ne0: forum

CLASS A RACING CUE CARD

Track Cars, Configs, Turbo Profiles, and Garage Build Order

CLASS A TRACK-BY-TRACK QUICK INDEX

Hey, I'm ne0. I'm still a pretty new Torn player - a little over a year in - but I'm a huge racing nerd, so I went deep into Class A guides, forum posts, scripts, lap records, and old comments trying to understand what people are actually using and why.

I wrote this down after a lot of reading because a lot of racing information gets passed around as "this is the meta" or "just use this car" without much of a source trail. This is my attempt to condense the useful bits into a practical Class A racing cue card: which car to use on each track, which surface build to run, whether to use short ratio or long ratio, and whether the car wants T2 or T3 turbo.

Why this exists: This is not meant to be a perfect physics model, a new authority, or a replacement for testing, I'm not looking into stealing someone's thunder. It is just a cleaner index for people like me who want the quick answer while still being able to click through to the source material. Where the community disagrees, I tried to mark the disagreement instead of hiding it.

Credit first: This post is only a condensed index of the data, testing, guides, scripts, and community knowledge built by other racers over the years. It is not a replacement for the original guides. It is an easier-to-read cue card for quick car selection, written from the perspective of a newer racer trying to organize the work that better racers already did.

Huge thanks to Baldr, RD_ / WhatTheFoxSay, the_khush, Rogue, Grando, Madmiketyson, AirQuotes, Bobdawildman, Andyman, Rosti, and everyone who has tested, corrected, scripted, logged laps, and shared racing information.

Use this page to pick quickly.Use the full guides when you need the reasons, exceptions, point strategy, recovery advice, racing skill details, and deeper testing.
Start with:Baldr's Racing FAQ, RD_'s Racing Guide 2.0, khush's Class A notes, Rogue's No Points Recovery Guide, Grando's upgrade path guide, and Rosti's Racing Helper script.

Important note: Torn racing still has RNG, driver skill, upgrade completion, race-size variance, and record-lap vs consistent-placement differences. Treat this as the current community-meta starting point, not an official guarantee.
Primary sources:khush's Class A upgrade notes, Racing Official Race Helper script, Torn Wiki Raceway track list, and Class A top-10 lap records.

Visual Garage Snapshot

Edomondo NSX

9 tracks across SR/T2, SR/T3, and dirt LR/T3 builds

Veloria LFA

Withdrawal, Speedway, and practical Uptown alternate

Lambrini Torobravo

Uptown specialist pick

Volt GT

Docks specialist

Mercia SLR

Convict specialist

Echo R8

Stone Park specialist

Colina Tanprice

Mudpit specialist, despite being Class B

Setup Legend

Tarmac build: Track Tires, Paddleshift Gearbox, Adjustable Coil-over Suspension.

Dirt build: Rally Tires, Rally Gearbox, Group N Rally Suspension.

LR: Long Ratio   |   SR: Short Ratio.

T2/T3: Stage 2 Turbo / Stage 3 Turbo. T3 is the default on most builds; T2 only appears on a few specialist short-track builds.

Turbo 2 vs Turbo 3: why it is still debated

Simple version:T3 is the default safe answer for most Class A builds, especially while you are still building your garage. T2 is not automatically worse, though. It shows up on a few specialist builds because some short/technical tracks appear to value acceleration more than the extra top-speed ceiling from T3.

Why T2 can make sense: community explanations usually describe the tradeoff as T2 = slightly better acceleration and T3 = slightly better top speed. So on tracks with more braking, corners, or repeated acceleration zones, T2 can sometimes beat the paper-stat logic of "higher stage must be faster".

Why it is not settled: Torn racing has RNG, driver skill, car history/possible car experience theories, race size, and lap-to-lap variance. RD_'s guide specifically treats some T2/T3 calls as very close, and older forum discussion has people reporting both sides depending on car, track, and testing sample. That is why this cue card only uses T2 where the current Class A table/helper meta actually calls for it.

How I read the meta here: T2 is marked only for Underdog, Commerce, Hammerhead, and Sewage. Everything else stays T3 unless you are deliberately testing your own car and accepting that your results may vary.

