Fill the tank before every session. Then 34 F / 32 R hot — don't touch it.
1:41.x is repeatability, not new speed. One rebuild: T5 — brake later and harder, −1.0 g, bleed it through turn-in, throttle at the apex. Four corners where you just drive your good version every lap: Carousel, T2/T3, T7a/T7b, T7 — cues in the table below.
Go early — 8:00 I2, 8:25 Advanced. T8 is flat, no lift. T1: brake light over the bumps, release fast — that alone is ~0.2 s.
Saturday in one paragraph
Six runs. The pressure lesson came first — 30 F / 29 R cold built to 40 / 39 hot and the car went loose; bled to 32 / 30 mid-session and the lap answered immediately, and the day settled on a 34 F / 32 R hot target that then held all day. The PB landed in run 2 — 1:42.70 on lap 22 — immediately after the fuel-starve key cycle, so the car that set it was a car you had just reset, not a car on a magic lap. Afternoon was heat and traffic. Then the evening run at 16:01 went 1:43.34 (lap 50) and 1:43.61 (lap 48) — the day's second-fastest run, at the end of the day, on the oldest rubber. Three laps under 1:44 total. Fuel starve capped laps 19–21 only; everything after lunch was clean.
What that combination means for Sunday: there is no tire cliff in this data and no heat penalty in this data. The evening run was faster than the middle of the day, whole-lap peak lateral g was higher at the end (1.30 g in R6 vs 1.25 g in R2), and every apparent within-run decline traces to one traffic lap. So take the early sessions because the track is cool and the traffic is thin — not because the tires are dying.
Sunday: where the time is
1:41.x needs no new speed. It needs the corners you already drove on Saturday to happen on the same lap.
Your best-segment stitch is 1:41.24 — 1.46 s under the PB — and the contributing segments come from 14 different laps, including three from the very first run of the morning. That is the signature of a repeatability problem, not a pace ceiling.
Ranked by your own spread — your median window time in that corner minus your own best in that corner, over 35 clean laps. Same car, same tires, same day, so this is time you have already demonstrated.
#
Corner
Your spread
vs May
Verdict
Do this, every lap
1
T13/T14 Carousel
+0.37 s
−0.21
attack
No brake. Hit the apex, whole apex curb, stay off the exit seam. Drifting between apexes → tiny lift, not hands.
2
T2/T3
+0.36 s
−0.28
attack
One arc, not two: enter outside/later, trail-brake for rotation, use the left exit rumble strips. Your fast laps already do this.
3
T5 Charlotte's Web
+0.34 s
+0.27
rebuild
Brake later and harder — −1.0 g — bleed through turn-in, throttle AT the apex. Reference: the ground dropping between the 2 and 1 board.
4
T7a/T7b
+0.31 s
−0.14
attack
Left tires on the whole left curb, light brake ending before the right apex curb, ride the seam. Do not take the apex curb.
5
T7
+0.30 s
−0.23
attack
Do not over-slow it: brake soft and early, hover the pedal at the apex curb, graze the left curb.
Top five total
1.68 s
The honest version. Measured against the May M440i template you are now ahead everywhere on minimum speed (+1 to +15 MPH) and ahead on peak lateral g in every corner. Only 0.51 s of the lap still sits above that template at all — T5 is 54 % of it (+0.27 s), then T15 (+0.11), T12 (+0.09), T6 (+0.03). So Sunday is not a hunt for new speed. It is one rebuild (T5) plus four corners where you simply need your good version every lap.
Read your PB honestly. Lap 22 beat a typical 1:44.5 by 1.59 s over the common distance, and it did it in five places at once — T4 crest −0.26, T15 −0.19, T2/T3 −0.19, Carousel −0.16, the back-straight link −0.14, T7a/b −0.12, the run to T1 −0.13. It also gave back 0.16 s in T1 itself. The PB lap has a visible mistake in it and still went 1:42.7. Lap 50 did T1 properly and conceded T2/T3 and the Carousel. Combining just those two laps' T1 lines is worth ~0.2 s on its own.
