How Red Bull found its straight-line speed in 2026, and why Mercedes lost it

Jatin Garg

Independent analysis of public Formula 1 timing data

10 October 2026

Main findings

Red Bull closed most of its pace deficit. Its qualifying gap to pole fell from — in rounds 1–5 (95% CI — to —) to — in rounds 12–17 (— to —). Corrected race pace went from — to — off the best.
Straights were about half of the qualifying gain. Of the — per lap closed on Mercedes, — came on full-throttle sections. On the longest straight in races, Red Bull is now level with Mercedes (— per pass; CI — to —), not ahead.
The gain came at corner exit, not top speed. Red Bull now starts the longest straight — relative to Mercedes (it was —), and ends it at —. Its qualifying trap speed changed by — relative to the field (CI — to —).
The energy-deployment explanation is not confirmed. Once entry speed is controlled for, Red Bull gains less speed along straights than Mercedes. Public data carry no battery channel, so the hypothesis cannot be ruled out, only left unconfirmed.
Mercedes' race straight-line edge faded, but the compression-ratio test shows no immediate effect. Mercedes changed by — per pass relative to the field (CI — to —). In the six races after the test, Mercedes-powered cars moved — in race top speed against Ferrari-powered cars (— confidence of any loss). The larger drop from Monza onward coincides with Ferrari's second engine upgrade and also hits Red Bull-Ford engines. A strongest-case analysis (— specifications) found slower acceleration at 290–340 km/h across all four Mercedes teams, but nothing that survives correction for multiple testing.
Tyre wear cannot separate the teams with this much noise.

Abstract

The 2026 Formula 1 rules made roughly half of a car's peak power electric and placed it under a per-lap energy budget and a speed-dependent limit. I study how Red Bull Racing closed its early-season deficit using — sessions, — timed laps and — passes along full-throttle straights. Laps are aligned by projecting car positions onto a reference racing line and reconciled to official lap times. Race pace is corrected for fuel, tyre age, compound and dirty air. Every comparison carries a 95% bootstrap interval over rounds. Red Bull's qualifying gap to pole fell from — to —, and about half of its gain on Mercedes came on straights. The straight-line gain is a corner-exit gain: Red Bull's speed-trap reading fell relative to the field, and it accelerates less than Mercedes once on a straight. The hypothesis that Red Bull won time by spending its electric energy better is consistent with some of the data but not supported by the entry-controlled test. Mercedes' straight-line advantage in races faded. A difference-in-differences test against Ferrari-powered cars finds no immediate effect of the 1 June compression-ratio test.

Keywords: Formula 1, telemetry, energy management, power units, difference-in-differences, FastF1

Introduction

In March Red Bull qualified more than a second off pole at Suzuka. By October it had taken consecutive poles at Sepang and Singapore. The paddock explanation is that Red Bull "found straight-line speed" and is now quicker than Mercedes on the straights. A second claim is that the FIA's compression-ratio test of 1 June took away a Mercedes engine advantage. Both are hard to test from speed traps: in 2026 a car can lose fifty kilometres per hour on a straight while its driver keeps the throttle pressed (Section 4.3).

This note asks four questions. Where on the lap did Red Bull find its time? How much of its straight-line gain is its own, and how much is Mercedes slowing down? What does the shape of the gain say about its cause? And did the compression-ratio test leave a measurable mark?

The 2026 power unit

The terms below carry the argument. All rule values are from the FIA 2026 Technical Regulations [1] unless stated.

ICE: The 1.6-litre V6 combustion engine. After the first review period the FIA rated Red Bull-Ford's as the best on the grid [4].
MGU-K: The electric motor-generator. It can deliver up to — to the wheels, or absorb the same to charge the battery.
Energy budget: The battery's state of charge may swing by — on track. Recovery is capped at — per lap, and lower at some events: Monza qualifying used —.
Overtake mode: Extends full electric power to higher speed. It is on all lap in practice and qualifying, and in races only within the detection gap to the car ahead [2].
Super-clipping: Harvesting energy while the driver stays at full throttle, so the car decelerates with the pedal pressed. It was always allowed; from Miami the cap rose from — to — [3].
Straight Mode: Active aerodynamics that flatten both wings inside FIA-defined zones. Its state is not in the public data.
ADUO: Extra engine homologations for manufacturers whose ICE trails the best. Red Bull received none. Audi, Ferrari and Honda raced upgrades from Barcelona, Austria and Zandvoort; Ferrari added a second at Monza [4].
Compression-ratio test: From 1 June the 16.0:1 limit must also hold at 130 °C. It was aimed at a reported Mercedes design that raised the ratio when hot; Red Bull-Ford was early on reported to use a similar idea [11][12].

