Methodology — Operational Workflow, Assumptions, and Current Snapshot
​
1) Pair Universe and Notation
The algorithm is evaluated across a universe of seven pairs. Pair notation specifies leg quantities. For example, 50SLV130PSLV means long 50 shares of SLV and short 130 shares of PSLV.
​
2) Pairs and Indicative Leverage
The leverage figures below are indicative values observed when each pair is tested in isolation under the broker’s margin model at the time of testing. Margin requirements are price- and size-dependent and therefore can vary materially over time.
Leverage is derived from broker margin requirements and is non-linear in size. In practice:
-
Leverage is approximated as: gross notional divided by the broker’s initial margin requirement.
-
Gross notional is approximated as: (absolute quantity of leg A × price of leg A) + (absolute quantity of leg B × price of leg B).
​
Example (4GS16MS): gross notional is approximated as (4 × GS price) + (16 × MS price), and leverage is approximated as that gross notional divided by the broker’s initial margin requirement.
Initial and maintenance margin requirements are typically visible pre-trade in the broker interface.
Pairs and indicative leverage (single-pair tests):
​
-
100EWZ111ILF : 8.10
-
100FENY15XOP : 42.24
-
50SLV130PSLV : 29.36
-
24XOM20COP : 17.84
-
4GS16MS : 11.90
-
11V5MA : 10.49
-
8HD9LOW : 19.30
​
Key nuance: the broker’s assigned leverage/margin can change with lot count (e.g., “2 lots” may be assigned meaningfully different requirements than “1 lot”). In some ratio/size ranges, initial margin can appear extremely low, while maintenance margin is generally still required. Commissions and financing/interest effects remain applicable.
​
3) Capital Segmentation and Monitoring Logic
For simplicity, treat the capital allocated to the algorithm as 100 units. The baseline structure segments this into seven equal sleeves (~14.29% each)—one sleeve per pair—while signals are monitored and acted upon pair-by-pair. Full concurrent deployment across all pairs is not mandatory; operationally, one or several pairs may be monitored and/or traded at a given time.
Due to margin mechanics, it can be feasible in some regimes to carry the seven-pair set with an initial margin footprint on the order of ~10% of total allocated capital, although the realized requirement is broker-, size-, and regime-dependent. Some pairs may consume materially less than their sleeve allocation; any unused portion can remain in USD cash and may accrue broker-dependent interest.
​
4) Data and Charting Requirements
If monitoring is performed via a charting platform (TradingView):
-
Live U.S. equity/ETF market data must be enabled on the charting platform (TradingView – U.S. Stock Markets Bundle).
-
A live market-data subscription is also required at the broker.
-
The charting subscription tier must support synthetic spread visualization (e.g., 50*SLV − 130*PSLV) if that expression is used for the spread chart.
​
5) Signal Output and Execution Convention
Upon each signal/trade trigger, the system outputs:
-
Entry level
-
Position-size percentage
-
Stop-loss level
-
Take-profit level
​
Intrabar convention: if both stop-loss and take-profit are reached within the same bar, the stop-loss is assumed to take precedence. For stop-loss and take-profit evaluation, a bar is considered to have “hit” the level if price touches the relevant threshold (standard high/low touch logic).
​
Trade-level details are not stored as a single consolidated external log file. Stop-loss and take-profit levels remain traceable, and TradingView “strategy results” are used as a practical review layer (entries and failed trades included). The custom simulator maintains a different metrics set, while still retaining the necessary trade-level information internally.
​
6) Simulation Assumptions and Constraints
-
Candlestick bars are used; no alternative bar constructions (Heikin-Ashi, Renko, PnF) are used.
-
The custom simulator does not model partial fills; given the liquidity profile of the selected pairs and non-extreme lot sizing, this approximation is accepted at the current stage.
-
Strategy modifications (e.g., leg weights / lot allocations) are implemented by the quantitative researcher as needed; depending on scope, changes may be completed within hours or up to one day.
​
7) Historical Backtest Snapshot (No Code Modifications)
The following summary is produced over approximately 20,400 bars spanning slightly more than two years, with no manual code changes. Results are current through 2025-12-12 20:25.
For comparability, CAGR is annualized using an approximate ~2-year normalization (the true window is slightly longer; values are therefore approximate). (First - Starting capital End = Last capital )
​
​
​Pair First End Trades Win% MaxDD% CAGR% CAGR/MaxDD
100EWZ111ILF 100 188.00 108 66.66 9.97 37.11 3.72
100FENY15XOP 100 199.41 80 61.26 14.66 41.21 2.81
50SLV130PSLV 100 192.50 90 62.20 9.40 38.74 4.12
24XOM20COP 100 200.84 82 61.44 13.76 41.72 3.03
4GS16MS 100 209.70 83 61.17 10.12 44.81 4.43
11V5MA 100 203.98 164 61.46 11.10 42.82 3.86
8HD9LOW 100 205.00 244 62.00 10.00 43.18 4.03
​
​​
​​​​​​​​​​​​​​​Aggregate / Averages (equal initial capital per pair):
-
Total First Net Capital: 700
-
Total End Net Capital: 1399.43 (Average End Net Capital: 199.92)
-
Average Max DD: 11.39%
-
Total Valid Trades: 851
-
Trade-weighted success rate: ~62.30% (approximate due to rounding in pair-level win rates)
-
Average CAGR (ann., ~2y): 41.37%
-
Average CAGR/MaxDD: 3.71
​
8) Live Monitoring Status and Publication Policy
At present, I am monitoring the signals only and am not executing them as fully deployed live trades; accordingly, I do not recommend that others trade them live at this stage. Broker-specific margin and leverage treatment can differ materially, and leverage observed in isolated tests can compress when multiple pairs are held concurrently. The leverage figures reported here are included for transparency and do not represent curve-fitting, because margin is inherently price-dependent and variable; consequently, effective leverage (within non-extreme bounds) can be adjusted and aligned with the algorithm as needed.
​
While this alignment can be time-consuming to implement responsibly, it has been tested in practice. However, both very low and very high effective leverage are operationally problematic, and leverage should be treated as a bounded risk variable rather than a performance target. Finally, although commissions are often relatively small, they are not currently embedded into the algorithm’s reported accounting; in live conditions, commissions mechanically reduce net capital, worsen maximum drawdown, and therefore reduce CAGR and the CAGR/MDD ratio relative to backtested figures.
​
The historical tables will not be modified; every new signal across the seven pairs will be recorded going forward. For completeness and governance, readers should independently validate the integrity of the reported tables (including whether any values were altered post hoc) using the available source outputs and time-stamped records, to the extent such materials are made available. Signal publication in the Signals section will begin upon completion of the website.
​
SECURITIES
​
EWZ : ISHARES MSCI BRAZIL ETF
ILF : ISHARES LATIN AMERICA 40 ETF
FENY : FIDELITY MSCI ENERGY INDEX ETF
XOP : STATE STREET SPDR S&P OIL AND GAS EXPLORATION & PRODUCTION ETF
SLV : ISHARES SILVER TRUST
PSLV : SPROTT PHYSICAL SILVER TRUST
XOM : EXXON MOBIL CORPORATION (NYSE)
COP : CONOCOPHILLIPS (NYSE)
GS : GOLDMAN SACHS GROUP INC (NYSE)
MS : MORGAN STANLEY (NYSE)
V : VISA INC (NYSE)
MA : MASTERCARD INCORPORATED (NYSE)
HD : HOME DEPOT INC
LOW : LOWE'S COMPANIES INC (NYSE)
​