Tracing a Stacked SAFE Deal to Your Real Converted Ownership

Before you wire an angel or seed check, the number that matters is not the headline cap on the SAFE. It is what your check actually converts into once the whole stack of SAFEs and notes converts and the priced round carves a fresh option pool. This guide is for angels, syndicate leads, and platform and talent partners who get handed a term sheet plus an existing cap table and have to answer one question: what do I really own after everything settles. I carry a single deal from a headline cap to a true fully diluted percentage, showing the arithmetic at each fork and the three wrong turns that quietly shrink the answer.

Generic explainers define pre-money, post-money, discount, and pool one term at a time. That leaves you unable to run the sequence on the deal in front of you, because the terms interact. Here I run them together on one worked case so you can copy the steps onto your own sheet.

The deal I am tracing, and the three views I need

The worked deal: a company has raised a stack of post-money SAFEs and one convertible note, and is now raising a priced round that carves a new option pool. My job is to trace one investor's converted ownership through all of it.

Before any math, I insist on three separate views of the cap table, because a percentage is meaningless without a denominator and a date.

An investor reads their true stake off the pro forma fully diluted view. Term sheet prices and target ownership almost always assume full dilution, including the new option pool. If someone quotes me a percentage off the current outstanding basis, it is flattering and wrong for negotiation.

The trap hides in the words "fully diluted." One table may include the whole option pool but show SAFEs on a separate line; another may include SAFE conversions but exclude a proposed pool increase. I ask for a line-by-line definition of the denominator every time. Two tables can both be labelled fully diluted and disagree by several points.

Fixing the post-money SAFEs: division, not modeling

Under the standard post-money SAFE, conversion is a division. Ownership sold equals the investment divided by the post-money valuation cap, and that percentage is fixed the moment the SAFE is signed. The valuation cap is the highest valuation at which a SAFE converts; if the priced round values the company above the cap, the SAFE converts at the cap instead.

So a $500k check on a $5M post-money cap is exactly 10%. YC's own published math runs the same way: targeting a $1M raise at 15% ownership implies a post-money cap of $1M divided by 15%, roughly $6.7M. Raise less on that cap and you sell less: $500k is about 7.5%, $800k is about 12%.

The wrong turn here is reading a pre-money cap as post-money. It is the single most common error and it inflates your sense of ownership.

The second surprise about post-money SAFEs is what happens when several stack up. Under the 2018 post-money form, stacked SAFEs do not dilute each other. Each is locked against the post-money valuation it sets, and the entire compounding hit falls on common stock, meaning founders, until the priced round. The post-money SAFE gives its holder anti-dilution protection against other SAFEs and notes.

The consequence is arithmetic that looks impossible until you accept it. In a documented stack, $2.5M spread across three post-money SAFEs converted to about 22.9% of the pre-Series-A fully diluted table, and about 19.1% after the round closed. Each investor "felt" they bought a smaller slice; the total is larger because founders, not the other SAFE holders, absorbed the overlap.

For the worked deal I lock each post-money SAFE to its fixed percentage first. That set of numbers will not move as I add the note and the pool. They are the fixed points I build everything else around.

Why a note of the same size buys more shares

A convertible note is not a SAFE with a maturity date bolted on. It accrues interest, and that interest converts into equity, so a note and a SAFE of identical face value produce different share counts.

SAFEs carry zero interest and have no maturity date; they sit outstanding until a triggering event. Convertible notes typically accrue four to eight percent annually, most often five or six, and mature in 18 to 36 months. At conversion the accrued interest is added to principal, and the larger balance converts.

Here is the interest effect on real numbers.

Over 12 to 24 months, interest adds roughly 6 to 10% to the converting dollars, and therefore to the shares. If I model a note at its face check I understate the noteholder's shares and, because those extra shares dilute me too, I understate my own dilution. So for the worked deal I compute accrued interest to the expected conversion date, add it to principal, and only then apply the note's cap or discount.

The maturity date is a separate flag worth logging. A note that matures before the round closes forces a decision: repay, convert, or extend. A SAFE never does. That timing risk does not change my share math today, but it changes whether the note is even in the stack when the round prices.

A note is a SAFE that quietly grows while you are not looking, and its growth dilutes you too.

The cap-versus-discount test, done in advance

When a single instrument carries both a valuation cap and a discount, only one is used: whichever produces the lower price per share, and therefore more shares. The two never add together.

I compute both prices at conversion and keep the lower. In a documented conversion the cap price came to $0.72727 and the discount price to $0.77265, so the holder converted at the cap and received 137,500 shares. The discount was decorative.

Better still, the crossover is knowable before the round even prices. For a $5M cap with a 20% discount, the flip sits near a $6.25M pre-money valuation. Below that the discount governs; above it the cap governs. So if I expect a strong up-round, I already know the cap is load-bearing and the discount is window dressing. Modeling both is the only way to tag which term actually drives the conversion.