Numbering. Cards lead with the JTI label — the map JTI hands you, and the language of the paddock this weekend. The grey line under each gives Chin's club numbering plus the canon row index the data is keyed on. They do not line up: JTI carries 18 labels to Chin's 17 and to the canon's 15 rows. Charlotte's Web is JTI T5 (Chin 5/6, canon row 4). Do not hang a coaching cue on a bare number with a stranger — hang it on the landmark: museum building, spider sculpture, low point, crest.
Corner by corner
All speeds MPH. Three numbers per card: Sat best = your fastest minimum of the day · Sat median = your typical minimum over 35 clean laps · May = the M440i median. The gap between the first two is the corner's real problem. Camber: + is on-camber (inside low). Rank = position on the time-available list above. Bold first line of each card = the cue. Everything under it is the why — skip it in the paddock. The four outlined cards (Carousel, T5, T8, T15) plus the table above are the whole plan.
T13/T14Right · 2875 m · the Carousel
Chin 14/15 · canon row 13 · rank 1 · +0.37 s of your own spread
ON-CAMBER +3.3%BLIND APEXHARDEST CORNERcompression +0.28 g, then a 9.1 m climb
Sat best
72.8 MPH
Sat median
66.2
May
63.2
Brake pt
none 0/35 laps
Peak lat g
1.22
No brake. Eyes on the blind apex, hit it, use the whole apex curb, and stay off the exit seam. If it drifts between the two apexes: a tiny lift to settle it — hands stay quiet.
Biggest single prize on the lap: the largest repeatability gap on the circuit (best 6.570 s vs median 6.940 s) — your best Carousel was 0.37 s quicker than your average one. The coaching corpus independently calls this complex "perhaps the trickiest corner of the whole circuit": a blind apex whose accuracy sets your line two corners later, onto the front straight. You take it with no brake on any of 35 laps at 1.22 g, camber building to +6.3 % through the arc and the compression adding load — it takes more than it looks like it will. Running wide over the seam is what wrecks T14a and T15.
T5Left · 1055 m · Charlotte's Web
Chin 5/6 · canon row 4 · rank 3 · the one rebuild — 54 % of what is still above the May template
OFF-CAMBER −1.76%REBUILDthe lap's only threshold-brake zone
Sat best
43.5 MPH
Sat median
40.0
May
39.9
Brake pt
131 m pre-apex
vs May
+0.27 s
Brake later and harder — a genuine −1.0 g stab, the only place on the lap you see that number — then bleed the pressure through turn-in, minimum speed ~10 m before the apex, throttle AT the apex. Reference: the ground dropping away between the 2 and the 1 board. Stay a gear high. The off-camber is invisible to the eye — trust the data, not the view. "Patience" is retired: it was the error.
Your six best T5s against your six worst, and it is not close: on the slow attempts you start braking 17 m earlier, brake 0.4 g softer (−0.60 g vs −1.01 g), are completely off the brake 31 m before the apex instead of 12, overlap turn-in by 3 m instead of 13, hit minimum speed 24 m early instead of 9, and are back on the throttle 22 m before the apex. Entry speed 58 MPH instead of 77. Exit speed is identical either way — every bit of the 0.7 s spread is made and lost between the brake board and the apex. The coaching corpus names exactly this failure mode: braking too early and coming off too early moves weight rearward and produces understeer at turn-in — which then feels like "no grip, be patient", so you brake earlier still. That is the loop the cue breaks. Lap 29 did 11.466 s peaking at −0.96 g; lap 10 did 12.237 s peaking at −0.27 g and coasted 30 m. Same corner, 0.77 s apart.
T2/T3Right · 415 m
Chin 2/3 · canon row 2 · rank 2 · uphill to the apex, one long arc
ON-CAMBER +9.1%BLIND APEXcompression
Sat best
55.6 MPH
Sat median
52.3
May
50.3
Brake pt
105 m pre-apex
One long arc, not two corners. Enter outside and later, trail-brake for rotation, drive the left exit rumble strips. No scrub on track-out — it is slightly off-camber and it costs you the whole climb to T4.
