Building a quantitative portfolio is not "rank stocks by one factor and buy the top 10". Anyone who tried that naive approach quickly discovered that: (a) the portfolio becomes concentrated in a few sectors, (b) turnover eats returns via costs, (c) drawdowns are worse than necessary, (d) in some regimes the factor simply fails for months on end.

Serious systematic stock picking requires five complementary pillars. This post explains each one, and how VORTEX QSP connects all five.

Pillar 1: Multifactorial composite score

Instead of betting on a single factor, you combine several orthogonal factors (with low correlation to each other). Each captures a different market "regularity":

The choice to combine 5 factors (not 3 or 8) is empirical. More factors generate dilution — you end up buying "every stock". Fewer factors increase regime risk. VORTEX QSP fixed 5 before backtesting even began and hasn't changed it.

Equal-weight across the pillars

Why equal weight, rather than weighting by "which factor is best"? Because any optimized weighting is overfitting to the past. Equal-weight is the choice with the lowest performance variance across different regimes — you don't bet on which factor will lead over the next 12 months, you capture a bit from each one.

PRINCIPLE

"Equal-weight is the best estimate of optimal weights when you don't trust your own individual Sharpe estimate." — adapted from DeMiguel, Garlappi and Uppal (2009), who showed that 1/N portfolios beat mean-variance optimization in most out-of-sample windows.

Pillar 2: Hysteresis band

Here's the trick that separates operational strategy from academic backtest. Without hysteresis, a stock that barely misses the top-15 one month re-enters the next. Transaction costs become what defines the result — usually for the worse.

Solution: a stock enters the portfolio when it reaches the top-15. But it only exits when it falls outside the top-25. The 10 slots between 15 and 25 form the hysteresis band — stocks in that zone "stay" in the portfolio if they're already there, and "don't enter" if they aren't yet.

The effect: monthly turnover drops from ~40% to ~12-15%. Costs drop proportionally. And surprisingly, gross returns remain virtually identical — the band introduces minimal noise to selection but eliminates most unnecessary cost.

Pillar 3: Inverse-variance weighting

After selecting 15 stocks, how much weight should each get? Three common options:

  1. Equal-weight (1/15 each). Simple, transparent, but concentrates risk in the most volatile.
  2. Cap-weighted (proportional to size). Becomes basically concentrated IBOV — you lose stock-picking alpha.
  3. Inverse-variance (proportional to 1/volatility²). More volatile stocks get less weight, more stable stocks get more. Balances risk contribution across positions.

VORTEX QSP uses inverse-variance. The practical result: no single stock contributes more than ~12% of total portfolio risk, even if it has the highest composite score. Drawdowns are more controlled, and Sharpe ratio improves.

Pillar 4: Anti-concentration constraint

Without sector constraints, a multifactorial strategy typically concentrates in 2-3 sectors dominating the ranking at that moment. In 2020 it was tech/digital retail. In 2022 it was commodities. Sector concentration = macro bet disguised as stock picking.

VORTEX QSP constraint: no sector can represent more than 30% of the portfolio. If pure selection would violate this, the algorithm replaces the marginal stocks of the saturated sector with the next-ranked stocks from another sector.

Cost: slight return loss during periods when one sector dominates (and it would be optimal to concentrate). Benefit: much less dependent on any specific macro thesis, more predictable drawdowns, and the product makes sense for an investor who doesn't want extreme sector exposure.

Pillar 5: Disciplined rebalance

The most underestimated and emotionally most difficult pillar. All the engineering above only works if you actually rebalance on scheduled dates, even when:

Discipline beats intuition in the vast majority of cases. The average investor's intuition is good at detecting when "something's wrong" — and those are exactly the moments when the backtest would have followed the rules and the human investor quits.

VORTEX QSP generates signals systematically before each trading session opens. Execution is the subscriber's decision. But the signal exists without day-to-day bias.

How the 5 pillars connect

The pillars are not independent — they operate in sequence:

  1. Composite score (pillar 1) ranks the universe by combined z-score of the 5 factors.
  2. Hysteresis (pillar 2) applies the enter-15/exit-25 rule on the ranking, generating the month's tentative portfolio.
  3. Inverse-variance (pillar 3) calculates individual weights based on recent volatility of each stock.
  4. Sector constraint (pillar 4) checks if any sector exceeded 30%. If yes, replaces marginal positions.
  5. Rebalance (pillar 5) executes before market open, with costs explicitly debited.

What this means for the investor

Well-executed systematic stock picking is less about "finding the next Petrobras" and more about a replicable process that captures documented premiums in the literature, with controlled risk. Returns won't be glamorous — there's no exotic pick narrative that multiplied 10x. But historically it beats IBOV with far superior Sharpe, and without depending on perfect investor timing.

VORTEX QSP is exactly that operationalized for retail investors: the 5 pillars running every month, before market open, in direct interface. No reliance on manager opinion, no timing, no narrative of the moment. Process. And honest disclosure of what returned, what drew down, and what failed.

To wrap up

The 5 pillars look complicated in text. In practice they're fixed rules running automatically. The hard part is the discipline to follow the signal — especially when the market is hurting. If that's difficult for you (it is for most), systematization is exactly the value VORTEX QSP delivers: the hard part is already written in code.