How a Rebalance Works
This page walks through, in order, exactly what happens from the moment the agent notices something to the moment (if it happens) that a trade executes. Nothing here is simplified to the point of hiding a step — if you want to understand exactly what AERA does on your behalf, this is the page that answers it in full.
Step 1: Continuous monitoring, not scheduled check-ins
Unlike a human advisor who might review your account once a quarter, AERA's agent is always watching. Specifically, it continuously tracks:
- Your current holdings, compared against the target allocation you set (for example: "I want roughly 60% in a mix of tech and healthcare stocks, and 40% in stable yield").
- Live prices and volatility for every Stock Token you hold, so it knows not just what things are worth right now, but how much they've been swinging.
- Available yield rates on approved on-chain lending positions (for example, through Morpho), so it knows what the "safe" side of your portfolio is currently earning.
Step 2: Recognizing that something's changed enough to matter
A rebalance isn't triggered just because a price moved slightly — that would mean constant, costly trading for no real benefit. Instead, the agent is looking for two kinds of triggers:
- Drift: your actual holdings have moved meaningfully away from your target. A simple example: if one stock in your portfolio doubles in price while everything else stays flat, it now makes up a much bigger share of your total portfolio than you originally wanted — increasing your risk in a way you didn't explicitly choose.
- Opportunity or risk shift: something about market conditions has changed enough that your own stated rules call for a different mix — for instance, if yield rates rise significantly, your rules might call for shifting a bit more toward yield and a bit less toward equity, simply because that's what you told the agent to optimize for.
Step 3: Building a proposal, not just executing blindly
Before anything happens, the agent puts together a specific proposal: exactly which asset it wants to trade, how much, and — just as important — a plain-language explanation of the reasoning. This isn't an afterthought bolted on for marketing purposes; the explanation is generated directly from the same data that triggered the decision, and it's saved permanently alongside the proposal, whether or not the trade ends up happening.
Example of what this might look like in practice:
"Your tech holdings grew from 30% to 47% of your portfolio after a recent price increase, above your 35% limit. Proposing to sell approximately 12% of tech holdings and move the proceeds into your approved yield position, bringing tech exposure back in line with your target."
Step 4: Checking the proposal against hard, on-chain rules
Before the agent can act on its own proposal, the vault contract itself checks it against the rules you set — completely independent of whatever the agent "believes" is a good idea. This includes things like: is the asset it wants to trade into on your approved list? Is the trade size within the maximum you allowed? Would the trade's expected price impact (slippage) stay within your tolerance? Has the agent already made its allowed number of trades for this period?
If the proposal fails any of these checks, it simply does not execute. There's no override, no exception, no "trust me" — a failed check is a hard stop.
Step 5: A second check for anything merely unusual
Some problems don't look like a rule violation on paper, but still look wrong. Imagine the agent proposes something technically inside all your limits, but wildly different from anything it's done before — an unusually large trade, an unusual number of trades in a short period, or a rarely-used asset suddenly showing up. A separate system (we call this the circuit breaker) checks for exactly this kind of pattern. If something looks like an outlier, execution pauses and the proposal is held for manual human review, rather than proceeding automatically just because it technically passed the hard rules.
Step 6: Execution
If a proposal clears both checks, the trade actually executes on-chain. Rather than building custom trading infrastructure from scratch, AERA routes trades through established, already-live liquidity providers on Robinhood Chain (such as 1inch and 0x), which helps ensure you get a fair, competitive price given how much liquidity is actually available at that moment.
Step 7: Recording everything, always
Whatever happens — the trade executes, gets blocked by a rule, or gets paused for review — it's written permanently into an immutable log: the market data that triggered it, the reasoning behind it, and the outcome. This log is available to you at any time through your dashboard. Nothing about this process happens somewhere you can't later go back and check.