Simulation
The Nordschleife, and how wrong we are about it
We built a lap simulation of the 20.832 km Nürburgring Nordschleife and ran the fleet around it. Before any of those results are worth reading, here is how far off the model is.
Calibration
Four real cars with published laps, run through the identical engine. A negative error means the model was optimistic — it ran the car quicker than reality.
| Reference car | Simulated | Real | Error |
|---|---|---|---|
| Porsche 911 GT3 RS (Manthey MR)A dry, downforce-bearing performance car in the same envelope as this fleet. This is the honest anchor. | 6:51.3 | 6:49.3 | +2.0 s |
| Mercedes-AMG ONEExtreme-downforce prototype; the model over-estimates aero it cannot see. | 6:18.9 | 6:29.1 | −10.2 s |
| Ford GT Mk IVFastest internal-combustion lap on record. Also outside the calibrated envelope. | 6:02.3 | 6:16.0 | −13.7 s |
| Rimac NeveraHeavy EV whose real lap is limited by weight and energy management a point-mass model cannot represent. | 6:47.8 | 7:05.3 | −17.5 s |
The GT3 RS lands within two seconds on fully-known specifications. That is the anchor, and it is the one reference car in the same envelope as this fleet — dry, downforce-bearing, road-derived.
The other three read optimistic, and the reasons are structural rather than incidental. Two are extreme-downforce prototypes whose aerodynamics the model over-estimates. The fourth is a heavy EV whose real lap is governed by energy management that a point-mass solve cannot represent.
The consequence, which applies to every number below: the model runs a few percent quick for the very fastest cars. Our top times are best-case ceilings, and the true gap to a record car is larger than a raw comparison suggests.
The fleet, on track tyres
Cup 2R track tyres, otherwise standard. Vehicle inputs are the programme's own physics configuration — the same numbers behind the specifications page.
| Vehicle | Lap | To leader |
|---|---|---|
| Cyclotis | 6:27.0 | — |
| Vindicator | 6:29.9 | +2.9 s |
| 427 Turbo | 6:30.4 | +3.4 s |
| Foundry 427 | 6:35.5 | +8.5 s |
| Soray | 7:08.1 | +41.1 s |
| GT-1 | 7:12.9 | +45.9 s |
| Ember RS | 7:43.9 | +76.9 s |
| Velma | 8:06.1 | +99.1 s |
| Summit | 8:24.8 | +117.8 s |
| Ironside | 8:49.2 | +142.2 s |
Grip beats power
This is the result that changed how the programme thinks about the fleet. A tyre-compound step is worth 13 to 19 seconds a lap. Two hundred additional horsepower is worth about one.
These cars are already over-powered for the grip they can put down. Everything meaningful left on the table is rubber, wings and lightness — not the dyno. It is not the conclusion anyone wanted, and it is the one the model keeps returning.
Track spec
Race slicks, a genuine downforce wing, and a carbon diet — with no change to engine output. That constraint is the experiment, not a footnote. Compared against the Ford GT Mk IV run through the same simulator at 6:02.3, which is the only comparison that controls for the model's bias.
| Vehicle | Lap | To leader | vs Ford (sim) |
|---|---|---|---|
| Cyclotis ROn the locked 2,000 hp marketing figure. On the 1,200 hp physics config: 5:54.2. | 5:41.0 | — | −21.3 s |
| Vindicator R | 5:47.2 | +6.2 s | −15.1 s |
| Foundry 427 R | 5:51.8 | +10.8 s | −10.5 s |
| 427 Turbo R | 5:52.4 | +11.4 s | −9.9 s |
| Soray R | 6:31.5 | +50.5 s | +29.2 s |
| GT-1 R | 6:36.8 | +55.8 s | +34.5 s |
For scale, the Ford's real recorded lap is 6:16.0 — the fastest internal-combustion lap on record. We are not claiming a Schubert car has beaten it. No Schubert car has turned a wheel on that circuit, or on any circuit. What the table shows is how four concepts place against one reference car inside one simulation, with that simulation's error published above.
The interesting result is not the order at the top. It is that a $225,000 road-derived coupe runs with cars costing ten times as much, because it is light and it carries the highest real downforce in the fleet.
What we got wrong the first time
The first version of this study used class-precedent estimates for power, mass and downforce, on the belief that no consolidated specification data existed. It did. Re-running everything on the real configuration moved numbers substantially: Cyclotis from an assumed 2,000 hp to 1,200 in the physics config, the 427 Turbo from 1,400 hp to 850, Ironside from 2,600 kg to 3,580. Downforce fell across the board.
The first version also reported that three cars beat the Ford on track tyres. On real data that result disappeared entirely. It was an artefact of inflated inputs.
An earlier model version was worse still: it let a tyre spend all its grip cornering and all of it braking simultaneously, ignored the circuit's 300 m of elevation, and had no gearing limit, which produced top speeds near 500 km/h. Those are fixed. They are recorded here because a simulation that has never been wrong in public is a simulation nobody has checked.
Method: Forward/backward quasi-steady-state point-mass solve, v2, 2026-07-04. Source: NORDSCHLEIFE_DOSSIER, grounded on VehicleConfigs.h. All Schubert times are simulated. No Schubert vehicle has lapped the Nordschleife or any other circuit. Reference-car real times are the published figures for those vehicles. Every vehicle on this site is concept-stage; none are in production, and nothing here is for sale.