The community's "angle" line — the alternative to hugging the inside edge — is quoted at ~0.5 s here; measured on your car as lateral offset to the outside, it is real but worth about 0.09 s, not 0.5. Measuring your lateral offset through 380–450 m: the outside family ran 11.352 s and 84.0 MPH exit against the inside family's 11.439 s and 82.6 MPH, correlation r = −0.35. Weak, but the tell is which laps sit outside: 22, 24, 25, 26 and 14 — five of your six best laps, and lap 14 owns the fastest T2/T3 of the day at 11.078 s. So keep doing what you already do on the good ones. The real prize is not the line, it is your 0.36 s of spread (best 11.078 vs median 11.435) — rank 2 on the lap.
Left-side tires on the entire left curb. Light brake, done just before the right apex curb, trail it in, then ride the seam between apex curb and pavement. Do not take the apex curb — it feels fast and isn't.
Second-highest bank on the lap (+8.9 % at the apex, behind only T2/T3) and it drops 9.8 m through the section — the bank is what lets you carry a sharp apex at 45 MPH on your best laps, 42 on a typical one. Blind and downhill, so it is a seat-time corner. 0.31 s of your own spread sits here, and you are already 0.14 s under the May template — this is repetition, not bravery.
T7Right · 1630 m
Chin 8 · canon row 6 · rank 5 · entry to the Museum complex
ON-CAMBER +2.3%crest
Sat best
74.4 MPH
Sat median
66.9
May
59.6
Brake pt
121 m pre-apex
Do not over-slow it. Brake only enough to make the corner — softer and sooner than instinct — and hover over the pedal at the apex curbing instead of pressing it. Graze the left-side curb with your left tires so the following right comes without dumping speed.
Over-slowing here is the corpus's #1 listed mistake in this whole section. On your best six laps you are 10.6 MPH over the May template (median-to-median it is 7.3) and 0.23 s under it, but 0.30 s of your own spread remains — best 3.947 s vs median 4.246 s. Same corner, three tenths, no new technique required.
Brake a touch earlier and lighter — ease the pedal over the bumps so ABS stays out — release quickly after turn-in, don't overrun the apex curb, track fully out. No more total brake than you use now.
Already 6 MPH over the template and a proper U-shape. Two things the corpus adds that the data can't see: the brake zone is bumpy, so pressure held over the bumps brings ABS in and eats your entry; and the banked downhill entry makes overrunning the apex curb easy. The specific ask: your PB lap lost 0.16 s right here while lap 50 did it properly. Put lap 50's T1 on lap 22's lap and you have two tenths for free.
T4Right · 900 m · the crest
Chin 4 · canon row 3 · rank 7 · top speed of the lap
≈FLAT −0.5%BLIND APEXelevation max, long gentle crest
Sat best
83.5 MPH
Sat median
71.7
May
69.6
Brake pt
none 3/35 laps
Continue the arc from T2's exit, apex just past the crest, cross it as straight as you can — hands still while the car is light. Flat — you already are on 32 of 35 laps; the only job is arriving at the 985 m mark fast.
The evening run proved 113.9 MPH over the crest is repeatable — the fastest crest speed of the entire day, on the day's lowest fuel load. This was the single biggest contributor to the PB (−0.26 s vs typical). Camber sign on a corner with 2° of direction change is not meaningful; ignore the badge.
T6Right · 1400 m
Chin 7 · canon row 5 · rank 14 · "a bend in the straight"
NEUTRAL −0.2%flat
Sat best
96.1 MPH
Sat median
94.9
May
92.5
Brake pt
none 0/35 laps
vs May
+0.03 s
Flat. Nothing to do here except already be positioned for the T7 brake zone.
0.21 g, no brake event on any lap, 95 MPH minimum, and only 0.09 s of spread — the second-tightest corner on your lap. The +0.03 s vs May is noise on a bend in the straight. Ignore the camber badge on a near-straight.