One note on where these combined instruments come from: YC's current standard templates do not include a combined cap-and-discount SAFE, on the stated grounds that it did not encounter situations where the combined form was preferred. So if a term sheet in front of you carries both, it is off the standard menu, and worth reading closely.

The option pool shuffle: the quiet killer

The pool refresh, not the new money, is often what shrinks your ownership most. When a term sheet dictates that a pool be created or expanded on a strict pre-money basis, 100% of that dilution is absorbed by founders and existing shareholders, and the expanded pre-money denominator hits SAFE holders converting against it too.

The shuffle works like this: you pick a target post-money pool percentage, then create the shares pre-money, so existing holders absorb it before the new money arrives. That drops the price per share, which quietly hands more of the company to the new investor at no change in the headline valuation.

Here is the effect on one worked case: 10M shares, a $10M pre-money, a $5M check.

A 20-point pre-money pool cut the price from $1.00 to $0.80 and moved about 6 points of the company to the new investor, who paid no more for it. The pool is sized to 12 to 24 months of hiring and typically runs 10 to 20% of fully diluted shares at close: pre-seed and seed 10 to 15%, Series A tops up to 15 to 20%, later stages back to 10 to 15%.

Note the funnel is illustrating the new investor's gain; for the SAFE holder already in the stack, the same pool expansion runs the other way, shrinking their pre-money percentage. That is the point most models miss: the pool does not just tax founders, it taxes every earlier convertible.

Running the whole sequence, end to end

Here is the procedure I run on the worked deal, in order. Each step produces a defined output so you can check your work at every fork.

There is a genuine order-of-operations dispute at step six, and it changes the answer materially. The standard market default creates the pool pre-money, so founders and existing holders absorb it. But some models close the round first at the headline split, say 80% to existing holders and 20% to the new investor, then add the pool afterward and dilute everyone proportionally. That is a different result, and it favours founders. Do not assume which one applies; confirm what the term sheet specifies.

Where this goes wrong: the failure modes

Most bad SAFE math is not a hard error; it is a skipped step that looks complete. Here are the specific ways the model lies, and the check for each.

• Reading pre-money as post-money. A $1M at $4M deal looks like 25%. If the cap is pre-money it is 20%, one divided by five. Confirm the exact word on the cap.
• Ignoring the pool refresh. Dividing the check by the post-money cap and stopping. The pool is carved pre-money, so recompute price per share on the expanded denominator before you read your percentage.
• Missing an accruing note. Modeling a note at its face check. Add accrued interest to the conversion date; a SAFE-shaped assumption understates a note's shares by 6 to 10%.
• Treating cap and discount as additive. Reading "cap AND 20% off" as both. Compute both prices and keep only the lower.
• Reading the wrong cap table view. Quoting an outstanding-basis percentage. Demand a line-by-line definition of the fully diluted denominator, since two tables labelled the same can disagree.
• Assuming stacked SAFEs share dilution. Expecting later SAFEs to dilute earlier ones. Each is fixed; founders absorb the overlap.
• Order-of-operations drift. Adding the pool post-close proportionally when the term sheet says pre-money. Confirm the timing.

One more that sits outside the share math but changes what your ownership is worth. A 10% fully diluted stake does not automatically receive 10% of a sale. Liquidation preferences, participation, seniority, dividends, and outstanding debt can all reshape the payout. Fully diluted percentage tells you dilution, not proceeds. When the exit matters, model the waterfall separately.

Before you call the model done

Run this list against your sheet before you commit to a number or a wire. It is the difference between a headline percentage and a defensible one.

Keep the model live. Log every SAFE the day it is signed, with amount, cap, discount, MFN, and pro rata, and update your conversion model after each signature rather than reconstructing it all at Series A. The stack only compounds; a model that lags the last three SAFEs will misstate your dilution by more than the last one alone.

Instrument | Amount | Type (post-money SAFE / note) | Cap | Discount | MFN | Pro rata | Interest rate | Issue date | Maturity | Cap or discount driven | Locked %

Finally, a note on who does this work. In Refolk's index only 64 US professionals and 11 UK professionals list Cap Table Management as a skill, a roughly 5.8x US-to-UK gap, and the cohort clusters in the San Francisco Bay Area. Financial modeling tied to a venture or angel context is thinner still, around 19 US profiles, though that figure comes from a keyword filter and is directional rather than a like-for-like count.

The practical reading is that the "someone who checks the math" role is scarce, so the investor frequently has to be that person. When you do want a second set of eyes, Refolk can name the finance leads and fund analysts who have modeled these conversions before, so you are handing the sheet to someone who has already hit the forks in this guide. But the arithmetic here is division, interest, and a pre-money pool carve; once you have run it on one deal end to end, the term sheet in front of you stops being a black box.