Small remap note: when racers say "T2" in modern shorthand, they are usually talking about the turbo profile, not saying every Stage 2 engine upgrade is better. Current helper shorthand lists T2 as the Stage Two Turbo Kit with the stronger ECU remap, and RD_ also notes that lower remap stages tested worse than the stronger remap.
Sources:khush Class A table, RD_'s Racing Guide 2.0, Turbo 2 vs Turbo 3 thread, Best Cars simplified thread, and racing helper shorthand.

Best Class A Cars by Track

Table source key: khush Class A notes + Rosti/helper table. Row notes are quick caveats; fuller links are in Source Notes.

Track	Surface	Best Car	Old/Common Name	Ratio	Turbo	Notes
Uptown 2.25mi	Tarmac	Lambrini TorobravoDEBATED	Lamborghini Gallardo	LR	T3	Specialist; LFA is practical/consistency alt.
Withdrawal 3.4mi	Tarmac	Veloria LFA	Lexus LFA	LR	T3	Long tarmac.
Underdog 1.73mi	Tarmac	Edomondo NSX	Honda NSX	SR	T2	Short tarmac T2.
Parkland 3.43mi	Dirt	Edomondo NSX	Honda NSX	SR	T3	Dirt short-ratio T3 build.
Docks 3.81mi	Tarmac	Volt GT	Ford GT	LR	T3	Docks specialist long-ratio build.
Commerce 1.09mi	Tarmac	Edomondo NSX	Honda NSX	SR	T2	Short tarmac T2.
Two Islands 2.71mi	Dirt	Edomondo NSX	Honda NSX	LR	T3	Dirt long-ratio T3 build.
Industrial 1.35mi	Tarmac	Edomondo NSX	Honda NSX	SR	T3	Short tarmac T3.
Vector 1.16mi	Tarmac	Edomondo NSX	Honda NSX	SR	T3	Short tarmac T3.
Mudpit 1.06mi	Dirt	Colina TanpriceCLASS B	Sierra Cosworth	LR	T3	Class B car, but still the meta Mudpit pick here.
Hammerhead 1.16mi	Dirt	Edomondo NSX	Honda NSX	SR	T2	Dirt short-ratio T2 build.
Sewage 1.5mi	Tarmac	Edomondo NSX	Honda NSX	SR	T2	Short tarmac T2.
Meltdown 1.2mi	Tarmac	Edomondo NSX	Honda NSX	SR	T3	Short tarmac T3.
Speedway 0.9mi	Tarmac	Veloria LFA	Lexus LFA	LR	T3	Long tarmac.
Stone Park 2.08mi	Dirt	Echo R8	Audi R8	SR	T3	Stone Park specialist.
Convict 1.64mi	Tarmac	Mercia SLR	Mercedes SLR	LR	T3	Convict specialist.

Uptown Caveat: Lambrini vs Veloria LFA

vs

Guide/meta pick: Lambrini Torobravo, Tarmac, LR, T3.

Practical alternate: Veloria LFA, Tarmac, LR, T3.

The Lambrini page says the racing community is undecided on Uptown: Lambrini has better acceleration/braking, while Veloria has better top speed. The Class A lap-record list also shows LFA and Lambrini very close on Uptown, with Stormatti entries ahead because it is a special prize car.

Recommendation: build Lambrini if you want the specialist guide pick. Use Veloria LFA as the safer broad-coverage choice if you want one LR/T3 tarmac car that already covers Withdrawal and Speedway, and can still argue for Uptown.
Sources:Lambrini page, Veloria page, Class A records, and khush comment thread.

Garage Shopping List

Car Build	Tracks Covered
Lambrini Torobravo Tarmac / LR / T3	Uptown specialist. Optional if you use Veloria as the practical Uptown alternate.
Veloria LFA Tarmac / LR / T3	Withdrawal, Speedway, and possible Uptown alternate. One is enough if you do not need duplicate/dedicated cars.
Volt GT Tarmac / LR / T3	Docks
Mercia SLR Tarmac / LR / T3	Convict
Echo R8 Dirt / SR / T3	Stone Park
Colina Tanprice Dirt / LR / T3	Mudpit
Edomondo NSX Tarmac / SR / T2	Underdog, Commerce, Sewage
Edomondo NSX Tarmac / SR / T3	Industrial, Vector, Meltdown
Edomondo NSX Dirt / SR / T2	Hammerhead
Edomondo NSX Dirt / SR / T3	Parkland
Edomondo NSX Dirt / LR / T3	Two Islands

