, Signal and set-up, with two inputs:

  • As-of date (default: the last traded day). Everything below is read from the daily closes up to and including that day; nothing after it is used.
  • Trading capital in whole kg of silver (default 100).

From these it shows, for the selected strategy:

  • What to hold: the split of the capital into physical silver (kg), the Silver ETF (kg-equivalent and the number of Silver BeES units to hold at that day's ETF close, rounded down) and the trading-account cash (kg-equivalent and rupees), valued at the silver spot of the day.
  • Minimum cash in the trading account: the margin for the largest short the strategy takes (75% of physical + ETF), at the volatility-estimate margin % of that day and the SILVERM-I price, plus the 10% reserve. MTM losses are paid from cash, so the cash should be at least this; the ETF pledge credit (75% of the ETF value by default) can carry the margin itself. The panel says whether the allocated 10% cash covers it. The margin is an estimate, not exchange data: check the broker's figure.
  • Swing high and swing low in force on that day, with the dates they were set. While a hedge cycle is open these are the frozen ATH and the swing low the levels were built on; without an open cycle they are the running ATH and the accepted swing low.
  • Short now: the target % of the eligible holdings (physical + ETF) rounded down to whole kg, e.g. capital 100 kg gives 60 + 30 = 90 kg eligible; at 38.86% the short is 34 kg.
  • Hedge status: open since the breakdown close of a date and covers on a close above the frozen ATH, or none and the price below which it opens (short 75% = N kg).
  • The ladder table with the price of every level, the short % and the short kg, and a death-zone note when a close has fallen below 61.8% in the open cycle.

Save report (HTML) (next to the Excel button) downloads one stand-alone HTML file of the run on screen: strategy name, code and revision, description and guide link, the parameters (table and the JSON stored in the strategy master), the inputs of the run (capital, split, period), the KPIs, the equity and drawdown charts, the holdings charts, the monthly breakup, the day-one set-up, the latest hedge changes and short trades, and the notes. It has no scripts or outside links, so it can be mailed and opened later.

Run backtest uses the same capital and runs from the From date (default the whole history) up to the as-of date, so the result is what the strategy did before that day. The signal is a daily close from the database, not a live price.

3. Strategy definition​

FieldValue
Display nameFib Arbitrage - Swing hedge ladder
Strategy codefib_swing_hedge
Variantsfib_swing_hedge__mult10, __mult15, __mult25, __mult30
ScorecardArbitrage (hedge income)
Horizon / familySwing / volatility
InstrumentsSilver futures short against physical silver and Silver ETF
Signal timeframeDaily
Signal instrumentSILVERM-I
Revision4
StatusCandidate; out-of-sample split not run; gate not passed

4. Strategy rationale​

The hypothesis is that a daily close below the 23.6% retracement of the ATH-to-swing-low move may precede further weakness toward deeper retracement levels. Instead of immediately selling the underlying silver, the strategy opens a short-futures hedge. It reduces the short at deeper levels and rebuilds it when price recovers to shallower levels.

Physical silver is retained under normal strategy operation, but may be liquidated under the cash-reserve / margin rule above.

5. Fibonacci anchors and cycle rules​

The detailed definitions of ATH, accepted swing low, breakdown, breakout and death zone are maintained in Fib swing rules.

At breakdown, the ATH and swing low then in force are frozen. The Fibonacci levels do not move during that cycle, including after a daily close below 61.8% (death zone).

6. Entry, adjustment and exit rules​

6.1 Initial hedge entry - close-based​

  • Trigger: daily close below the frozen 23.6% Fibonacci level.
  • Action: short 75% of eligible silver holdings.
  • Entry: that day's close.

6.2 Hedge adjustment - touch-based​

  • If the day's low reaches a deeper level, reduce the short to that level's target percentage and book profit.
  • If the day's high reaches a shallower level, increase the short to that level's target percentage.
  • Fill at the level price; on a gap through a level, fill at the day's open.
Level touchedTarget short hedge
23.6%75.00%
38.2%55.09%
50.0%38.86%
61.8%22.91%
78.6%0.00%

Daily OHLC data does not reveal whether the high or low occurred first. The backtest assumes high first when open is above close; otherwise low first. Outcomes dependent on this assumed sequence are assumption-sensitive.

6.3 Full hedge exit - close-based​

  • Trigger: daily close above the frozen ATH.
  • Action: cover the entire remaining short at that close.
  • A recovery above 23.6% without a close above the frozen ATH does not close the hedge.

