What to expect — including the bad parts
The Dorlexa Portfolio is a concentrated, single-theme portfolio. At some point it will fall 35–40% or more from peak to trough. That is not a sign something broke — it is the normal price of the long-term return. The edge comes from staying invested through the downturns. Size your position so you can hold it when that happens.
Expect a large drawdown — it is built in, not a bug
In the survivorship-free backtest (2016–2026) the strategy had a maximum drawdown of roughly −38.7% — top to bottom — with several smaller 15–25% dips along the way. A concentrated AI/tech portfolio that delivers a high long-term return does so because it takes this kind of volatility, not despite it.
For context: even the best long-term equity strategies have survived brutal stretches (tech indices fell ~50%+ in 2000–02, ~50% in 2008, ~35% in 2022). A strategy claiming AI-like returns without large drawdowns is lying. Dorlexa doesn’t.
Why Dorlexa swings more than the market
Three reasons, all deliberate. Concentration: ~20 names, not 500 — fewer positions means bigger swings. One theme: the whole portfolio is AI/tech, the stocks are highly correlated, and there are no defensive sectors to cushion a sector-wide selloff. Winners are allowed to grow: the drift construction lets large positions run to 15%, which lifts returns in an upturn but means a single name’s bad stretch hits harder.
In a broad AI/tech selloff, expect Dorlexa to fall MORE than the S&P 500, not less. That is the design price of the concentration.
What the model does in a crash (and what it doesn’t)
The model has no market-timing overlay. It does not go to cash to dodge a decline — it stays ~fully invested in the top of its field and rides the downturn down and back up.
It only sells a name when it loses its RELATIVE rank (falls below rank 30) — not because the market falls. In a broad crash everything falls together, so relative rankings often change little: the model typically trades LESS in a crash, not more. It does not panic-sell, and neither should you. The retention rule exists precisely to hold long-term winners through a single bad stretch.
If a model that rides drawdowns fully invested doesn’t fit your risk tolerance, the answer is NOT to try timing in and out — it is to own a smaller position.
Sleeve sizing: the most important decision you make
You don’t control the model, but you control how much you put in it. That is your primary risk tool. Treat Dorlexa as a satellite/sleeve, not your whole net worth — it is an aggressive, thematic sleeve, not a core holding. Use only money you won’t need for at least 5 years.
Rule of thumb: imagine your Dorlexa position falls 40% tomorrow. Can you carry that — financially and emotionally — without being forced to sell? If yes, you are probably sized right. If no, the position is too big: cut it to that level NOW, calmly, rather than at the bottom in a panic. Consider building the position over time rather than all in one day.
Behaviour: where returns are actually won and lost
The biggest threat to your return is not the model — it is the impulse to sell when it hurts. The backtest’s returns assume you held all the way through, including the −38.7% stretch. An investor who sold at the bottom and bought back after the recovery would have done dramatically worse than the numbers show.
Practically: decide your sleeve size BEFORE a crash, while you are calm — write it down. When the downturn comes (and it will), your job is to do nothing unless the model sends an alert. Follow the model, not the news. Avoid checking the portfolio daily in a drawdown; it only raises the odds of a panic decision.
When is concern justified?
A large drawdown is NOT by itself a signal that something is broken — it is expected. What would be genuinely concerning is structural, not price-based: if the AI investment theme fundamentally collapsed over many years, or if the model’s selection edge disappeared in the ongoing control tests. We monitor the latter explicitly. A fall in price — even a big one — is a normal part of the journey, not a breach.
Bottom line
Dorlexa is built to maximise long-term, after-tax returns in the AI supply chain — and that kind of return costs volatility. We don’t hide it: expect a −35–40% drawdown at some point, expect to fall harder than the market in an AI selloff, and size your position accordingly. Do that, and your best strategy through the next downturn is the simplest one: stay seated, and let the model work.
Not financial advice. Dorlexa is not a licensed adviser; this page exists to set honest expectations, not to assess your situation. Historical and backtested results are no guarantee of future returns. Assess your own circumstances and seek professional advice if in doubt.