Recommended Build Order

1. Edomondo NSX - Tarmac / SR / T3: covers Industrial, Vector, Meltdown.
2. Edomondo NSX - Dirt / SR / T3: Parkland plus useful dirt fallback. 
3. Veloria LFA - Tarmac / LR / T3: covers Withdrawal and Speedway, and Uptown.
4. Edomondo NSX - Tarmac / SR / T2: covers Underdog, Commerce, Sewage.
5. Colina Tanprice - Dirt / LR / T3: Mudpit specialist. * 
6. One-track specialists: Volt GT, Echo R8, Mercia SLR, then Lambrini Torobravo if you want the Uptown specialist pick.
   
   * If you already have the Colina from the getgo, as many people use it to progress on B class, prioritize it.

Special Prize Car Note

Stormatti Casteon: the guide notes that Stormatti can replace Lambrini on Uptown and Mercia SLR on Convict, but it is a competition prize car, not a normal starter garage item.
Source:khush's Class A upgrade notes and Torn Wiki car list.

Track Context

Torn currently has 16 Raceway tracks, ranging from Speedway at 0.9mi to Docks at 3.81mi. Track length does not decide the car by itself, but it helps explain why some tracks favor long-ratio speed builds and others favor short-ratio acceleration/handling builds.
Source:Torn Wiki Raceway track list.

Source Notes

Main setup table:khush's Class A upgrade notes cross-checked with the Racing Official Race Helper script and older RD_ / WhatTheFoxSay racing guidance.

Beginner and deeper racing reading:Baldr's Racing FAQ, Rogue's recovery guide, and Grando's upgrade path guide.

Track list, lengths, and images:Torn Wiki Raceway.

Car classes and car images:Torn Wiki Cars list.

Uptown Lambrini/LFA caveat:Lambrini Torobravo page, Veloria LFA page, and Class A top-10 lap records.

Turbo 2/Turbo 3 comments:Turbo 2 vs Turbo 3 discussion, Best Cars simplified thread, RD_'s Racing Guide 2.0, and helper shorthand.

Mudpit Class B note:Torn Wiki Cars list Colina Tanprice is Class B, even though it is the recommended Mudpit class A specialist.

Thank You

Thank you again to everyone who has done the hard work before this post: building guides, testing cars, updating tables, answering questions, logging laps, writing scripts, and correcting outdated info.

This cue card exists because of that work. I only tried to make the first lookup faster, keep the sources close to the claims, and mark the places where the answer is still a little debated. The original guides are where the real depth lives.

Last checked: Sep 2026. Racing meta can move as more lap data and testing comes in.

OFF-META VIABLE CARS

leaderboard-aware alternates, legacy picks, and practical fallback cars

OFF-META DOES NOT MEAN BAD

This is a follow-up / companion post to the main Class A cue card. The goal is not to rewrite the meta table, but to mark the cars worth testing or using as fallbacks when the clean table answer is not available, or when your own lap data says something interesting.

After the Docks feedback, I did a track-by-track pass through ShadowSentinel's Unofficial Racing Leaderboards and tightened the list. Some cars got promoted, and one big one got demoted: Veloria LFA should not be listed as a Docks alternate.

Important: This is not a new meta table. The unofficial leaderboard is useful, but PB-based. The author says it cannot reliably prove exact upgrades or RS at the time of every lap, so treat this as testing leads, not gospel.
Start with: Unofficial Racing Leaderboards, khush's Class A notes, Rosti's Racing Helper, and LtCabel's Racing UI/UX Enhancements.

What changed from feedback: Weston on Convict stays. The leaderboard and player feedback both support mentioning it.
LFA on Docks is removed. The posted Docks table has LFA far behind Volt, NSX, and Lolo.
Stone Park and Mudpit get better alternates. Stone Park has Lolo/Veloria almost tied with Echo; Mudpit has Tsubasa much closer to Colina than the Class A fallbacks.

