The application
Projections
Monte Carlo scenarios, sustainable withdrawals, an income goal solver and a historical replay, all computed on your own positions.
On this page
Choose a time horizon, assumptions and contributions or withdrawals to compare possible portfolio outcomes. Projections start from your current positions, valued like the dashboard: an asset without a price is left out, and the starting point reads ≈ with the number of assets left out. The borrowing cost uses the average rate of the loans whose rate is known, applied to all the debt and marked ≈ when some are missing; with no rate known, the cost is not counted and the page says so. They help you explore a question; they do not predict what will happen.
Pro plan
Projections are included in Pro.
Hundreds of paths
Multiple possible paths with changes in prices, yield and borrowing costs.
The simulation first, then three scenarios
The figures at the top of the page, the band and its median come from the simulated paths. The Bear market, Base and Bull market scenarios, folded under them, are fixed-return paths: neither percentiles nor forecasts.
Withdrawal reference levels
The curve shows the share of simulations able to fund the selected withdrawal through the time horizon.
What is left out
Taxes, future reallocations, changes to your debt and your own behavior are not modeled. The result informs a decision; it does not replace it.
Start with the time horizon, then enter expected contributions or withdrawals. The figures at the top read the base scenario. To see what changes under another assumption, use Explore a scenario: it reruns the simulation on the same paths with the yield level and the crypto return you set, from −80 to 100% a year for all your crypto, and gives the median of the path-by-path gaps. A custom fixed-return scenario can also be added in the advanced options. By default, withdrawals rise by 2% a year with inflation; the advanced options also set a yearly compression of yields, from 0 to 20%.
Simulations include large variations and linked movements between assets according to the scenario's assumptions. Results depend on those assumptions and available data; a simulated probability is not a guaranteed real-world probability.
A loan's collateral counts in the projected net worth and follows its market; so does a crypto or tradfi debt, which weighs more when its market rises. A loan whose debt is the asset it pledges is therefore not liquidated by a price move, and a crypto debt against stablecoins can be liquidated on a rise: the lending panel then gives the price of the debt that liquidates it. When the collateral of the loans whose liquidation threshold and prices are known, weighted by those thresholds, falls below their debt, the simulation closes those loans: their debt is cleared and the protocol takes collateral worth that debt plus a liquidation penalty. A loan whose threshold or price is unknown is left out of that check and keeps its debt; the lending panel names it. The crypto collateral of the loans counts, with the crypto you hold, in the volatility the simulation draws for your crypto. Yield and borrowing rates are annual yields, turned into monthly rates by compounding, as on the Stable Yield page. Interest that the capital in yield cannot pay adds to the debt. A simulated sale credits its full proceeds: the share with no destination in your trading plan stays as cash, and a plan with no destination at all sends 70% to the main yield pocket and 30% to the reserve.
Understand the calculation
The chart band runs from the 10th to the 90th percentile of the simulated paths: about one path in ten ends below it, one in ten above it, and its middle line is the simulated median. The median often sits below a simple compounding of the current rate: a -20% month followed by a +20% month does not get back to even, and the simulation accounts for it where smooth compounding does not.
The fixed-return scenarios, folded at the foot of the lead card, use fixed returns, without volatility. Bear market and Bull market apply, every year, the low and the high crypto return that the simulation reaches 1 time in 10 over the whole horizon; Base applies the average returns. Because they ignore volatility, their results differ from the simulation: Base usually sits above the simulated median, which leads the page.
For lending positions, the projected health factors and the distance to liquidation are read with your own warning and critical levels, as everywhere else in the app: the distance is 1 − 1/HF, so a health factor at your critical level is the same alert in both. The liquidation banner states the share of paths in which a liquidation occurs, "< 1%" when it is a single path out of 300, and, over a horizon longer than a year, the share within the first 12 months. A loan below 1.00 or below your critical level is named in a critical warning, even when the combined health factor looks sound. A loan left out of the scenarios is named with its reason: a liquidation threshold to enter on the position, a collateral without a price, or a debt valued at zero.
The curve estimates the share of paths that fund withdrawals through the time horizon. Comfort corresponds to at least 95% success and median capital preservation; Stretch marks the amount at 90%; Danger is crossed below 75%; between 75 and 90%, your withdrawal reads Between stretch and danger. When the median ends below the starting capital even with no withdrawal, no withdrawal can preserve it, and comfort reads the 95% level alone. The zone of your own amount is the same at the top of the page, on the analysis row and in the curve panel: it reads the 300 simulated paths at that amount, and the curve uses the same paths at every level. When some paths fall short, the top of the page gives the typical month and the share of paths it concerns.
The income goal also estimates required capital and the gap from your current position, at your current average yield rate, after the interest on your loans. It is not an income commitment.
With a fixed amount, the sustainable withdrawal row reads the probability of serving it in full, the same figure as the top of the page. With a share of the yield, the withdrawal follows the yield and cannot fall short: the row then shows the comfort ceiling as a fixed monthly amount, and no zone warning applies.
