Stop Choosing Between Speed and Fit
Model speed and portfolios that actually fit the client, at scale. Describe your rules in plain English and PortfolioSolver builds trade proposals that honor every one, explains each trade, and watches for drift.
- AI-assisted, math-guaranteed
- You approve every rule and every trade
- Starts from the household targets you already use
Model speed, or a portfolio that fits.You have had to pick one.
Run every client on the same model and you move fast, but the constraints that make a portfolio genuinely personal get left out. Honor those constraints by hand and you can only do it for so many households before the day runs out.
A dozen constraints, one spreadsheet
Gains caps, cash floors, restricted holdings, target weights. You satisfy them by hand, one household at a time.
Promises that live in your head
“She’s sensitive about the Apple shares her father left her.” That is a mental note, not a policy, and it walks out the door when the advisor does.
Drift found at the annual review
A policy break in March surfaces in December, three quarters after it started costing the client.
PortfolioSolver ends the tradeoff. What each client needs becomes a rule the system enforces, on every allocation, for every advisor, every night, at the speed you would get from a model.
From a sentence to a defensible trade proposal
Four steps, and you are in the loop for all of them.
- 01
Describe the rules
Say a client’s constraints the way you’d say them out loud.
“Keep realized gains under $5,000 and don’t sell AAPL.”
- 02
Approve the translation
AI structures your words into rules. You approve each one.
Nothing is applied on the AI’s say-so.
- 03
Solve
Optimization satisfies every rule at once, down to whole shares.
Same inputs, same trades, every time.
- 04
Review and apply
Every trade carries its reason. Apply, compare, revert in one click.
Trade files still go out through your normal workflow.
Everything a model gives you.Everything a model cannot.
Capture what each client actually needs, construct trades that honor all of it at once, and prove the result line by line.
Say it once, in your own words.
The constraint leaves your head and becomes something the system can actually enforce.
Plain-English rules
Speak or type a constraint. AI structures it, you approve it.
Standing client rules
Attach a rule to the household and it holds for every future allocation.
A vocabulary for real constraints
Layered at the firm, the client, and the allocation.
- Asset-class weightstarget
- Cash minimumfloor
- Cash maximumcap
- Realized gainscap
- Hold, do not selllock
- Approved buy listallow
- Do-not-buy listblock
- Minimum yieldfloor
- Trade sizefloor · cap
Every rule, held at the same time.
The engine, and the reason you no longer have to choose between speed and fit.
Optimization-grade construction
PortfolioSolver does not work through your rules one after another. It solves them simultaneously, the same class of optimization institutional desks run, and it works in whole shares because whole shares are what you actually trade. That simultaneity is the thing that lets one portfolio be fast and personal at the same time.
Rule by rule
- Sell to reach the 3% cash floorrealized gains jump to $8,400
- Trim the sale to cap the gainscash drops back under the floor
- Rebalance the weights to compensateand you are back at step one
Each fix breaks the one before it. You stop when you run out of time, not when it is right.
All at once
- Cash floor3.0%
- Gains cap$5,000
- Hold AAPL0 sold
- Target weightson model
One solve. Every rule satisfied together, down to whole shares.
Tax-aware rebalancing
Cap realized gains so a rebalance stays inside the client’s tax budget.
Household targets
Weights pre-fill from the models your firm already maintains.
Check and Fix Minimally
Built the trades by hand? The solver changes the fewest it can.
Nothing you cannot defend.
Every proposal arrives with its reasoning attached, and it keeps getting checked after you approve it.
A reason for every trade
Each trade links back to the exact rule that drove it.
Conflict detection
Some sets of rules genuinely cannot all be met. When that happens you get the rules that collide and the smallest change that clears them, rather than a dead end.
Safe and reversible
Applied provisionally, tagged as solver-generated, reverted in one click.
Nightly monitoring
Re-checked every night. Violations surface as badges and feed your reports.
A solver doesn’t return a checkmark.It returns a bound.
Every rule comes back with your limit, the value the proposal landed on, and whether the constraint is binding or has room to spare.
AI-assisted. Math-guaranteed.
The AI reads your language and explains the results. It does not decide the trades. The portfolio math is deterministic optimization: reproducible, auditable, defensible line by line. That is the part you get to lean on.
What the AI does
- Turns your words into structured constraints
- Explains why each trade was proposed
- Answers rule-check questions on demand
What the math does
- Holds every constraint at the same time, not one after another
- Returns a reproducible, auditable answer
- Tells you when the rules cannot all hold, and names the ones that collide
Same inputs. Same trades.
Run the solve as many times as you like. The answer will not move. That is what deterministic means, and it is why a proposal holds up when someone asks how you got there.
- SELL40VTI
- BUY12BND
- HOLDAAPL
One run so far.
Client rules become firm infrastructure
The rules one advisor captures for one client end up serving the whole practice: memory that survives a departure, a supervision trail, and the same policy applied by everyone.
Institutional memory
Client preferences live in the system, not in one advisor’s head. Transitions stop being a risk to client-specific promises.
Supervision and defensibility
Every run records what was checked, what passed, what broke, and why. That is a documented supervision trail, not a reconstruction after the fact.
Consistency at scale
The same rules produce the same trades for every advisor. Policy enforced by construction, not by memo.
Capacity
Rebalancing collapses from hours per household to seconds, so the same team serves more clients without making any of them generic.
Human-in-the-loop by design
AI translates and proposes; advisors approve. A stronger story with regulators than “the AI trades for you.”
What advisors ask first
Short answers. We are happy to go deeper on a call.
Does PortfolioSolver place trades?
No. It proposes and checks trades. You review every proposal and generate trade files through your existing workflow. Nothing reaches a custodian without you.
Is the AI deciding what to trade?
No. The AI translates your language into rules and explains the results. The portfolio math is deterministic optimization, so the same inputs always produce the same trades.
Do I have to give up my models?
No, and that is the whole point. PortfolioSolver starts from the household targets your firm already maintains, then layers each client’s rules on top. You keep the model. The client still gets the exceptions that matter to them.
How precise are the tax figures?
Realized capital gains are calculated at the tax-lot level on taxable accounts.
Do I have to give up building trades by hand?
No. Run a rule check on your own draft to see exactly what breaks, then use Fix Minimally to have the solver adjust the fewest possible trades. It preserves your intent rather than replacing your work.
See PortfolioSolver™ on your own book
Book a walkthrough and we will show you how your firm’s rules translate, what the solver proposes, and how the nightly checks work.
Already a FinTurk client? Talk to your account team about switching it on.
Important disclosures
- PortfolioSolver proposes and checks trades. It does not place or execute them.
- Figures and screens shown on this page are illustrative.