Track	Cue-card pick	Viable / tested alternate(s)	Verdict / sources
Uptown	Lambrini Torobravo 1:11.820	Veloria LFA 1:11.800 Stormatti Casteon 1:11.610 Weston Marlin 177 1:12.360	Veloria is a real normal-garage alternate. It is basically tied with Lambrini on the posted PB table. Stormatti is faster, but special/rare. Weston is legacy-viable, but less convincing than Veloria. Sources: Class A PBsCue-card
Docks	Volt GT 2:40.120	Edomondo NSX 2:41.780 Lolo 458 2:43.399	Correction: remove Veloria LFA. The posted LFA PB is 2:55.433, way behind Volt, NSX, and Lolo. NSX is the believable temporary fallback if you do not own Volt yet. Sources: Class A PBs
Commerce	Edomondo NSX 0:49.360	Lolo 458 0:50.240	Lolo is the only close-ish listed alternate. NSX still wins the posted table, but Lolo is within +0.880s, so it is worth remembering as a fallback. Sources: Class A PBs
Vector	Edomondo NSX 1:00.130	Lolo 458 1:01.139 Weston Marlin 177 1:01.442 Echo R8 1:01.849	Lolo is viable as a fallback. NSX is still fastest, but Lolo is close enough to mention. Weston also looks decent here, but treat it as legacy/fallback rather than replacement. Sources: Class A PBs
Mudpit	Colina Tanprice 0:35.420	Tsubasa Impressor 0:36.035 Edomondo NSX 0:40.050 Volt GT 0:41.314	Tsubasa is the real Colina backup. It is only +0.615s on the posted Class B table. NSX/Volt/Echo are Class A fallbacks, but much further behind. Sources: Class B PBsClass A PBs
Hammerhead	Edomondo NSX 0:54.720	Echo R8 0:55.470 Colina Tanprice 0:56.210 Veloria LFA 0:56.731	Echo is a believable backup. NSX still wins the posted Class A table, but Echo is within +0.750s. Colina is not crazy as a dirt fallback, though slower than Echo here. Sources: Class A PBsClass B PBs
Meltdown	Edomondo NSX 0:56.470	Lolo 458 0:57.630 Volt GT 0:57.942 Echo R8 0:58.030	Useful fallbacks, not a new meta. Lolo is closest at about +1.160s, with Volt/Echo also not terrible, but NSX still owns the row. Sources: Class A PBs
Speedway	Veloria LFA 0:26.548	Stormatti Casteon 0:26.200 Weston Marlin 177 0:26.778 Mercia SLR 0:26.889	Weston and Mercia are credible backups. Stormatti is fastest on the posted table but special/rare. LFA stays the normal cue-card pick. Sources: Class A PBs
Stone Park	Echo R8 1:11.480	Lolo 458 1:11.500 Veloria LFA 1:11.520 Edomondo NSX 1:11.992	Lolo and Veloria are very real here. They are basically tied with Echo on posted PBs. NSX is still practical, but not the closest table-backed alternate. Sources: Class A PBs
Convict	Mercia SLR 0:59.993	Weston Marlin 177 0:59.750 Stormatti Casteon 0:59.100 Veloria LFA 1:00.128	Weston absolutely deserves the mention. The posted table has Weston ahead of Mercia, matching the feedback. Stormatti is fastest but special/rare; Veloria is close enough to be a practical fallback. Sources: Class A PBsRD old guide

Checked, but not promoted

• Withdrawal: Veloria still clearly leads. Mercia is the closest normal fallback, but not a serious replacement. PB table
• Underdog: NSX still leads. Lolo is the closest normal fallback, around +1.4s. PB table
• Parkland: NSX is far ahead; I would not list a strong alternate from this table. PB table
• Two Islands: NSX is far ahead; GT-R/Volt are interesting context, not replacement picks. PB table
• Industrial: NSX is far ahead; Lolo is the best non-NSX PB but still too far back. PB table
• Sewage: Leave as NSX/meta for now. The Class A table has odd-looking entries, so it needs dedicated testing before changing the guide. PB table

Testing note: Rosti vs LtCabel scripts

Rosti's Racing Helper is still the quick car/setup selector. LtCabel's maintained Racing UI/UX Enhancements, based on Lugburz's older script, is the one I would add for testing/logging because it shows RS to 5 decimals, last gain, race results, race links, PDA support, and CSV export.
That makes it useful for T2/T3 and off-meta testing, especially 50-100 lap comparisons instead of judging from one lucky PB.

Bottom line: Off-meta viable means worth testing, not automatically better. The strongest corrections from this pass are: Weston Convict is real, LFA Docks is not, Tsubasa belongs in the Mudpit backup conversation, and Lolo/Veloria deserve mention on Stone Park.