Electric power is limited by speed (Article C5.2.8). In normal race mode, with $v$ in km/h and power in kW,

P_{K}^{max} (v) = {\begin{matrix} 350, & v < 290 \\ 1800 - 5 v, & 290 \leq v < 340 \\ 6900 - 20 v, & 340 \leq v < 345 \\ 0, & v \geq 345 \end{matrix}

(1)

With Overtake on, the limit is $P_{K}^{OT} (v) = min (350, max (0, 7100 - 20 v))$ , which permits full power to —. Figure 2 plots both. These are maxima: whether a car uses them depends on its energy budget. From Miami, races also apply FIA-defined sectors capped at 250 kW below 310 km/h, and Monaco and Singapore use a lower curve. The consequence for this study is that qualifying and races test different things. In qualifying the battery may supply full power almost to the end of a straight. In a race the electric contribution fades from — and is gone by —, so the ICE carries the top end.

Figure 1. How the rules shape one straight (schematic). Electric power is limited at corner exit only by the 350 kW cap (i), fades with speed (ii), may be harvested while the driver stays flat out (iii), and braking recovers more (iv). Which phase a team gains in is a hypothesis tested in Section 4.5.

Figure 2. Maximum MGU-K power against speed, from Eq. (1) and the Overtake variant.

Data and methods

Data

The data are every 2026 session available on 10 October, from Australia to Singapore qualifying, downloaded from Formula 1's live-timing archive with FastF1 [6]. They cover lap timing, car telemetry (speed, RPM, gear, throttle, brake) at about 4 Hz, car positions at about 4 Hz, weather and race-control messages. The 2026 Bahrain Grand Prix was held at Sepang and is treated as Sepang throughout. The public feed carries no battery, Overtake or wing-state channel, and the legacy DRS field is zero for all of 2026.

Exclusions. Monaco is excluded from every straight-line and race measure: several cars had power-unit failures, it has no Straight-Mode zones, and its race position feed stops halfway. Hungary's race position data are degraded, so it is excluded from race straight-line measures. Only laps on slick tyres are used, and qualifying uses each team's fastest non-deleted lap with valid telemetry.

Alignment on the racing line

For each circuit I build a reference line $𝐫 (s)$ from the position samples of a clean pole lap, densified to 1 m. Every position sample $𝐩_{i}$ is mapped to the track coordinate

s_{i} = \underset{s \in [0, L)}{\arg \min} ‖ 𝐩_{i} - 𝐫 (s) ‖

(2)

and car-data samples inherit $s$ by interpolation in time. Each driver's clock offset between timing and telemetry is estimated from the line crossings it implies (it reached 21 s in the rain-delayed Sepang race). Laps are rejected when they leave the racing line, when their reconstructed duration differs from the official lap time by more than 0.25 s, or when a residual clock offset remains. That removes — of timed qualifying and race laps. The accepted laps have a median duration error of — (99th percentile —), and each is then rescaled so that its time matches the official lap time exactly.

Straights, corners and clean air

A circuit's straights are the full-throttle zones of its reference lap that are at least 200 m long. A zone that runs through the start/finish line is one straight, measured from the end of the previous lap. Straights are timed over their first 90%, which keeps the comparison clear of braking points. Corners are classed by reference apex speed: slow below 120 km/h, medium 120–200 km/h, fast above that. The gap $g$ to the car ahead is the time since any other car crossed the line. A lap is clean when $g > 2$ s, and laps with no measurable gap are left out.

Race-pace model

For each race, green-flag laps on slick tyres, excluding the first lap, in- and out-laps, and laps more than 4% off the median, are fitted by ordinary least squares:

t_{d, ℓ} = α_{d} + γ_{c} + β_{c} a_{d, ℓ} + φ n_{ℓ} + δ_{1} 𝟙 [g < 1] + δ_{2} 𝟙 [1 \leq g \leq 2] + ε_{d, ℓ}

(3)

where $d$ is the driver, $c$ the compound, $a$ tyre age and $n$ the lap number. The lap term $φ$ absorbs fuel burn and track evolution together; only their sum is identified. A team's pace is its better driver's prediction at common conditions. The model is descriptive: it assumes the same compound effects for every team, and the driver term mixes car and driver. In a typical race, $- φ$ was —, and a lap within a second of the car ahead cost $δ_{1}$ = —.

