Tactical Defensive crosses below the benchmark

The news this week is a change of side, not a big number. Tactical Defensive went from above the benchmark to below it. A week ago it stood at 2.74% since seeding against 2.24% for the equal-weight market. Now it is 3.16% against 3.90%. It rose 0.40% on the week and still lost ground, because the benchmark moved faster. It sits at -0.74 percentage points.

Our yardstick is the equal-weight panel: 605 names, the same window for everyone, 18 June to 15 October, 3.90%. Quoting a portfolio's gain without the reference over the same stretch is half the story, so we print both.

The worst figure of the week is still Momentum + Core: -4.33% in seven days and -10.28% since seeding, with a NAV of 89,700. A week earlier it was at -6.22%. It is -14.18 percentage points against the benchmark. It is the weakest strategy in the catalogue and there is nothing to soften there.

Two sit above the reference. Core S&P 500 moves up to 4.73% since seeding (from 3.95%), with +0.74% on the week and +0.83 points over the benchmark. The Diversified Basket drops 0.62% on the week, falling from 4.94% to 4.29%, though it stays +0.39 points ahead. Both have 83 days since seeding.

Outside that comparison, Trend BTC+ETH is at 21.84% over 117 days with a flat week (+0.08%) and a NAV of 121,836. Multi-asset 6 classes, seeded on 21 August, is at 0.24% over 39 days with +0.20% on the week. Different windows from the equal-weight panel, so we do not line them up against that number.

On the events work, forward multipliers read: baseline 0.866 over 9,641 cases, earnings 1.135 over 583, guidance 1.058 over 68, exec 0.894 over 250. The effect of an event is measured against the baseline, not from the raw ratio. In backtest, earnings gives 1.089, guidance 1.154 and exec 1.115, with a baseline of 0.964. Exec sits above its baseline in both, but in forward by a much smaller margin (0.894 against 0.866) than in backtest (1.115 against 0.964), and that difference in size is what is worth logging.

The code week was cleanup plus Core. Two bots retired from the catalogue, one published strategy per family, one backtest figure per bot. Core tested from 2022 onwards across 21 start calendars. On crypto, the exit study says the per-coin stop does not work; an A/B with crypto_momentum_risk runs in parallel.

None of this is investment advice. Every strategy here is still running in simulation with notional capital, and we publish the bad weeks on the same terms as the good ones.