Small API change request for the racing records endpoint: please expose racing lap times with one extra decimal place, or add a separate millisecond/exact-time field.

The current /v2/racing/{trackId}/records?cat={class} response returns the top 5 lap records for a track and class, but the returned lap_time is rounded to two decimal places. That creates a lot of apparent ties in record boards, especially on short or highly optimized tracks.

Example That Triggered This

The original racing forum discussion started here: Did I break the E-class Speedway Global Record?

A player ran an E-class Speedway race where a racing script calculated their best lap as 32.130. Torn's event log showed the same lap only as 00:32.14.

The question was simple: did they beat the global record, tie it, or miss it? With only two decimals exposed, the answer cannot be proven from Torn's own public data.

The race endpoint also returned best_lap_time: 32.14, and the Speedway E records endpoint returned five records all shown as 32.14. The API clearly has an ordered list, but the returned precision hides the actual differences.

Test performed:

I queried every current racing track returned by /v2/racing/tracks.

Classes tested: E, D, C, B, A.

Total tested: 16 tracks x 5 classes = 80 leaderboards.

Each leaderboard returns 5 records, so the sample was 400 returned record rows.

Result Summary

34 / 80 track/class leaderboards had at least one duplicate returned lap time.

82 / 400 returned record rows were part of a visible tie group.

44 placements could not be uniquely explained from the returned two-decimal lap_time alone.

The clearest example was Speedway Class E, where all five returned records showed exactly 32.14.

By Class

By Track

Examples

Speedway E: 32.14, 32.14, 32.14, 32.14, 32.14

Speedway D: 30.31, 30.34, 30.34, 30.34, 30.35

Mudpit C: 54.64, 54.69, 54.70, 54.70, 54.70

Vector A: 60.13, 60.15, 60.15, 60.22, 60.22

Where This Matters Most

The biggest impact is on short, highly optimized tracks such as Speedway. In the test, Speedway had visible ties in all five classes, and 16 / 25 returned Speedway record rows were part of a tie group.

Example: if the real stored Speedway E times were 32.136, 32.139, 32.141, 32.143, and 32.144, the current API would flatten all of them into 32.14. A millisecond field would show the actual order immediately.

That is exactly why a player with a script-calculated 32.130 lap cannot prove the result from public Torn data. With three decimals or milliseconds, the answer becomes clear instead of depending on hidden ordering or guesswork.

Why One More Digit Helps

Two decimal places group all times into 0.01 second buckets. On optimized racing records, many top laps are close enough that different laps collapse into the same returned value.

Adding one more decimal place would reduce that bucket from 0.01s to 0.001s. That is ten times more precise and would separate many of the current apparent ties.

This would make record boards, racing tools, league analysis, fastest-lap comparisons, and progress tracking much more reliable without exposing any new private data.

Suggested Backward-Compatible Change

Please keep the existing lap_time field for compatibility, but add one precise field:

{
  "lap_time": 32.14,
  "lap_time_ms": 32143
}

Another option would be:

{
  "lap_time": 32.14,
  "lap_time_exact": 32.143
}

The integer millisecond version is probably safest for developers because it avoids floating point formatting issues.

Request

Please expose racing lap records to at least millisecond precision in the API.

This is a small additive API improvement, but it would remove a real limitation for racing tools and analysis. The endpoint already has the ordered records; the current precision just hides too much of the timing difference.

you should NOT grind XP while you don't have at least a handful of cars to cover most tracks if you want to take racing seriously, or else you're probably going to have a very hard time recovering from that, as you're getting matched with people with perfect garages and in the same RS bracket you are. 
grind points until you can cover most tracks at least. like Rogue mentioned, get a NSX TS3 to get you started. 

read this up:
https://www.torn.com/forums.php#/p=threads&f=21&t=16562147&b=0&a=0

but basically, it's unrelated to torn itself, it's player ran. 