T8Left · 1880 m · the track-out
Chin does not number this · canon row 8 · rank 15 · lowest point on the circuit, 181.8 m
COMMIT — CAUTION DROPPEDAPEX CAMBER −1.3%one 5 m station reads −6.4%
Sat best
73.7 MPH
Sat median
70.5
May
67.9
Through it
+15.2 MPH
Brake pt
none 0/35 laps
Full throttle, no lift, hands unwinding, eyes through to T9. The compression at the low point gives you grip — take it. Yesterday's caution is retired.
Honest reading of our own survey: apex camber is −1.3 % and the corner mean is −1.7 %; the alarming −6.4 % is camber_min, a single 5 m station, and the same corner also contains a +6.9 % station. One station off-camber, apex fine. And the car agrees emphatically: across all 35 clean laps speed rises from 71.6 to 86.8 MPH through the section, peak lateral g is 0.80 against May's 0.79 — identical, zero of 35 laps show any deceleration inside the window, and the wobble index is the lowest of the entire descent group — T8 is the calmest place on this part of the lap. The mid-window dip in lateral g that looks like a grip loss is just the sign change from T7b (right) into T8 (left). Repeatability gap +0.067 s, dead last of 15. The corpus is blunter still: "turn 8 is, well, not really a corner".
Brush the brake (don't lift), get the direction change done early while the bank is still paying, left-side tires on the whole left curb, no apex curb. Take T9/T10 as straight as you can.
The mechanism to carry in your head: the banking is there at T9 entry and gone by the T10 apex, so the car rotates easily on entry and then refuses to change direction later — that is the surface, not you. Only 0.19 s of spread — this is nearly done.
T10Right · 2190 m
Chin 11 · canon row 10 · rank 12 · ends far left
ON-CAMBER +1.1%BLIND APEXflat, climbing
Sat best
90.0 MPH
Sat median
83.7
May
82.2
Brake pt
none 1/35 laps
Flat. Late apex, direction change finished before the apex, then let it run all the way to the far left. The banking is gone by exit — the car will not turn for you there, so don't ask it to.
The brake event has effectively disappeared in the Z4 (1 lap in 35, against May's 36 m). Lap 50 found 0.13 s here over a typical lap; it is one of the few places the evening run beat the PB.
T11Left · 2560 m
Chin 12 · canon row 11 · rank 6 · into the second chicane
ON-CAMBER +2.8%flat, level ground
Sat best
90.6 MPH
Sat median
82.6
May
71.3
Brake pt
96 m pre-apex, 31/35 laps
Positioning, not braking: inside half of the track for both halves of the chicane, aim two-thirds to three-quarters through the right apex curb. On an early lap, try it brake-free in 4th. Early entry cuts the steep curb and unsettles the car; late entry runs you off.
The corpus says a real fraction of drivers do this pair with no brakes at all, in a tall gear — that is the ceiling to test, and on your best six laps you are already 15 MPH and 0.35 s over the May template here, so you have the room. If it works, you keep 0.29 s of spread and the T12 entry gets easier too.
T12Right · 2675 m
Chin 13 · canon row 12 · rank 8 · the compression
NEUTRAL +0.2%BLIND APEXcompression +0.27 g
Sat best
66.2 MPH
Sat median
63.1
May
63.1
Brake pt
57 m pre-apex
vs May
+0.09 s
Brake into the squat at the low point, trail it through while already placing the car for the Carousel, aim two-thirds to three-quarters through the right apex curb, squeeze power, cross the rise as straight as possible. Two-corner thought, not one.
One of only four corners still above the May template, and the only corner where your median minimum speed exactly equals May's — 63.1 MPH both. Begins on level ground with a blind apex. The track hands you grip exactly where you need it: +0.27 g of compression at the low point. 0.26 s of spread and 0.09 s owed to May: there is still real time sitting here.