The historical replay applies real past price windows, a bear market or a recovery, to your current positions. Only prices are replayed; yield and borrow rates are today's, held constant (the yearly yield compression of your parameters still applies, and the panel says so), so it is a price-shock test, not a reconstruction of the period. A window with a month missing from the stored history is not offered. The panel gives the lowest net worth and the month it is reached, and dates a liquidation in the calendar month it would have happened. The sensitivity analysis moves one factor at a time on the base path at fixed returns, by one typical deviation (1 sigma) of its average over your horizon, as the simulation draws it (yield decay and withdrawal inflation move by 3 and 2 points), and shows the lower and the higher outcome for the net worth and the monthly income. A combined bear shock applies the three correlated moves together: crypto down, yields compressed, borrow rates up. Their interaction is shown when it weighs at least 0.1% of the net worth.
Under the parameters, two cards gather the analyses. Risks: the sustainable withdrawal, lending under stress, the historical replay and the sensitivity analysis. Levers: the income goal, the trading plan and the yearly summary, with the scenario explorer and the optimizer comparison at the foot of the card. Each row shows its figure and opens the full analysis in a side panel.
The trading plan row reads the plan on the simulated paths, compared path by path with the same paths without the plan: the median effect on the final net worth, the share of paths it raises, and the change in probability in points. A negative figure means that on the typical path the plan lowers the final net worth under these assumptions; switching the plan off in the parameters is one of the options. The yearly summary reads the simulation too: the median net worth at the end of each year, its range in 8 cases out of 10 and the median monthly income.
The scenario explorer reruns the simulation with two other values, the yield performance and the crypto annual return, and reads each path against the same path at the base values: the median gap on the final net worth and on the income at the horizon, and the probability before and after. The optimizer comparison takes the three allocations Stable Yield offers for your yield capital, at the rates the page opens on, and replays the base path at fixed returns with each rate: the gaps show the effect of the rate alone, not the swings of the market. An allocation that places nothing is left out.
The copilot analyzes these projections only when you ask, with the parameters on screen. When you change them afterwards, the card says that its reading is dated.
Not modeled
Taxes, future manual reallocations, changes to your debt, and your own behavior. Trading plan effects, applied by default when you have a plan (a switch in the parameters turns them off), assume perfect and immediate execution of every signal. Projections inform decisions; they do not replace them, and leaving things as they are is always one of the options.
Frequently asked questions
How should I read the band and the fixed-return scenarios?
The band and its median are percentiles of the simulated distribution: about one path in ten ends below the band, one in ten above it. The Bear market, Base and Bull market scenarios, folded under the lead card, are fixed-return paths: low crypto returns reached 1 time in 10, average returns, and high crypto returns reached 1 time in 10. They show what those returns would give without volatility. Neither these scenarios nor the band are guarantees or forecasts.
Why does the median look lower than a simple rate extrapolation?
Returns compound: after -20% and then +20%, an initial 100 becomes 96. The simulation accounts for this effect, but the result still depends on the chosen assumptions.
What is the sustainable withdrawal analysis?
It compares your withdrawal with the share of simulations that keep it funded through the time horizon. The 95%, 90% and 75% reference levels help you read the curve. Between levels, use the displayed probability rather than assuming a guarantee.
How does the income goal work?
Set a target monthly income, by default your own withdrawal, kept while the page is open; the solver returns the required capital, the gap from where you are, the rate that would be needed with your current capital, and an estimate of months to reach the target with your contributions. The income is net: the interest on your loans comes out of the yield first, so the required capital and the rate include it. A target today's net income already covers reads Already reached. The withdrawal covered 9 months out of 10 is read on the base scenario; each fixed-return scenario gives its own below it. With no yield rate known, the required capital and the gap read – with the reason; with no capital in yield, so does the rate needed.
How is the trading plan's effect measured?
The simulation runs a second time without the plan's tiers, on the same paths, and compares the two path by path. The row shows the median of those differences on the final net worth, the share of paths the plan raises, and the change in the probability the page reads (serving every withdrawal for a fixed amount, preserving the capital for a share of the yield), in points. The gap between the two medians is not shown: it can read as a gain while the plan lowers most paths. A sale in a steady rise always costs on a smoothed path; on the simulated paths it can also protect. The panel then lists the tiers the plan would trigger on the fixed-return paths. Execution stays manual: the projection assumes every tier executed as configured. As on the Trading page, a sell tier sells its share of the crypto held when the first sell tier fires, and a buy tier spends its share of what the source held when the first buy tier fires: tiers of 30, 30 and 40% sell the whole holding.
What does the historical replay do?
It replays your current portfolio through real past windows (for example the 2022 bear market or the 2023 recovery), using actual price paths. Only prices are replayed; yield and borrow rates are held constant. Read it as a price-shock test, not a full reconstruction of the period.
What do projections not model?
Taxes, future manual reallocations, changes to your debt, and your own behavior. Trading plan effects, applied by default when you have a plan, assume perfect and immediate execution of every signal. Projections inform decisions; they do not replace them.
See also