7. Margin, MTM and ETF management​

The portfolio starts with 10% cash, 30% ETF and 60% physical silver. Cash is the first liquidity buffer. If cash falls below zero or the margin is not covered, physical silver is sold under Rule A; ETF held above its target is converted to physical silver under Rule B (both in section 2.3).

The ETF reinvestment rule in the backtest:

  • Invest cash above required margin plus a 10% margin reserve into ETF.
  • Purchase only in whole 1 kg-equivalent steps and only when excess cash is worth at least 1 kg.
  • ETF holdings are included in the base holdings to which the hedge percentage applies.
  • Rule B then moves the part above the ETF target into physical silver the same day.

Backtest caveat: the original test (section 9.1) used 60 kg physical silver, 25 kg-equivalent ETF and no cash reserve. Section 9.2 uses the 60 / 30 / 10 allocation with the older liquidation order (ETF first, no ETF cap). Section 9.3 is the run with the rules of section 2.3.

8. Backtest configuration​

ParameterBase valueTested values
Breakdown trigger23.6%Fixed
Hedge ladder75 / 55.09 / 38.86 / 22.91 / 0%Fixed
Minimum swing size20 x ATR(50) at breakdown10, 15, 20, 25, 30
ATR length50Fixed
Margin reserve10% of margin requiredFixed

Data and execution assumptions​

  • Signals: SILVERM-I continuous nearest-month futures daily OHLC.
  • Fills: real dated SILVER contract, calculated as level_price x (that_day_settle / SILVERM-I_close).
  • Costs: silverhedge/config/costs.yaml (futures brokerage, exchange charges, STT, GST, stamp duty and ETF costs).
  • Margin is a volatility estimate; no historical margin file was used.
  • Silver BeES is the ETF price proxy.
  • Dated-contract fills are approximations.
  • Daily high/low order uses the assumption described in section 6.
  • 28 SILVERM-I opens and 34 highs/lows were missing or inconsistent and repaired from neighbouring prices.

9. Backtest results - 20 x ATR(50)​

9.1 Original configuration (60 kg physical, 25 kg-equivalent ETF, no cash)​

MetricStrategyBuy and hold
CAGR15.04%9.59%
Maximum drawdown-31.20%-48.90%
Calmar ratio0.480.20

Additional statistics:

  • Hedge changes: 39 (about 2.6 per year).
  • Average short: 44.8% of holdings.
  • Costs: INR 82,649.
  • Closed short round trips: 87; win rate: 63.2%.
  • Net closed-round-trip result: INR 5.4 million.
  • Three open lots at end; unrealised MTM: INR 5.5 million.
  • Closed round trips plus open lots reconcile to the futures MTM ledger to the rupee.
  • Margin calls met by selling ETF: 298.
  • Physical silver sold: none. Unresolved shortfall: none.
  • ETF purchases from excess MTM: 218.

9.2 Rerun on the revised allocation (100 kg capital = 60 physical / 30 ETF / 10 cash)​

Same period (2012-01-02 to 2026-10-23), same rules, engine behaviour of section 2.3.

MetricStrategyBuy and hold
CAGR15.03%9.59%
Maximum drawdown-31.26%-48.90%
Calmar ratio0.480.20
  • Hedge changes: 39 (2.6 per year); average short 44.8% of holdings; costs INR 94,346.
  • Closed short round trips: 87; win rate 62.1%; net INR 6.4 million; three open lots, unrealised INR 6.5 million.
  • Margin calls met by selling ETF: 286; ETF purchases from excess MTM: 238.
  • Physical silver sold: none. Unresolved shortfall: none.

The 10% cash reserve did not reduce the ETF churn noticeably (286 sales against 298 before). The headline result is unchanged to two decimals of CAGR.

9.3 Rerun with the capital rules of section 2.3 (revision 4)​

100 kg capital = 60 physical / 30 ETF / 10 cash, 2012-01-02 to 2026-10-23, physical silver sold first for margin, ETF capped at its 30 kg target.

MetricStrategyBuy and hold
CAGR15.00%9.59%
Maximum drawdown-30.97%-48.90%
Calmar ratio0.480.20
  • Hedge changes: 39 (2.6 per year); average short 44.8% of holdings; costs INR 94,453.
  • Closed short round trips: 95; win rate 67.4%; net INR 6.3 million; three open lots, unrealised INR 6.5 million.
  • Margin calls: 169, all met by selling physical silver (303 kg sold in total); the ETF was never sold for margin. Unresolved shortfall: none.
  • ETF purchases from excess MTM: 252, each followed by a conversion of the excess to physical silver (410 kg bought in total).
  • Holdings at the end: physical silver 167 kg, ETF 30.6 kg-equivalent (it stayed between 30.0 and 31.0 kg throughout); physical silver ranged from 59 to 168 kg.