T14aRight · 3050 m
Chin 16 · canon row 14 · rank 13 · highest point on the circuit, 203.5 m
ON-CAMBER +7.6% AT APEXEXIT FALLS AWAYcrest, 203.5 m
Sat best
54.5 MPH
Sat median
50.9
May
47.9
Brake pt
57 m pre-apex, 21/35 laps
Lift off the gas AT the turn-in point — weight onto the nose — then turn. Feed the power in; no full throttle before the apex, because the road falls away on exit.
New technique, straight from the corpus. If you stayed off the Carousel exit seam you arrive on the right side of the track, which is what makes this work. Anticipate the off-camber exit (apex is +7.6 % but it drops 2.2 m to exit). Carry as much speed off the turn as you can, because this is the middle link of the sacrifice chain into T15. Only 0.15 s of spread — you already do this well; the change is the deliberate lift, not more speed.
T15Left · 3210 m · onto the front straight
Chin 17 · canon row 15 · rank 10 · the end of the sacrifice chain
ON-CAMBER +6.1%BLIND APEXTRACK-OUT OFF-CAMBERexit curb you are not using
Sat best
66.9 MPH
Sat median
62.7
May
60.8
Brake pt
none 0/35 laps
vs May
+0.11 s
Earliest throttle of the lap, bought two corners earlier: stay off the Carousel exit seam, take T14a on the deliberate lift so you arrive here placed and pointed, and get on the gas early. Reference: about three-quarters of the way into the apex curb. Keep the steering wound and accelerate up the hill — do not unwind and hope. Use the exit curb. A small lift before turn-in is fine.
This is a three-corner sacrifice chain, not a corner. The corpus is explicit — deliberately compromise the two corners before it to buy the earliest possible throttle here, because early throttle onto the front straight pays you across the longest full-throttle stretch of the lap. Blind apex; the exit curb is highly usable and most drivers never touch it; the track-out is slightly off-camber. It contributed −0.19 s to the PB and your best T15 of the day (5.580 s) came from lap 5 — an opening-run 1:45.6. You have driven the right version once; it is not a mystery, it is a choice.
Fuel — the one thing that can end your Sunday
Mechanism, once, so the rule makes sense. The G29 has a saddle tank straddling the driveshaft tunnel with one delivery module — pump, collector, level sender — on the active side; the passive side is scavenged by a flow-driven venturi jet pump, not a second electric pump. Sustained lateral g pushes fuel to the far side; low vertical g removes the fuel head. The collector empties, the pump draws air, flow stops — and because the jet pump is flow-driven, transfer stops too, so the passive half can hold gallons you cannot reach. Rail pressure sags, the DME throws P0087 and latches a reduced-power map to protect the HPFP, turbo and cats. It does not self-clear. An ignition cycle clears it.
The rule: top off before EVERY session. $0, no parts, no compromise, and it is genuinely the whole answer. The "quarter tank" threshold is street folklore — your car cut Saturday at only ~6 gallons under full, about 56 % of the tank, over the T4 crest at ~112 MPH. Aux transfer pumps and surge tanks fix it properly but there is no confirmed G29 part; Hydramat and an LPFP upgrade do not address this cascade.
Where it hits — two candidate places, and you should respect both. Your own confirmed event was the T4 crest, where four factors stack: sustained right-hand lateral, near-zero vertical g at the elevation maximum (the decisive one — it removes the fuel head), peak full-throttle demand, and hot fuel. The closest published analogue — a 2007 Z4 M Coupe at Barber — starved instead at the Carousel and the full-throttle climb after it: 1.3–1.55 g at part throttle for ~7 seconds, then WOT still at ~1 g for another ~7, engine cutting for 2–3 s at the top of the hill and recovering under the next braking zone, at ~4 gallons. Different corner, same physics, and the Carousel-plus-climb is the longest sustained lateral load on the lap. Following the fill rule makes the location argument moot, which is why the rule is the headline.
