What Data Does Acubic Use? Methodology Explained

Key takeaways

• Acubic's portfolio engine uses standard third-party market and reference data: historical prices, company profiles and sectors, financial ratios, and US Treasury yields as the risk-free rate.
• It uses no sentiment feeds, news, chart patterns or alternative data, and it does not try to predict short-term price moves.
• Your constraints are applied first; eligible tickers are then ranked by a method you choose, such as market capitalisation, trailing Sharpe ratio, momentum, lower volatility or company fundamentals, or a blend of them; returns-based measures look back three months to three years (twelve months by default).
• Weights come either from one of three portfolio optimisers (hierarchical risk parity, mean-variance or CVaR) or from simple rule-based weightings such as equal weight, rank-based weight or capped weights. Nothing is preselected: you choose the method, and the Beginner and Intermediate menus mark a rule-based weighting as the recommended option for your risk level.
• On a connected account the target is recomputed on the rebalance frequency you choose (monthly, quarterly by default, or annually); the 15-minute check is only a polling interval.

Every article on this blog eventually gets asked the same question in different words: what is actually driving the numbers. Acubic proposes portfolios and, for connected broker accounts, keeps them aligned with a target over time, but a reasonable reader wants to know what data feeds that process, how assets are ranked and weighted, and how often anything actually changes. This article answers all three plainly, using the same terms the underlying method uses rather than marketing language. For the step-by-step of the product itself, see how does Acubic work; this piece goes one level deeper into the mechanics.

Where the numbers come from

Acubic's portfolio engine runs on licensed third-party market data from Financial Modeling Prep (FMP), served through Acubic's own market-data store. That single source feeds several distinct types of information into the pipeline, not just a stock price:

• Historical price series, used for the return and risk calculations behind ranking and allocation.
• Company profiles and sector classifications, used to describe and group holdings.
• Financial ratios and key company metrics, used by some ranking methods.
• US Treasury yields, used as the risk-free rate in risk-adjusted return calculations such as the Sharpe ratio.

None of this is proprietary sentiment data, alternative data, or a signals feed. It is standard, third-party market and reference data, the same category most portfolio software and research desks draw on. What differs is how it gets used, which is the subject of the next few sections.

Building the investable universe

Before any ranking or weighting happens, Acubic narrows a starting universe of several thousand listed stocks and funds, drawn from its market-data universe and limited to what the supported brokers can trade, down to the tickers that fit the preferences you gave in the guided conversation, such as your risk tolerance and any exclusions or limits you set. This screening step is a hard filter: a ticker either meets your stated constraints or it is not eligible for selection, regardless of how attractive its numbers look on their own. A stock you have explicitly excluded, or one that falls outside a limit you set, is removed before ranking begins, not weighted down afterward and hoped away.

The order matters here. Constraints are applied first, on the full universe, so the ranking and allocation steps that follow only ever see eligible candidates. That is different from building an unconstrained portfolio and then trying to patch it toward your preferences afterward, which tends to produce compromises rather than a portfolio that actually reflects what you asked for. We cover why constraint-first thinking matters in top portfolio construction mistakes, and how constraints shape the final allocation in how to optimize a portfolio that fits risk.

How assets are ranked before they go in

Within the eligible universe, Acubic ranks candidates using a ranking method you choose. The main ones are market capitalization, trailing risk-adjusted return (Sharpe ratio), price momentum measured over a trailing window, lower volatility, and company fundamentals such as value and quality measures, and they can be blended. The returns-based methods look back over a historical period you can adjust, from three months up to three years, with twelve months as the default. An advanced ranking selector exposes this choice for users who want it; most people never touch it and get the twelve-month default automatically.

It is worth being precise about what this ranking step is and is not. It orders candidates by a historical statistic measured in months, not a forecast of where a price is headed tomorrow. A stock with strong twelve-month momentum is not a prediction that it will keep rising; it is one input, drawn from the past, used to decide which tickers are eligible for the optimization step that follows. The full explanation of how ranking feeds construction is in quantitative portfolio construction that holds up.

The choice of lookback window changes what the ranking rewards, and that trade-off is worth understanding rather than treating as a technical detail. A shorter window, such as three or six months, reacts faster to a company's recent numbers but is more exposed to a single unusual quarter. A longer window, up to three years, smooths out short-term noise at the cost of reacting more slowly when a company's underlying picture genuinely changes. Twelve months is the default because it is a reasonable middle point for most goals; it is not presented as the objectively correct choice, which is why the option to change it exists for users who want to.

