Zandvoort FIA F4 Case Study: 0.61s Faster Repeatable Pace
A personal best is useful, but it is also one lap. For this Zandvoort FIA F4 program, we wanted to answer a harder question: did the driver actually become faster in a way that showed up across normal laps and then carried into race running?
The dataset says yes. Across 55 valid archived laps recorded by Valor between August 20 and August 24, 2026, the driver moved from an early representative median of 1:36.250 to a late representative median of 1:35.639. That is a 0.611-second improvement in repeatable pace.
Over the same representative sample, Valor's modeled turn loss fell from 2.182 seconds to 1.799 seconds, a 17.5% reduction. The best recorded lap reached 1:34.810, 1.566 seconds faster than the first archived lap, but the stronger result is that the improvement was not confined to that single lap. The gain also showed up in race conditions.
The program at a glance
Car: FIA F4
Track: Zandvoort Grand Prix layout
Analysis window: August 20–24, 2026
Valid archived laps: 55
Offline Testing: 6 laps | best 1:36.119 | median 1:36.250 | average modeled turn loss 2.259s
Practice: 6 laps | best 1:34.810 | median 1:35.767 | average modeled turn loss 1.739s
Lone Qualify: 4 laps | best 1:35.835 | median 1:35.960 | average modeled turn loss 2.001s
Race: 39 laps | best 1:35.081 | median 1:35.917 | average modeled turn loss 1.967s
First archived lap: 1:36.376
Best archived lap: 1:34.810
Peak first-lap-to-best improvement: 1.566 seconds
Representative pace gain, first 10 vs. last 10 normal laps: 0.611 seconds
Representative modeled turn-loss reduction: 17.5%
First-lap-to-best modeled turn-loss reduction: 59%
For the representative first-10-versus-last-10 comparison, obvious nonrepresentative laps above 1:40 were excluded. Those laps can still be valid telemetry records, but traffic, mistakes, race interruptions, or other delays make them poor evidence of underlying pace.
Why we did not use the personal best as the main result
The 1:34.810 lap is the fastest lap in the dataset, and it matters. It proves the driver could put together a substantially faster lap than at the start of the program. But a PB by itself can hide a lot. A driver can gain a second because one lap happened to connect three unusually good corners. The next five laps may still be at the old pace. That is useful for identifying potential, but it is not the same as driver development.
So the primary progression measure here is representative pace. The first 10 normal laps produced a median of 1:36.250. The final 10 normal laps produced a median of 1:35.639. That is 0.611 seconds of repeatable improvement.
The modeled turn-loss number moved at the same time: 2.182s to 1.799s. This matters because it gives us a second signal underneath the stopwatch. The driver was not only circulating faster; the corner-by-corner loss profile was becoming smaller.
Session-by-session development
Offline testing: establishing the starting point
The first six archived laps came from offline testing. The best was 1:36.119, the median was 1:36.250, and average modeled turn loss was 2.259s. The pace was already fairly tight: all six laps were between 1:36.119 and 1:36.505. That is useful because the later improvement is not simply the result of comparing a messy first lap with a polished final lap. There was a clear early plateau around the low 1:36s.
Practice: the ceiling moves
Across six archived practice laps, the best reached 1:34.810, the median was 1:35.767, and average modeled turn loss was 1.739s. The headline is the 1:34.810, but the median matters just as much. Practice pace had moved well below the original 1:36.250 testing median. Average modeled turn loss also fell by roughly half a second compared with offline testing: 2.259s to 1.739s.
Lone qualifying: pace remains in the 1:35s
Across four archived lone-qualifying laps, the best was 1:35.835, the median 1:35.960, and average modeled turn loss 2.001s. Qualifying did not reproduce the 1:34.810 practice peak, but it stayed meaningfully faster than the original testing median. That is another reason not to build the story around the PB alone.
Race: development survives contact with race conditions
There are 39 archived race laps in the dataset. The best race lap is 1:35.081, the overall race median is 1:35.917, and average modeled race turn loss is 1.967s. Thirty-three of those race laps were below 1:40. The remaining valid laps include slower race-running records that are useful telemetry but not representative of clean underlying pace.
