When one asset in a portfolio rises sharply, it does not just make you money. It also quietly changes the shape of your portfolio in a way that can expose you to risks you never agreed to take on. This question came up today in the Convictions portfolio, where Hyperliquid (HYPE, a decentralized trading platform that lets crypto users trade futures contracts without going through a centralized exchange like Binance or Coinbase) has risen roughly 40% since the portfolio launched two weeks ago. That gain is a good thing. But it has also pushed HYPE from its 30% target weight to nearly 38% of the total portfolio value, and now a rebalance is coming.
Here is how the mechanics work. When you launch a portfolio with four assets at equal or fixed weights, each one starts as a specific fraction of the total value. But assets do not move in lockstep. Over days and weeks, the faster-rising ones grow into a larger share and the slower ones shrink. The portfolio that started as 30/30/30/10 slowly becomes 38/25/27/9. The structure you designed has quietly drifted away from you.
The reason this matters is concentration risk. If you deliberately chose to put 30% in an asset, that was a judgment call: at 30%, you wanted meaningful exposure but not the kind of dominance that would make the portfolio's fate depend on one bet. At 38%, the same asset now controls a much larger share of whether the portfolio goes up or down on any given day. You have taken on more risk than you decided to accept, not by trading, but simply by letting a winner run.
The fix is called rebalancing. You trim the winner and use the proceeds to top up the losers, restoring everyone to their target weight. The idea can feel uncomfortable, because you are literally selling something that has been working. But the logic is sound: you are not selling because the thesis is broken, you are selling because the position now represents more of your portfolio than the thesis justifies. You still believe in Hyperliquid; you are just re-sizing it back to the bet you intended.
The Convictions mandate handles this with a simple two-tier rule. If a position drifts more than 5 percentage points from its target weight, it gets flagged as a rebalance candidate. If the drift exceeds 10 percentage points, the rebalance happens immediately, regardless of the day. For drifts between 5 and 10 percentage points, the execution waits until Monday. This delay serves a purpose: markets are noisy, and a surge that looks like a drift on Friday might partially reverse over the weekend. Waiting a few days avoids selling into short-term momentum only to have the price settle back anyway.
This framework comes from standard portfolio construction theory, formalized in academic finance since the 1950s by Harry Markowitz's work on modern portfolio theory. The core insight is that the expected return of a portfolio is the weighted average of its parts, but the risk is lower than the sum because assets do not move together perfectly. When you let weights drift, you are implicitly changing your portfolio's risk profile in ways that may no longer match what you set out to do.
The practical takeaway is this: in a well-designed portfolio, a winning position eventually creates an obligation to trim. Not because the asset is bad, but because the portfolio structure has to mean something. If you never rebalance, you end up with a portfolio that is whatever the market made it, not the one you chose. Rebalancing is how the mandate stays honest over time.
One important caveat: rebalancing has a cost. Every sale is a taxable event if the portfolio is held in a taxable account, and there are transaction costs. This is why most well-run portfolios do not rebalance continuously but use threshold rules like the ones Convictions follows. The goal is to stay close enough to the target without churning the portfolio into unnecessary friction. The right threshold depends on the portfolio's cost structure and how volatile its assets are. For a crypto portfolio, where assets can move 20-30% in a week, a 5-percentage-point drift threshold is relatively tight.