If it cuts anyway:do not lift abruptly — you are RWD, light, and mid-corner. No braking mid-crest. Neutral hands, expect a torque surge just before the latch, then point-by and pit: limp can be walking pace, so do not try to complete a lap at speed. Cycle the ignition in the paddock and add fuel. If you have a spare minute this weekend, pull codes to confirm P0087 rather than an IAT-driven torque pull — they feel identical from the seat and have different fixes.
Between sessions — the data loop
Expect a lag, and plan around it. Lap headers and lap times land within minutes of upload. The 25 Hz telemetry does not: Saturday's flush Lambda timed out on a 147k-row session and had to be retried, and the day's own log put lap_fix up to ~2 h behind the lap headers, so budget 30–60 minutes at best for per-corner analysis on a big session (TDAI-139, successor to TDAI-105). Nothing is lost when this happens — Aurora has every lap; it is the lakehouse flush that stalls. So upload immediately after each session and go do something else; don't stand at the laptop waiting for corners.
Where it lands: laps in tdai_scratch.barber_aug2026_weekend_laps, the May comparator in tdai_scratch.barber_aug2026_corner_baseline.
The one-liner to ask for between sessions:
per-corner window time this session, ranked by
(my median − my best) desc, with min MPH vs the May median
That is the single most useful view, because Sunday's problem is spread, not pace. Ask for two corners, never more. Secondary: session peak lateral g (1.30 was Saturday's ceiling, in the evening run; under 1.10 means you are not using the tire). Do not diff April Z4 corner numbers against anything — those logged at ~1.05 Hz and miss the apex minimum entirely.
Instructing Taylor
Taylor Whatley · novice · 2020 M4, car #33. More power and more weight than he can use yet, on a track with seven blind apexes.
Sunday schedule: his Novice sessions are 9:15, 12:45, 2:00 and 3:35, with classroom at 8:00 and 1:15. Your own runs are 8:00 I2 and 8:25 Advanced — so you get your fast laps in before his 9:15. The 8:00 classroom overlaps your I2 session, so agree with him last thing tonight that he goes to it solo and you debrief at 9:00. Confirm every time on the board at check-in — only your own 8:00/8:25 are confirmed in the mesh.
One thing per session — four sessions, four ideas, no more. Your own six seconds in May came from a single concept applied fifteen times, not fifteen fixes. Suggested order: 9:15 vision and looking further ahead · 12:45 braking in a straight line, releasing smoothly · 2:00 one corner, T7 entry · 3:35 putting a whole lap together.
Blind apexes are a seat-time problem, not a bravery problem. T2/T3, T4, T7a/T7b, T10, T12, the Carousel and T15 all hide the apex from the braking reference. Tell him that up front so he doesn't read his own inconsistency as a lack of commitment.
In an M4, T7 entry is where he'll lose the most time. Over-slowing it is the corpus's #1 listed mistake and a heavy, powerful car makes it worse. Soft and early beats late and hard; hover the pedal at the apex curbing.
T5 Charlotte's Web is the one to keep him conservative at (his cue is the opposite of yours, on purpose) — off-camber, downhill approach, the primary passing zone, and the corner that most tempts a novice into too much entry speed. Straight-line braking first; rotation later in his career. Note the off-camber is invisible to the eye, so "it looks fine" is not evidence.
T8 needs a name, not a warning. Chin's numbering skips it, so it may not be on his map at all. Frame it as "track-out, hands unwinding, full throttle" — our own data says it is the calmest place on that part of the lap, so don't plant a fear there.
The Carousel is where a novice M4 gets loose. It is the "confidence turn" and the corpus notes loose cars start to drift between its two apexes. Teach the fix in advance: lift slightly to settle, don't correct with steering.
Managing your own day: instructing eats sessions, but Saturday's evidence says the tires are not the constraint — the 16:01 run was the day's second-fastest on the oldest rubber, and peak grip was higher at the end of the day than in the middle. Take the early sessions for cool track and clear traffic, then teach without feeling like you are spending your lap time. And if lap times ever do stop responding, that is the tires, not you — don't chase it with setup changes at the track. Pressures stay at 34 F / 32 R hot; the target held all day and there is no reason to touch it.