The Sharpe ratio used here is the standard measure of return per unit of risk. Its inventor, William Sharpe, described it as "designed to measure the expected return per unit of risk for a zero investment strategy" (Sharpe, 1994), which in practice means return above the risk-free rate divided by volatility. That is why Treasury yields appear in the data list above.

How weights are allocated: HRP, MVO, and CVaR

Once a ranked shortlist exists, Acubic hands it to one of three portfolio construction methods to decide how much to hold of each name.

The optimisers are not the only route. On the Beginner and Intermediate strategy menus the recommended allocations use simple rule-based weightings instead, such as equal weight, weights by rank, or weights capped at a fixed share per holding. None of these methods involves predicting which asset will outperform next; they turn the ranking order, market capitalisations or historical risk and return statistics into a set of weights that satisfy your constraints. The sensitivity of MVO is well documented: a 2009 study of 14 optimisation models found that "the gain from optimal diversification is more than offset by estimation error" (DeMiguel, Garlappi and Uppal). The full comparison, and why Acubic offers HRP and when it fits, has its own article.

How often the portfolio actually rebalances

For connected broker accounts, a background process checks every fifteen minutes for accounts that are due for a rebalance. That interval is a technical polling frequency, not how often your portfolio actually changes. The interval that matters to you is the rebalance frequency you choose when you connect a broker: monthly, quarterly, or annually, with quarterly as the default. When your chosen interval falls due, Acubic recomputes target weights using the same method and constraints used to build the portfolio, then either flags the trades for your approval or places them automatically, depending on the automation mode you selected.

This is a deliberate design choice. A portfolio that rebalanced every time a price moved would be reacting to noise, not managing risk, and would generate far more trading activity than the underlying method calls for. A fixed calendar cadence keeps the system doing what it is meant to do: bringing weights back toward a stated target on a schedule, not chasing daily price action. The mechanics of timing and thresholds are covered in more depth in our portfolio rebalancing strategy model guide.

Two automation modes govern what happens once a rebalance is due, and they apply the same underlying calculation differently. In one-click mode, the default, Acubic computes the trades needed to bring the portfolio back to target and flags them for your approval; nothing reaches your broker until you act on it. In full-auto mode, available for users who explicitly opt in, the computed trades are placed automatically at the scheduled date. Either way, the calculation itself, and the schedule that triggers it, is identical; automation mode only changes whether a human confirms the trade before it happens.

What this data pipeline does not do

Being specific about the limits is part of being specific about the method. Acubic's data pipeline does not attempt to predict short-term price moves, does not use technical indicators or chart patterns to time entries or exits, and does not read news or sentiment to trade around events. Technical values such as moving averages can be used as screening filters, some built into the Beginner and Intermediate presets, never as timing signals. The ranking methods select which assets are eligible based on historical statistics measured in months; the allocation methods decide how much to hold of each. Neither step is a forecast of tomorrow's price, and nothing in the pipeline is designed to react to a single day's headline or a single day's price swing.

This is also why a rebalance only happens on the schedule you set, not whenever the market moves. If the data pipeline were trying to call short-term direction, a fixed monthly, quarterly, or annual cadence would not make sense; you would want it reacting constantly. Instead, the schedule reflects what the underlying statistics are actually measuring: risk relationships and return patterns that hold over months, not signals that expire by the next trading session. Acubic also does not promise a return or guarantee that any portfolio will outperform a benchmark. It builds and maintains a portfolio according to a stated, quantitative method and the constraints you set, and the results should be judged on that basis, not on whether any single position moved in your favor this week.

How to check this yourself

None of this needs to be taken on faith. Build a portfolio without connecting a broker and look at which tickers were selected and why, using the ranking method and lookback window as a reference point. Read the fuller explanation of the method on the methodology page, or start directly with the AI portfolio builder if you would rather see it in action. Our guides collection covers individual pieces of this process in more depth, and you can read about who runs the product on the about page.

If you are earlier in the process and want the general method before the product, how to build an investment portfolio from scratch covers the steps any investor can follow, with or without software.

The data pipeline exists to serve one output: the target weights your connected account is moved toward. How that final step behaves depends on the broker, since a standard eToro connection presents each rebalance for approval while a Trading 212 connection can run it on a schedule.

This article describes method and data, not results. It is general information, not financial advice, and capital is at risk.

Sources

1. The Sharpe Ratio, William F. Sharpe, The Journal of Portfolio Management, Fall 1994 (author's copy, Stanford University), accessed 12 October 2026.
2. Optimal Versus Naive Diversification: How Inefficient is the 1/N Portfolio Strategy?, V. DeMiguel, L. Garlappi and R. Uppal, The Review of Financial Studies 22(5), 2009 (abstract via RePEc IDEAS), accessed 12 October 2026.