The latest larger race block was particularly useful: 24 archived laps, 19 normal sub-1:40 laps, a best of 1:35.081, a median of 1:35.757, and average modeled turn loss of 1.858s. That is the transfer signal we were looking for. The driver did not simply learn how to produce one fast practice lap. A meaningful portion of the gain remained present when the session type changed to racing.
First lap versus best lap: what changed underneath the stopwatch
The first archived lap was 1:36.376 with 2.400 seconds of modeled turn loss. The best lap was 1:34.810 with 0.984 seconds of modeled turn loss. That is a 1.416-second reduction in modeled corner loss, or approximately 59%.
On the best lap, 12 of the 14 mapped corners showed less modeled loss than on the first archived lap. One corner was unchanged and one was slightly worse. That distribution is important: the faster lap was built from improvements across the circuit rather than one isolated hero corner.
Turn 1
Modeled loss moved from 0.301s to 0.184s, an improvement of 0.117s. Entry speed was nearly unchanged at roughly 140.1 mph versus 140.6 mph. Minimum speed moved from 61.6 to 61.9 mph, while exit speed improved from 82.3 to 83.2 mph. This is a good example of why raw minimum speed does not tell the entire story: the corner improved by more than a tenth without a dramatic change in the headline speed numbers.
Turn 2
Modeled loss fell from 0.076s to 0.030s, removing 0.046s. The corner was never a major loss center, but reducing small losses matters when they appear repeatedly across a lap.
Turn 3
Modeled loss went from 0.101s to 0.000s on the best lap. Minimum speed increased from approximately 69.9 mph to 70.4 mph, exit speed from 85.6 mph to 86.1 mph, and peak brake input fell from roughly 45.2% to 41.0%. It became effectively neutral against the modeled reference on the best lap.
Turns 4, 5 and 6
These three corners were already relatively strong early in the program, but all three improved further. Turn 4 moved from 0.055s to 0.021s, Turn 5 from 0.050s to 0.010s, and Turn 6 from 0.037s to 0.018s. Together they removed another 0.093 seconds of modeled loss.
The biggest gain: Turn 8
Turn 8 produced the largest first-lap-to-best-lap improvement in the entire dataset. Modeled loss moved from 0.310s to 0.022s. That is 0.288 seconds recovered in one corner.
Entry speed: approximately 114.8 mph to 118.9 mph
Minimum speed: approximately 90.8 mph to 100.8 mph
Exit speed: approximately 103.4 mph to 105.7 mph
Peak brake: approximately 56.5% to 34.6%
The minimum-speed increase is roughly 10 mph, while peak brake demand fell by more than 20 percentage points. That is a very different corner execution. The faster version carried more speed into and through the corner while requiring less peak brake input. This is exactly the kind of change a lap-time delta alone cannot explain.
Turn 7: another quarter second nearly disappears
Turn 7 went from 0.257s of modeled loss to 0.009s, a 0.248-second improvement. Entry speed changed only slightly, from roughly 128.5 to 129.1 mph, but minimum speed jumped from about 99.0 to 108.2 mph and exit speed from 109.3 to 113.1 mph. Peak brake fell from roughly 41.5% to 26.8%. Again, the major difference is not simply braking later. The best lap carried substantially more minimum speed and reached a stronger exit while using a lower peak brake input.
Turn 12: a difficult loss center becomes smaller
Turn 12 began as the single largest loss on the first archived lap at 0.415 seconds. On the 1:34.810 lap it was down to 0.162 seconds, a gain of 0.253 seconds. Minimum speed rose from approximately 55.4 mph to 58.7 mph. Exit speed stayed almost identical at roughly 84 mph, which suggests the gain was created primarily before the exit rather than by a dramatic increase in terminal exit speed. Turn 12 remained an opportunity even on the PB, but it stopped being the four-tenths problem it was at the beginning.
Turn 13: more speed with less brake
Turn 13 improved from 0.302s to 0.146s, recovering 0.156 seconds. Entry speed increased from roughly 102.7 to 103.5 mph, minimum speed from 82.4 to 87.4 mph, exit speed from 94.8 to 95.9 mph, and peak brake dropped from about 46.7% to 29.9%. The combination is familiar by this point: similar entry speed, higher minimum speed, slightly better exit, and a significantly lower peak brake demand.
