Hedge fund engineering, built for long-term investors.
AI Labs is the research and engineering function behind the Great Investments Programme. We build the data pipelines, AI models and simulation systems that power the Great Investments Programme — the same systems we use to manage our own capital. Not algorithmic trading. Not a black box. Rigorous investing, thoughtfully translated for the person who owns the portfolio.
The financial world has spent two decades obsessing over speed. High-frequency trading, microsecond arbitrage, event-driven algo strategies. Fortunes have been made there. None of it has any relevance to the person trying to build a portfolio.
We do the opposite. AI Labs exists to apply the same rigour — the data, the filters, the statistical methods, the AI — to the timeframe that actually matters for private wealth: months and years, not microseconds.
We call it Low-Frequency Investing. LFI isn't a slower version of HFT. It's a distinct discipline with different priorities: quality of research, robustness through drawdowns, compounding over decades, and clear thinking under uncertainty. It is the category that retail investing should have been built on. Nobody was building it. So we did.
- Holding period: milliseconds
- Arbitrage, latency, execution
- Institutional only — inaccessible to retail
- Returns from speed, not insight
- Holding period: months to years
- Quality, valuation, drawdown control
- Designed for private investors
- Returns from research, not reaction
Our benchmark isn't another investing course.
It's the communication quality of a private bank — translated for the person who owns the money.
Nine systems, in active daily use.
AI Labs is not a future-state brochure. Every system listed here is running in production. Members of the Great Investments Programme see the outputs. The inputs and methods are documented below.
AI builds portfolios. A hedge fund manager signs them off.
Building an optimal portfolio used to be a job for a desk of PhDs. Now, with large language models paired with portfolio optimisation engines, it's a job any investor can run — if the underlying logic is sound.
AI Labs uses custom AI models to construct and test model portfolios for each member risk profile. The AI generates candidates. Monte Carlo simulates thousands of outcome paths. Alpesh, an FCA-regulated fund manager, reviews the final output. Members see both the recommendation and the reasoning.
- Input Member's risk profile, time horizon, currency, existing holdings
- Process AI candidate generation → Monte Carlo stress test → human validation
- Output Model portfolio, expected drawdown range, path distribution, rebalancing schedule
- Not A personal recommendation. Education and information only. Members make every final decision on their own money.
Thousands of possible futures, not one hopeful projection.
Most "what return should I expect?" conversations end with a single number. "Seven percent a year." "Ten percent a year." A single number is a story — and it is the wrong one to tell an investor.
AI Labs runs Monte Carlo simulations on every model portfolio and every member's bespoke allocation. The simulation generates thousands of possible market paths, drawing on decades of historical return and volatility data. The output isn't a number. It's a distribution.
Members see the median outcome, the good case, the bad case, the 1-in-20 worst case. They see whether their plan survives a 2008-scale drawdown. They see whether the sequence of returns in the decade before retirement is likely to leave them short. Uncertainty, made visible.
- Simulations 10,000 paths per portfolio, refreshed monthly
- Variables Volatility, drawdown depth, sequence-of-returns risk, withdrawal rate sustainability
- Member output A chart they can read in 30 seconds — showing what their portfolio might look like under 10,000 possible futures
AI Labs is invisible. The output isn't.
The worst quantitative investing experiences hand the user a dashboard full of Greek letters. Ours does the opposite. Every complex calculation is translated into a single, actionable output a member can read in seconds.
| What AI Labs does (behind the scenes) | What the member actually sees |
|---|---|
| 14-factor scoring across 17,000 stocks & funds — CROCI, Piotroski, Altman, momentum, volatility, quality, value, growth | A monthly shortlist of ~200 names, ranked, matched to their risk profile. No spreadsheet to interpret. |
| AI-generated model portfolios stress-tested across 10,000 Monte Carlo paths | Three model portfolios — "Sleep Easy," "Balanced Growth," "Higher Conviction." Each with an expected drawdown range and recovery curve. |
| 13F institutional filings ingested weekly, pattern-recognised for sentiment shifts | Telegram alert: "Goldman, Morgan Stanley and Citadel all opened new positions in [sector] this quarter. Here is what that likely means." |
| 1,200-page private-bank outlook document synthesised by LLM agents | A 5-minute video. Three bullet points in a weekly email. One chart. |
| Academic paper on factor robustness (Journal of Finance, 40 pages) | A plain-English one-pager explaining what changed in the evidence, and whether we're updating our approach. |
| Portfolio volatility and drawdown modelling per member holding | The Sleep Easy app: one screen showing each position's worst historical drop, and what it would have done to the overall portfolio. |
Four principles that shape every output.
AI proposes. Humans decide.
Every AI-generated output — portfolio candidate, ranked list, pattern signal — is reviewed by an FCA-regulated fund manager before it leaves the building. No member ever sees unreviewed machine output. The tools are fast. The judgement is slow. That sequence is non-negotiable.
Inputs are checked against inputs.
AI Labs never relies on a single data vendor. Every factor is cross-validated against at least two independent sources. Every AI research summary is audited by a second model with a different prompt architecture. Agreement across independent methods is the standard — because that's what real research looks like.
If we can't explain it, we don't ship it.
Every output a member receives can be traced back through its logic — which factors, which data sources, which AI process, which human decision. No black boxes. No "trust the model." This is education and information, not a faith-based product.
Long-term over clever.
We don't chase what's hot this quarter. Every system AI Labs builds is tested against decades of historical data — through crashes, recessions, interest-rate cycles and real investor behaviour. The test isn't "does it work in a bull market." It's "does it survive the bear."
What AI Labs is not.
Equally important to say, out loud. AI Labs does specific things well. It deliberately does not do several things that adjacent firms claim to.
High-frequency trading
We don't trade at speed. We don't compete with Citadel's microsecond execution or Virtu's market-making. Those firms are magnificent at what they do. It has nothing to do with building a portfolio.
A black-box robo-adviser
Robo-advisers hand investors a pre-packaged portfolio with little transparency. AI Labs is the opposite: the logic, the data, the decisions — every layer is explained to the member. Education, not outsourcing.
A regulated investment manager
Alpesh runs an FCA-regulated fund (Rootbridge Capital) — that is legally and structurally separate from the Great Investments Programme. AI Labs produces education, information and analysis. Not personal advice. Not a regulated service.
A guarantee of performance
Every system here is sophisticated. None is magic. Markets are uncertain; the future isn't in the data. What AI Labs offers is better-informed decisions, a clearer picture of risk, and disciplined process. Not promises about returns.
An algorithmic trading service
Members don't connect a broker API and let our models trade. They see the research, the filters, and the model portfolios. They click every buy and sell themselves. Control stays with the person who owns the money.
For sale to third parties
AI Labs serves one customer: members of the Great Investments Programme. We don't license the engine, we don't white-label the feeds, we don't sell research to hedge funds. It exists to serve private investors. That's it.
Members of the Programme use AI Labs every day.
The Great Investments Programme is the only way to access the outputs of AI Labs. Every system described on this page is built, maintained and used to support programme members — and the team who manages our own capital. Start with a free portfolio review, or enrol directly.