Comparing against the field, with uncertainty

Each team's measure $y_{T}$ is taken relative to the field median $\tilde{y}$ in each round and averaged over rounds 1–5 and rounds 6–17:

Δ_{T} = {⟨ y_{T} - \tilde{y} ⟩}_{R6–17} - {⟨ y_{T} - \tilde{y} ⟩}_{R1–5}

(4)

A negative $Δ_{T}$ means team $T$ became quicker relative to everyone else. The boundary follows the FIA's first engine-review period (rounds 1–5); it was not chosen from the data, and it is not pre-registered. Intervals are 95% percentile bootstraps that resample whole rounds; with five or six rounds per phase they are wide, and many include zero.

Results

Pace through the season

Table 1 summarises the season in three phases. Mercedes took the first ten poles. Red Bull's biggest single step came at Miami: its qualifying gap was — at Suzuka and — at Miami (Figure 3). Red Bull had the best corrected race pace in Austria, Baku and Sepang. The Baku qualifying dip is Mercedes at its most dominant; Red Bull's best lap there was Hadjar's, not Verstappen's, whose engine had a fault.

Table 1. Headline measures by phase, with 95% bootstrap intervals over rounds. Percentages are of lap time; seconds are Red Bull minus Mercedes, positive when Red Bull is slower. Monaco is excluded from the race and straight-line rows.
Measure	R1–5	R6–11	R12–17

Figure 3. (a) Gap of each team's best qualifying lap to pole and (b) race pace from Eq. (3) relative to the fastest team, by round. Lower is better. Dashed lines mark (a) the Miami package and energy-rule change, (b) the compression-ratio test and (c) Red Bull's last major package. SEP is the Bahrain Grand Prix held at Sepang.

Where the lap time came from

Splitting each qualifying lap against Mercedes by segment (Figure 4) shows that straights were Red Bull's largest deficit early and closed the most: — per lap in rounds 1–5 against — in rounds 12–17 (CI — to —). Segment times now reconcile to official lap times: the — closed by the segments compares with — from official times. Racing Bulls, with the same engine and a different chassis, is a weak control. Measured against the field, its qualifying-straight change was —, about — of Red Bull's, and in races it moved the other way (—). The shared engine therefore explains less of Red Bull's straight-line gain than the first version claimed.

Figure 4. Qualifying lap time relative to Mercedes by segment type, phase means with 95% intervals: (a) Red Bull, (b) Racing Bulls. Positive values are time lost.

Losing speed at full throttle

Figure 5 shows why speed traps mislead in 2026. On Albert Park's back straight both cars stay at or above 97% throttle from corner exit to the braking point. Both peak about halfway down and then lose around 50 km/h. With harvest power $P_{h}$ and mass $m$ , the added deceleration is

Δ a \approx \frac{P_{h}}{m v}

(5)

so 350 kW at 300 km/h on an 800 kg car gives about 5 m/s², or 19 km/h per second, the order of the slopes observed. The feed cannot distinguish harvesting from simply reduced deployment, so Figure 5 shows a signature consistent with either.

Figure 5. Speed along Albert Park's back straight, fastest qualifying laps of Russell (pole) and Hadjar (P3). Shaded: throttle at or above 97%.

The straight-line turnaround

In races, in clean air, Red Bull's time over each circuit's longest straight relative to Mercedes averaged — per pass in rounds 1–5 and — in rounds 12–17 (CI — to —; Figure 6). Red Bull reached parity rather than overtaking. Against the field (Figure 7), Red Bull changed by — per pass (CI — to —) and Mercedes by — (CI — to —). Both intervals just include zero. In qualifying, — of — teams lost straight-line time relative to the field after round 5, so the slide of the Mercedes-powered teams there is not unusual.

Figure 6. Median time over the first 90% of each circuit's longest straight on clean-air race laps, relative to Mercedes. Positive is slower. Straights through the start/finish line are measured whole. Monaco and Hungary are omitted.

Figure 7. Change relative to the field, Eq. (4), with 95% intervals, for (a) race time on the longest straight, (b) qualifying time on all straights and (c) the qualifying speed trap. Negative time is a gain; for (c), positive is more top speed.

