Piertobia Laporta

This document presents a systematic multi-asset algorithmic strategy based on cross-asset statistical arbitrage, filtered by market regime and implemented with adaptive position sizing (ATR-based scaling). The investment thesis exploits medium-term structural inefficiencies (H4/D timeframes) across FX pairs, indices, and commodities, avoiding direct competition with market makers and HFT firms on intraday horizons. The system does not seek entry precision; instead, it builds serial exposure with parameterizable outlier stops, per-asset circuit breakers, and cross-asset diversification. Empirical validation covers 10 years of data (5Y In-Sample / 5Y Out-of-Sample), testing on 130+ independent symbols, Monte Carlo simulations (200 runs), and MFE/MAE analysis. The results demonstrate cross-sectional robustness that rules out dependence on specific market regimes or parametric overfitting.

Operating philosophy: We do not predict price direction. We exploit temporary deviations from the historical spread between two assets, normalized by volatility (ATR), and build progressive exposure. If the market enters a strong trending regime, scaling is halted. An outlier stop (% of balance) closes the exposure only in the event of structural invalidation of the thesis.

Exposure management: The system does not increase exposure geometrically. It uses an adaptive grid:

Number of orders: capped per symbol

Grid spacing: symbol-specific

Volume: calibrated on a target % of balance or % risk per order, with ATR/points fallback

Each order carries independent SL/TP levels proportional to local volatility.

Empirical Validation — Reference Period

Backtest window: 01/05/2016 – 05/05/2026 (10 years)

In-Sample (IS): 01/05/2016 – 01/05/2021

Out-of-Sample (OOS): 01/05/2021 – 05/05/2026

The equity curve maintains consistent slope, volatility, and drawdown profile across both segments, with no evidence of performance decay or parameter degradation in the OOS period.

Monthly Performance ($) — Supporting Backtest Data

2021 (OOS start, from May): May 9,417.31 | Jun 8,986.90 | Jul 9,986.18 | Aug 8,083.46 | Sep 8,335.16 | Oct 9,842.24 | Nov 11,084.69 | Dec 7,412.60 || YTD 73,148.54

2022: Jan 10,462.23 | Feb 1,486.22 | Mar 10,143.65 | Apr 920.84 | May 11,914.23 | Jun 12,544.87 | Jul -2,689.74 | Aug 16,529.02 | Sep 13,020.37 | Oct 11,812.17 | Nov 14,560.17 | Dec 13,319.30 || YTD 114,023.33

2023: Jan 15,170.47 | Feb 9,518.59 | Mar 18,363.97 | Apr 13,970.63 | May 17,505.52 | Jun 15,956.04 | Jul 17,567.72 | Aug 14,798.57 | Sep 10,633.74 | Oct 15,618.53 | Nov 13,743.28 | Dec -2,839.42 || YTD 160,007.64

2024: Jan 9,278.35 | Feb 8,373.83 | Mar 11,624.69 | Apr 17,623.75 | May 11,277.61 | Jun 12,434.59 | Jul 18,326.51 | Aug 16,445.53 | Sep 22,484.77 | Oct 14,147.06 | Nov -2,198.50 | Dec 16,354.54 || YTD 156,172.73

2025: Jan 15,672.40 | Feb 15,409.58 | Mar 13,234.87 | Apr 16,663.43 | May 11,477.25 | Jun 13,245.42 | Jul 18,001.38 | Aug 18,585.02 | Sep 17,116.62 | Oct 18,984.68 | Nov 12,892.55 | Dec 13,408.81 || YTD 184,692.01

2026 (through 05/05): Jan -5,935.58 | Feb 13,652.44 | Mar 18,974.21 | Apr 9,683.29 | May -9,315.47 || YTD 27,058.89

Note: 2021 begins in May (start of the Out-of-Sample period). 2026 data through 05/05/2026.