# Northstar daily tactical portfolio ## Purpose This is a 500,000 DKK research model for daily-frequency tactical allocation. It is not an intraday scalping system: a position must be held for at least one completed US trading session, and the model may make at most one batch of trades per US session. It uses no leverage, derivatives, short sales or borrowed money. The strategy is deliberately mechanical. Research documents persistence in price trends and useful information in simple moving-average rules, but neither finding guarantees that a short-horizon implementation will outperform after costs. Research on volatility management supports reducing exposure when risk rises, although later evidence finds that the improvement is not universal. The SEC describes active day trading as highly risky and warns that costs and leverage can make losses substantially worse. Those limitations motivate the liquid-ETF universe, turnover controls and cash reserve below. ## Fixed universe - Equity beta: SPY, QQQ, IWM, EFA and EEM. - Equity sectors: XLE, XLI, XLV and XLP. - Diversifiers: GLD and IEF. - Reserve asset: BIL. Residual DKK cash is permitted. - Benchmark: ACWI, converted to DKK over the identical measurement interval. Only unadjusted prices may be used for executable share prices. Adjusted closes must be used for signals and performance continuity. Splits and distributions must never be mistaken for market returns. ## Daily decision rule 1. Use one common completed US session for every signal. Never use the current execution quote in signal construction. 2. For each risk asset, calculate adjusted-close returns over exactly 5 and 20 trading sessions, its 20-session simple moving average and 20-session annualised realised volatility. 3. An asset is eligible only when its close is above its 20-session average and both its 5-session and 20-session returns are positive. 4. Rank eligible assets by `0.40 × 5-session return + 0.60 × 20-session return`. Retain an existing eligible holding while it remains in the top five; fill empty sleeves from the highest-ranked candidates. This buffer limits churn. 5. The broad regime is risk-on when ACWI is above its 20-session average and its 20-session return is positive. A regime change normally requires two consecutive completed-session confirmations. 6. In risk-on, hold three 30% risk sleeves and 10% BIL. In risk-off, hold at most two 15% sleeves selected from XLV, XLP, GLD and IEF, with the balance in BIL. Whole shares and residual DKK cash cause small deviations from these targets. 7. A holding becomes an exit candidate when it is no longer eligible. It cannot be sold on its purchase date. There is no intraday stop because the model is permitted only one daily decision; risk is controlled through sizing, diversification and the next scheduled review. 8. Do not place an order below the greater of 25,000 DKK or 5% of current net asset value. Do not trade merely to remove a small target-weight difference. 9. The single 18:00 Europe/Copenhagen market-analysis run may execute one consolidated trade batch for a US session, using supported contemporaneous regular-session quotes. Check the journal first; an existing batch for the session prevents any further trades, including on a rerun. If the market is closed, a common signal date is unavailable or execution quotes are not supportable, make no trades. ## Accounting and audit trail - Ignore commissions, spread and tax in reported model P&L, as requested, but enforce the minimum-trade and ranking buffers as if costs matter. - Store quote time, market status, price, FX rate, shares and DKK notional for every trade. Never describe a stale close as a live execution. - Append one value and ACWI benchmark point for each completed decision day. Use matching timestamps and FX where possible; disclose any proxy or mismatch. - Every buy and sell requires a self-contained reason with the raw signal values, ranking, regime and which rule caused the action. Operational valuation notes do not belong in the main Analysis feed. - Preserve the complete trade journal. Never rewrite or delete an old trade. ## Initial model allocation The initial screen uses 3 September 2026 signals and the 4 September completed close as a transparent seeding convention because 7 September is a US market holiday. It is not represented as an executable 7 September fill. | Asset | 5-session | 20-session | 20-day average | Signal score | Initial weight | |---|---:|---:|---:|---:|---:| | XLE | +3.74% | +11.11% | $62.54 | 8.16 | 30% | | XLV | +0.98% | +5.36% | $171.02 | 3.61 | 30% | | EFA | +0.18% | +0.79% | $107.98 | 0.55 | 30% | | BIL | reserve | reserve | — | — | 10% | ACWI closed at $161.94 on 3 September, above its $161.01 20-session average, with a +1.15% 20-session return. The initial regime is therefore risk-on. The three selected risk assets passed every eligibility test and were the three highest eligible scores. BIL is the mandatory reserve sleeve. ## Research basis and limitations - Moskowitz, Ooi and Pedersen, *Time Series Momentum*: https://fairmodel.econ.yale.edu/ec439/mosk.pdf - Hurst, Ooi and Pedersen, *A Century of Evidence on Trend-Following Investing*: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2993026 - Brock, Lakonishok and LeBaron, moving-average and trading-range rules: https://onlinelibrary.wiley.com/doi/10.1111/j.1540-6261.1992.tb04681.x - Moreira and Muir, *Volatility Managed Portfolios*: https://www.nber.org/papers/w22208 - Cederburg et al., evidence that volatility-management results are not universal: https://www.sciencedirect.com/science/article/pii/S0304405X2030132X - SEC, *Day Trading: Your Dollars at Risk*: https://www.sec.gov/about/reports-publications/investorpubsdaytipshtm This is a forward model, not a backtest. Its short 5/20-session implementation is an explicit extrapolation from broader trend-following evidence and must be judged by its subsequent out-of-sample record against ACWI.