Speed along the straight

Panel (c) of Figure 7 contradicts the paddock story. Red Bull's trap speed changed by — relative to the field (CI — to —), one of the few changes whose interval excludes zero. In Australia its trap read — against Mercedes, yet its time on the straights was — against Mercedes. Figure 8 shows where the difference sits. With each circuit weighted equally, late-season Red Bull is — faster than Mercedes over the first 40% of the longest straight (CI — to —; slower in — of — races) and — over the last 10%. The time over a straight is

T = \int_{a}^{b} \frac{d s}{v (s)}, δ T \approx - \int_{a}^{b} \frac{δ v (s)}{v {(s)}^{2}} d s

(6)

so a gain of 1 km/h at 200 km/h is worth ${(300 / 200)}^{2} \approx 2.25$ times one at 300 km/h. The early advantage is carried in from the preceding corner: Red Bull now starts the straight — relative to Mercedes, against — early in the season. Panel (b) removes that by measuring speed gained since the start of the straight. On that basis late-season Red Bull is — over the first 40% (CI — to —) and gains less than Mercedes thereafter. A car that enters faster gains less on any power-limited straight, because drag rises with speed, so panel (b) is conservative. But it gives no sign that Red Bull accelerates harder once on the straight. The thesis that Red Bull found its time by deploying electric energy early, and letting a strong ICE carry the top end, predicts the opposite pattern and is not supported here. In qualifying the late-straight difference is — (CI — to —), within noise.

Figure 8. Red Bull minus Mercedes along each circuit's longest straight in 10% bins, each circuit weighted equally. The band is the 95% interval over circuits for late-season races; faint lines are the individual late-season races.

Did the compression-ratio test matter?

If the hot compression-ratio test took power away from Mercedes engines, Mercedes-powered cars should have lost straight-line speed against cars whose engines it did not target, and the loss should be largest late on race straights, where the electric contribution has faded. Ferrari-powered cars (Ferrari, Haas, Cadillac) are the natural control. Red Bull-Ford is estimated separately, because it was reported to use a similar idea. The model has team and round fixed effects:

y_{T, r} = α_{T} + λ_{r} + \sum_{w} (β_{w} M_{T}^{w} + ρ_{w} R_{T}^{w}) + ε_{T, r}

(7)

where $M_{T}^{w}$ and $R_{T}^{w}$ mark Mercedes- and Red Bull-Ford-powered teams in window $w$ : rounds 7–12, right after the test, and rounds 13–17, from Monza. Confidence is the share of 1,000 round-resampled estimates pointing the way a power loss predicts; 50% means no information.

In the six races right after the test, Mercedes-powered cars moved — in race peak speed against Ferrari-powered cars (CI — to —; — confidence of a loss). Late-straight speed gives —, race time on straights — (—), and qualifying straights —. Converted with $P \propto v^{3}$ near top speed, the data rule out an immediate loss larger than about —, so they cannot confirm or exclude the 10–15 hp press estimates. From Monza onward Mercedes-powered cars are at — (CI — to —), but Red Bull-Ford drops too (—, — confidence). That timing matches Ferrari's second engine upgrade at Monza, which made the control faster, not a June change to Mercedes engines. Red Bull-Ford cars got quicker on straights against Ferrari after the test (— in races; — confidence of a loss), so whatever version of the idea Red Bull used, losing it left no visible mark.

The strongest case. To guard against a test designed only to find nothing, a second analysis was set up to build the strongest possible case for an effect and to report every specification it tried. It estimated —. The best supporting pattern is acceleration at matched speed: between 290 and 340 km/h, Mercedes-powered cars' acceleration changed by — relative to Ferrari-powered cars (interval — to —). That is the largest change of all 35 possible four-team groupings, and all four Mercedes teams point the same way. Against it, the loss sits between 290 and 310 km/h, where electric power still contributes, and vanishes above 310 km/h (—), where an ICE-only loss should be clearest. A model with separate steps for Miami, the test, Austria and Monza places the peak-speed drop at Ferrari's Austria upgrade (—; interval — to —) and gives the test step the opposite sign (—). RPM and gear at peak speed do not shift. — of the — specifications pass a nominal 5% randomisation threshold and — have intervals excluding zero; none survives correction for multiple testing. Its verdict is suggestive evidence of a Mercedes-group change and weak evidence that the test caused it, which agrees with the analysis above.

Figure 9. Mercedes-powered and Red Bull-Ford teams minus Ferrari-powered teams by round, relative to Canada (R5). The dashed line is the compression-ratio test; dotted lines are Ferrari's engine upgrades at Austria and Monza.

