How to price a flip with a cash floor, not a hunch

2026-09-11 — pricing method for used-car flippers, about 1,100 words.

A flip goes wrong in two directions. The first is famous: you overpay for a car and can't retail it. The second is quieter: you underpay for a car that turns out to have a title issue, a hidden mechanical problem, or a market you misread, and now you're stuck sitting on inventory that ties up your capital. Both problems are pricing problems in disguise.

Most flippers price a deal off one number: comps. They pull a KBB Fair Purchase Price, look at three or four active listings on Autotrader and CarGurus, and decide whether the ask is "under-market." That's a ceiling number — the top of what a retail buyer might pay. It tells you the upside if everything goes right. It tells you nothing about the downside if the car turns out to be worse than described.

A better method uses three numbers on every candidate: ask, ceiling, and floor. The floor is the one nobody teaches, and the one that keeps you out of trouble.

The three numbers

Number one: the ask

Straight from the listing. Autotrader by-owner. Facebook Marketplace. Craigslist. Whatever the seller is asking today. This is the input to everything else. Note the date — if the ask has been sitting for 60 days, the seller is negotiable in a way they weren't on day one.

Number two: the retail ceiling

What a retail buyer would pay if you did the work: cleaned the car, took real photos, waited two-to-four weeks for the right buyer, and haggled through the offer. KBB Fair Purchase Price is the standard proxy. It's a fair proxy but it's a retail number, not a private-party number. If your exit plan is a private Craigslist buyer rather than a dealership, KBB's private-party value is closer to your real ceiling and typically runs 8–15% below FPP for a comparable mileage and condition.

The ceiling is a soft number. It depends on how long you're willing to hold, how good your photography is, and how patient the buyer market is that month. Two flippers looking at the same car will have different ceilings.

Number three: the cash floor

The number an instant-offer buyer will pay you tomorrow, with a tow truck, no negotiation, and no waiting for a retail buyer. In the FlipScout stack that's an Algo (Auto Lenders) instant offer: submit the VIN, get a real range within the hour, drive the car in with clean title and take the check. The floor is a hard number. It's what you actually get if the deal falls apart, if the car needs a repair you didn't budget for, if you get bored, if the market shifts.

Not every car has a real cash floor. Pre-2005 vehicles, exotics, and some unsupported ZIPs come back with no estimate — that's a real "no" from the buyer, not a system error. When there's no floor, the deal is only as safe as your read on the retail market. That is a rougher place to be.

The safety inequality

The math the floor lets you write is one inequality:

ask + fees ≤ floor.high  →  deal is arithmetically safe
ask + fees ≤ floor.low   →  deal is safe on the low end too

In Texas, "fees" is roughly 6.25% × ask + $40 title + $125 Algo handling fee. In Illinois it's the model-year use-tax table plus $165. In your state it's whatever the DMV and the buyer charge. Bake it in — a $300 rounding error kills a $1,000 flip.

A deal that clears the low-side inequality is a deal you can walk away from with cash in hand even if the car is exactly as bad as described. A deal that clears only the high-side is one where you're betting the estimator lands closer to the top of the range — usually fine, sometimes not. A deal that clears neither is not a flip; it's a bet on the retail market moving in your favour before your capital runs out.

Worked example

2013 Honda CR-V EX-L, 168k miles, listed by owner in Austin at $6,286. KBB Fair Purchase Price for this trim + mileage: $8,415. So the ask is 25% under FPP, which reads as a good deal.

Fees in Texas at that ask: 6.25% × $6,286 + $40 + $125 = ~$558. So the all-in cost is $6,844.

Ceiling: KBB private-party value for a 168k-mile CR-V in Austin, excellent condition, runs about $7,300–$8,100 in this segment. That gives you $500–$1,300 of retail headroom, minus your detailing time, listing time, and the two or three tire-kickers who don't show. Thin.

Floor: this is where you run the instant-offer request. Suppose Algo comes back with $6,500–$7,100. That means:

• Low-side check: $6,844 > $6,500 → fails. On the low end you'd lose ~$344.
• High-side check: $6,844 < $7,100 → passes. On the high end you'd clear ~$256 without touching the retail market.

Now the deal is a real decision, not a hope. If you can talk the seller to $5,900 you clear the low-side inequality by ~$150 and the ceiling turns into pure upside. If they hold at $6,286, you are gambling on the estimator and on your retail patience for a ~$500 spread. That is the wrong risk-to-reward for a used CR-V.

Why the floor changes behaviour

Two things change once you're pricing off three numbers instead of one. First, you stop chasing deals that look great on KBB alone but have no exit. Second, you start negotiating differently: you know your walk-away number cold, and you can say it to the seller with numbers ("Algo will pay me $6,500 tomorrow, I need $200 of margin, I can do $5,900 today") instead of arguing about KBB.

The trap the ceiling-only method sets is that both mistakes look the same going in. A car that's 25% under KBB with a hidden problem and a car that's 25% under KBB because the seller is moving to Alaska are the same row in your spreadsheet. The floor is what separates them: the first car's Algo estimate will be disappointing, the second car's will be surprisingly close to KBB. That signal is worth the ~40 seconds it takes to get a quote back.

What FlipScout does with this

FlipScout builds the three-number picture automatically. Every scanned listing carries the ask (Autotrader), the ceiling proxy (KBB FPP on the search row plus, on demand, a KBB private-party pull), and — when you explicitly ask for it with flipscout quote --yes — the Algo floor range. The floorSpread field in the report is exactly the inequality above, computed for you per row. Positive = safe; negative = walk. See the docs.

None of this is complicated math. The trick is having all three numbers on every candidate, cheaply, before you drive across town to look at the car. That's the deal FlipScout tries to make routine.