# Why We Sell Our Best-Performing Position to Buy the Laggards
If you glanced at the Convictions portfolio today, you would see something that looks backwards: the system sold some of its biggest winner and bought more of the positions that have been losing. Hyperliquid (HYPE, a decentralized trading platform token) is up over 64 percent since the portfolio first bought it. Bitcoin is down 22 percent, and the broader crypto index (CMC20) is down 25 percent. So why did we trim the winner and add to the losers?
The answer is portfolio rebalancing, and it is one of the most counterintuitive but mathematically sound ideas in portfolio management.
The Convictions portfolio runs a fixed target: 30 percent Bitcoin, 30 percent Hyperliquid, 30 percent the top-20 crypto index, and 10 percent Pendle Finance (a protocol that lets you trade future yield streams). These target weights are not just preferences; they are the strategy's way of expressing equal conviction across four distinct bets. When one position grows much larger than intended, it means the portfolio has accidentally concentrated its risk into that single bet, even if you never made a conscious decision to do so.
Here is the math behind what happened. Convictions launched with 10,000 dollars, with 3,000 in each of the three main positions. HYPE went up sharply. Even after today's seven percent decline, HYPE now represents 48 percent of the total portfolio. That means nearly half of the portfolio's future gains or losses depend on one token performing well, when the original plan was never to have more than 30 percent in any single position. This drift happened passively, just by prices moving.
The rebalance today is triggered by a specific rule in the mandate: if any position drifts more than ten percentage points above its target weight, rebalance immediately without waiting for the next scheduled review. 48 percent versus a 30 percent target is an 18-point drift, well past the threshold.
What does the rebalance accomplish? It locks in some of HYPE's gains by converting them into positions that have lagged. It restores the portfolio to its original risk architecture. And from a mathematical standpoint, systematic rebalancing tends to improve long-term returns by a mechanism called rebalancing alpha: you are forced to buy more of things when they are cheaper and sell some of things when they are more expensive, which is the opposite of what emotions would push you to do.
Critically, the rebalance does not change the thesis on any position. The pnlPct metric in this portfolio tracks how much each token's price has moved since we first bought it, and it never resets on a rebalance. HYPE's tracked gain of 64 percent reflects the actual price performance of the token from our entry point. We are simply changing how much capital sits in that position, not our view on the token itself.
One thing worth noting: the rebalance did not happen because of Arthur Hayes (a well-known crypto investor) announcing he exited HYPE, which was news today. The mandate explicitly says no news-driven trades for Convictions. The rebalance would have triggered today regardless, purely because the drift exceeded the mechanical threshold.
The lesson for anyone building their own portfolio: decide your target allocations in advance, set a rebalance rule (a time interval or a drift threshold), and commit to following it mechanically. The hardest part is selling the thing that has been working. But if your original reason for the position was 30 percent conviction, not 48 percent conviction, then letting it grow to 48 percent is letting the market make your allocation decisions for you.