The headline numbers are close to section 9.2. What changes is the shape of the holdings: the ETF stays at about 30 kg, and the profits end up as physical silver.

10. Parameter comparison​

Original configuration (section 9.1):

VariantCalmarCAGRMax DDChanges / yearRound tripsWin rate
10 x ATR(50)0.4212.97%-30.61%3.710649.1%
15 x ATR(50)0.4514.09%-31.38%2.58160.5%
20 x ATR(50)0.4815.04%-31.20%2.68763.2%
25 x ATR(50)0.4815.04%-31.20%2.68763.2%
30 x ATR(50)0.4815.04%-31.20%2.68763.2%

The 20, 25 and 30 ATR variants produced the same cycles and reported results. The 10 and 15 variants accepted smaller swings and performed less well on the reported metrics. The variants have not been rerun on the 60 / 30 / 10 allocation.

11. Observed behaviour​

  • Across the full sample, the strategy beat buy-and-hold on CAGR and maximum drawdown while retaining underlying silver exposure.
  • Cycles can last years; one 2011 cycle remained open until the 2020 breakout.
  • The same frozen levels are reused throughout a cycle.
  • The original test recorded 4-36 margin calls per year.
  • With no cash reserve, MTM losses on the short were funded by selling ETF and MTM gains bought ETF back, causing frequent ETF turnover (218 purchases and 298 sales). With the 10% cash the rerun shows the same pattern (238 and 286).
  • With the rules of section 2.3 the ETF stays near 30 kg, margin calls are paid with physical silver (169 times) and ETF gains are converted to physical silver (252 times), so physical silver ends at 167 kg. The same churn now shows up as physical silver being sold and bought back.

12. Risks and limitations​

  1. No out-of-sample validation has been run.
  2. Daily OHLC cannot establish the true intraday high/low sequence.
  3. Historical exchange/broker margin data was not used.
  4. Silver BeES is a proxy for ETF pricing.
  5. Dated-contract fills are approximations.
  6. Some missing or inconsistent data was repaired from neighbouring prices.
  7. The section 9.1 results use 60 kg physical + 25 kg-equivalent ETF + zero cash, not the 60 / 30 / 10 allocation; section 9.2 does.
  8. The liquidation sizing (whole kg, least needed) and the ETF cap are implemented as in section 2.3 but have only the whole-sample test (section 9.3); no brokerage is booked on physical silver, so the churn of physical sales and purchases is cost-free in the model. Real physical trading has a spread and costs that would reduce the result.

13. Validation work required before deployment​

  • Out-of-sample / walk-forward validation.
  • 2x transaction-cost test.
  • Margin-shock scenarios.
  • Neighbouring hedge-ladder values.
  • Rerun using 60% physical / 30% ETF / 10% cash (done, section 9.2).
  • Implement and run the physical-first liquidation and the ETF cap (done, section 9.3).
  • Test a cost / spread on physical silver purchases and sales, and a larger minimum liquidation lot.
  • Sensitivity analysis for daily high/low ordering assumptions.

Decision: Keep as a candidate. Do not deploy based only on the current whole-sample results.

14. Run instructions and report location​

Run from Silver Arbitrage > Strategy tester and select Fib Arbitrage - Swing hedge ladder (enter the capital in whole kg), or run:

python -m silverarb.fib_hedge run --capital 100

Original base-run reports:

reports/strategies/fib_swing_hedge/mult20_2012-01-02

This directory contains the daily ledger, events, hedge orders, short trades and seed trades.

15. Revision history​

RevisionDateChanges
12026-11-01First version: close-based start, touch-based ladder, excess MTM invested into ETF.
22026-11-01Parameter JSON includes engine, family, instruments; display name changed to "Fib Arbitrage - Swing hedge ladder". Rules and results unchanged.
32026-11-01Reorganised document; added agreed 60% physical / 30% ETF / 10% cash allocation (stored in the JSON as allocation) and separated futures hedge ratio from capital allocation. Strategy tester takes the capital in whole kg and an as-of date and shows the signal (swing high, swing low, kg to short, ETF units, minimum cash); the backtest runs up to the as-of date. Rerun on the new allocation (section 9.2). Trading rules unchanged.
42026-11-01Capital rules added to the strategy JSON (liquidity): physical silver is sold first, in whole kg, for a margin shortage; the ETF is capped at its 30% target and any excess of 1 kg or more becomes physical silver. Implemented in the engine and rerun (section 9.3). Entry, adjustment and exit rules unchanged.