Table 2. Eq. (7) estimates against Ferrari-powered cars, with 95% intervals and the confidence that each effect points the way a power loss predicts. Times are percent of the round median (positive = slower); speeds are km/h (negative = slower).
Measure	Window	Engine	Estimate	95% interval	Confidence of a loss

Upgrades and rule changes

Red Bull declared — parts in the FIA car-presentation submissions after its launch specification [5] (Figure 10). The Miami package was the largest: floor, sidepods, front wing, engine cover, rear corners and a rotating Straight-Mode rear wing. Press reports also describe about 6 kg of weight saving, which was not part of the FIA declaration. The rotating wing was withdrawn after failures in Austria qualifying and at Silverstone. Spa, run without it, produced Red Bull's worst qualifying trap deficit of the season, —. The wing returned in Hungary. At Sepang Red Bull was — per qualifying lap against Mercedes on the straights. That weekend McLaren's team principal judged Red Bull best in a straight line "by a decent margin" [9], and Mercedes' own new package was "very draggy" [10].

Figure 10. Red Bull parts declared per event (circle area proportional to the count; filled for major packages) and rule or engine changes.

Tyres

Red Bull's median tyre-degradation slope in rounds 12–17 was — against Mercedes' — (Figure 11), but the intervals overlap widely. The 2026 tyres degrade little, and these slopes cannot separate the teams.

Figure 11. Median per-race degradation slope with 95% intervals, in milliseconds of corrected lap time per lap of tyre age.

Discussion

Each explanation below is stated as a hypothesis with what the data predict and what they show.

Red Bull leaves corners faster. Red Bull's straight-line gain is carried in from the preceding corner, and its medium-speed corner losses halved. This fits the Miami and Austria packages and a more driveable car. Supported.
Red Bull-Ford spends its energy better. If so, Red Bull should accelerate harder along straights at matched entry speed. It does not, and its trap speed fell. The data cannot observe deployment directly, so the hypothesis is not supported but not excluded.
Mercedes lost its straight-line edge. Its race straight-line change against the field is the largest of any team, with an interval that just includes zero. Probable; cause unclear.
The compression-ratio test cost Mercedes power. No measurable effect in the races right after it; the later drop matches Ferrari's engine upgrade and hits Red Bull-Ford too; the strongest-case analysis finds a consistent but statistically fragile acceleration loss. Not established; an effect below about 13 hp cannot be excluded.
The rotating rear wing helped on the straights. Spa, without it, was Red Bull's worst trap deficit; a single event is weak evidence. Plausible.

Limitations

Energy use is inferred from speed traces, because the public feed carries no state of charge, Overtake state or wing position. The phases contain five or six rounds on different circuits, so track mix is controlled only through comparisons with the field or a control group in the same round. Qualifying uses each team's best lap, which mixes in driver effects and changes of driver (Lawson replaced Hadjar at three events). The race-pace model is descriptive and assumes common compound effects. The compression-ratio test has one change date and several overlapping confounders: the Miami energy rules, three other manufacturers' engine upgrades and Mercedes' own aerodynamic changes. A "supported" verdict means the data fit and nothing contradicts them, not proof.

References

FIA. 2026 Formula 1 Technical Regulations, Section C, Issue 20, 5 August 2026. Articles C5.2.7–C5.2.10, C5.4.3, C5.12. fia.com
FIA. 2026 Formula 1 Sporting Regulations, Section B, Issue 09, 1 October 2026. Article B7.2. fia.com
FIA. Refinements to the 2026 FIA Formula 1 regulations agreed by all stakeholders, 20 April 2026. fia.com
FIA. FIA announces ADUO results following first and second review periods, 26 August 2026. fia.com
FIA. Car presentation submissions, every 2026 Grand Prix (decision documents). fia.com
P. Schaefer et al. FastF1, version 3.8. docs.fastf1.dev
GPblog. Mercedes compression ratio saga over as Monaco GP change comes into force, 2026. gpblog.com
Crash.net. Max Verstappen details Red Bull's true F1 engine reality after FIA ADUO verdict, September 2026. crash.net
PlanetF1. Max Verstappen's Malaysia pole and Red Bull's advantage, data analysis, October 2026. planetf1.com
F1-Fansite. Mercedes' Sepang upgrade backfires as W17 loses straight-line speed, October 2026. f1-fansite.com
PlanetF1. Christian Horner: Mercedes and Red Bull pushing F1 2026 limits, not cheating, 2026. planetf1.com
Motorsport.com. Red Bull confident new F1 power unit is legal, albeit "on the very limit" of the rules, 2026. motorsport.com
Autosport. Explained: F1 2026's compression ratio furore, 2026. autosport.com

Data from Formula 1 live timing via FastF1. Upgrade lists from FIA car-presentation submissions. This is an independent analysis, not affiliated with any team or with Formula 1.