The full corner-loss map
Turn 1: 0.301s → 0.184s | gain 0.117s
Turn 2: 0.076s → 0.030s | gain 0.046s
Turn 3: 0.101s → 0.000s | gain 0.101s
Turn 4: 0.055s → 0.021s | gain 0.034s
Turn 5: 0.050s → 0.010s | gain 0.040s
Turn 6: 0.037s → 0.018s | gain 0.019s
Turn 7: 0.257s → 0.009s | gain 0.248s
Turn 8: 0.310s → 0.022s | gain 0.288s
Turn 9: 0.214s → 0.136s | gain 0.078s
Turn 10: 0.147s → 0.174s | loss increased 0.027s
Turn 11: 0.000s → 0.000s | unchanged
Turn 12: 0.415s → 0.162s | gain 0.253s
Turn 13: 0.302s → 0.146s | gain 0.156s
Turn 14: 0.135s → 0.072s | gain 0.063s
The math reconciles with the overall modeled-loss change: 2.400 seconds on the first lap versus 0.984 seconds on the best, a 1.416-second reduction.
What the best lap still says about remaining opportunity
A 1:34.810 is the fastest lap in this program, but Valor still modeled 0.984 seconds of turn loss on it. Turn 1 still carried 0.184s, Turn 9 0.136s, Turn 10 0.174s, Turn 12 0.162s, and Turn 13 0.146s. Those five corners alone account for roughly 0.802 seconds of the modeled loss on the PB. In other words, the lap was a major improvement and still contained a clearly defined next development path.
Peak pace versus race pace
The difference between the best practice lap and best race lap was only 0.271 seconds: 1:34.810 versus 1:35.081. The race-best lap carried 1.334 seconds of modeled turn loss, compared with 0.984 seconds on the practice PB. It did not perfectly reproduce the best practice execution, but the stopwatch remained close. The driver was operating much nearer the developed pace ceiling than at the beginning of the program.
What we can and cannot claim
This is an observational case study, not a controlled laboratory test. The data was recorded during real sim-racing sessions across several days. We have not normalized every lap for fuel load, tire state, setup changes, track temperature, track usage, weather, traffic, tow, or race circumstances. Those factors can influence lap time and individual corner behavior. For that reason, we do not claim that Valor alone caused every tenth of improvement.
What the dataset does support is more specific: the driver's representative pace became faster; the modeled turn-loss profile improved at the same time; the best lap showed major corner-level changes in several independent parts of the circuit; the improvement was distributed across the lap rather than concentrated in one corner; a meaningful portion of the gain transferred into qualifying and race sessions; and the final data still identified specific areas where additional pace remained available.
Why this matters for telemetry coaching
A conventional lap delta tells a driver where the clock moved. That is valuable, but it does not automatically tell the driver why. At Turn 8, the difference was not simply “brake later.” The faster execution combined a higher entry speed, about 10 mph more minimum speed, a stronger exit, and materially less peak brake pressure. At Turn 12, the exit speed barely changed, but the modeled loss still fell by more than a quarter second. At Turn 1, the headline speed values changed only slightly, yet more than a tenth was recovered.
These are different driving problems. Treating all of them as the same generic instruction would throw away the useful part of the telemetry. Valor's goal is to connect lap time to the sequence underneath it: braking, release, rotation, minimum speed, throttle commitment, exit behavior, and the driver's relationship to a reference.
The result
The cleanest summary is not the PB. It is this: 55 valid archived laps. Representative pace improved by 0.611 seconds. Modeled turn loss fell 17.5%. The first-lap-to-best loss profile fell 59%. And the faster pace showed up in race running. That is a much stronger development signal than one isolated fast lap.
Want to see how Valor uses baselines and corner-level telemetry? Explore the Valor Baseline Library at https://www.valorsimracing.com/baselines or get Valor at https://www.valorsimracing.com/downloadvalor.
Comments