May be it's time to get admins involved and make 500 car races available for these qualifiers lol. It's stressful to get in, yeah. 

it was a lot of fun indeed, thanks for that, ggs all around.

https://www.torn.com/forums.php#/a=0&b=0&f=67&p=threads&t=16488820

I'd go for the Veloria LFA, TL3, then afterwards build a Dirt NSX DS3. 
Only then I'd go for the TS2 NSX then colina/volt gt and so on. 

oh hell yeah

you're a friend finch. <3

then you can just join the race with the wrong car and always win a point, then everybody would pick the worse meta, making it the same effort as I don't know... winning?
just follow the guides as people have mentioned, if you still can't, join a mechanic job for a little while and trade in some points. 

Right, that's kinda what I meant. I agree with you, but I don't think it's "that simple".

I think the result is predetermined when the grid locks, but the "predetermined" result could still come from seeded RNG.

Something like this:

RaceTime = Lap1 + Lap2 + Lap3 + ... + LapN

Each lap:

LapTime = ExpectedLap + LapRNG + SectionRNG + IncidentRoll

Where:

ExpectedLap = TrackBase - RSBonus - CarFitBonus - UpgradeBonus

So let's say we have two drivers:

Driver A = RS76 + perfect Volt
Driver B = RS21 + Veloria

Because Driver A has better RS and a better build, the expected lap is better:

Driver A expected lap = 60.00 sec
Driver B expected lap = 60.25 sec

So Driver A should gain:

0.25 sec per lap
5.00 sec over 20 laps
25.00 sec over 100 laps

Now add seeded rolls:

LapRNG = normal lap variation
SectionRNG = braking/cornering/acceleration checks
IncidentRoll = rare bad lap, mistake, traffic, slowdown, etc.

Example formula:

LapTime = ExpectedLap + random(-0.75 to +0.75) + 5 section rolls + possible incident penalty

Same formula. Different seed.

Seed 2:

20 laps:
Driver A: 1204.393 sec
Driver B: 1206.512 sec

Driver A wins by 2.119 sec

100 laps:
Driver A: 6015.832 sec
Driver B: 6042.854 sec

Driver A wins by 27.022 sec

That is what you'd expect. The better RS/car/build averages out ahead.

But now use another seed.

Seed 47:

20 laps:
Driver A: 1205.603 sec
Driver B: 1205.484 sec

Driver B wins by 0.119 sec

100 laps:
Driver A: 6024.357 sec
Driver B: 6023.586 sec

Driver B wins by 0.771 sec

Same formula, but different seeded rolls.

In Seed 47, Driver A was still the better expected setup. Driver A had a 25 second expected advantage over 100 laps. But the random/incident side swung the other way hard enough to erase that advantage.

That is the issue I mean.

It is not that RS/car/build do not matter. In the formula, they clearly do matter because they lower ExpectedLap.

The problem is if this part:

LapRNG + SectionRNG + IncidentRoll

is big enough to overpower this part:

RSBonus + CarFitBonus + UpgradeBonus

Then a weaker RS/car can still beat a stronger RS/car, even over 100 laps.

On short races, that is totally understandable. On 100 laps, it feels like the deterministic part should have more weight, or the RNG variance should be narrower.

It would make sense having huge crashes involving multiple cars in ovals, like IRL, but considering how expensive on points every crash is on this game... it's best it doesn't happen. 

Been wondering about this for a while too. I don’t think RS and car choice are unimportant, but I also don’t think the race formula is fully deterministic.

It feels like there are repeated RNG checks throughout the race, maybe not literally one roll, but a bunch of small checks for braking, cornering, acceleration, consistency, etc. Otherwise the person with the best lap would almost always have the best overall time, and that clearly doesn’t always happen.
Think of like every lap we have multiple dice rolls. I kind of see it as Torn simulating race inconsistency: bad braking zones, overheating, traffic, poor exits, tiny mistakes adding up over 100 laps. So better RS and the right car probably improve your odds, but they don’t guarantee the outcome every single race.

was this ever fixed?

Small API change request for the racing records endpoint: please expose racing lap times with one extra decimal place, or add a separate millisecond/exact-time field.

The current /v2/racing/{trackId}/records?cat={class} response returns the top 5 lap records for a track and class, but the returned lap_time is rounded to two decimal places. That creates a lot of apparent ties in record boards, especially on short or highly optimized tracks.

Example That Triggered This

The original racing forum discussion started here: Did I break the E-class Speedway Global Record?

A player ran an E-class Speedway race where a racing script calculated their best lap as 32.130. Torn's event log showed the same lap only as 00:32.14.

The question was simple: did they beat the global record, tie it, or miss it? With only two decimals exposed, the answer cannot be proven from Torn's own public data.

The race endpoint also returned best_lap_time: 32.14, and the Speedway E records endpoint returned five records all shown as 32.14. The API clearly has an ordered list, but the returned precision hides the actual differences.

Test performed:

I queried every current racing track returned by /v2/racing/tracks.

Classes tested: E, D, C, B, A.

Total tested: 16 tracks x 5 classes = 80 leaderboards.

Each leaderboard returns 5 records, so the sample was 400 returned record rows.

Result Summary

34 / 80 track/class leaderboards had at least one duplicate returned lap time.

82 / 400 returned record rows were part of a visible tie group.

44 placements could not be uniquely explained from the returned two-decimal lap_time alone.

The clearest example was Speedway Class E, where all five returned records showed exactly 32.14.

By Class

By Track

Examples

Speedway E: 32.14, 32.14, 32.14, 32.14, 32.14

Speedway D: 30.31, 30.34, 30.34, 30.34, 30.35

Mudpit C: 54.64, 54.69, 54.70, 54.70, 54.70

Vector A: 60.13, 60.15, 60.15, 60.22, 60.22

Where This Matters Most

The biggest impact is on short, highly optimized tracks such as Speedway. In the test, Speedway had visible ties in all five classes, and 16 / 25 returned Speedway record rows were part of a tie group.

Example: if the real stored Speedway E times were 32.136, 32.139, 32.141, 32.143, and 32.144, the current API would flatten all of them into 32.14. A millisecond field would show the actual order immediately.

That is exactly why a player with a script-calculated 32.130 lap cannot prove the result from public Torn data. With three decimals or milliseconds, the answer becomes clear instead of depending on hidden ordering or guesswork.

Why One More Digit Helps

Two decimal places group all times into 0.01 second buckets. On optimized racing records, many top laps are close enough that different laps collapse into the same returned value.

Adding one more decimal place would reduce that bucket from 0.01s to 0.001s. That is ten times more precise and would separate many of the current apparent ties.

This would make record boards, racing tools, league analysis, fastest-lap comparisons, and progress tracking much more reliable without exposing any new private data.

Suggested Backward-Compatible Change

Please keep the existing lap_time field for compatibility, but add one precise field:

{
  "lap_time": 32.14,
  "lap_time_ms": 32143
}

Another option would be:

{
  "lap_time": 32.14,
  "lap_time_exact": 32.143
}

The integer millisecond version is probably safest for developers because it avoids floating point formatting issues.

Request

Please expose racing lap records to at least millisecond precision in the API.

This is a small additive API improvement, but it would remove a real limitation for racing tools and analysis. The endpoint already has the ordered records; the current precision just hides too much of the timing difference.

Edomondo NSX - Tarmac / SR / T3: covers Industrial, Vector, Meltdown.

Edomondo NSX - Tarmac / SR / T2: covers Underdog, Commerce, Sewage.

Veloria LFA - Tarmac / LR / T3: covers Withdrawal and Speedway, and can be your practical Uptown alternate.

Edomondo NSX - Dirt / SR / T3: Parkland plus useful dirt fallback.

Colina Tanprice - Dirt / LR / T3: Mudpit specialist.

This should be your basic garage... then you move into the single track cars. Probably Volt GT for Docks and so on. 
Pick what you see fit from there. 

you know... you should mention that this is paid. 

18:52:27 - 20/06/26 You bought some Xanax on Deadringers's bazaar at $836,400 each for a total of $836,400

I have a suggestion if you don't mind.
The car selection, takes in consideration the latest car used apparently to qualify your "decent" or "perfect" car condition. 
Since you already pull the data on the cars owned... why don't you use that and change the label on that to "recommended" and "owned", because eventually it will show data primarily for other cars that have little use.

It would make that less confuse, and then inside the track, you can qualify by your averages like you already do.

nothing major, just a pet peeve seeing that, lol. 

it damaged your ECU :)

yeah, unfortunately there isn't any situation where it excels. like I mentioned a few posts ago, if you have any patch of dirt, it will slow you up so much that you're better off concentrating on dirt entirely for that race, the time lost on these sections is far easier to win with dirt tires than on tarmac with race tires.
rally cross tires tend to be mediocre on both situations, and in turn... slower.