TMX Replay Investigation

A report by donadigo & Wirtual
Part II — other players

This part provides analysis of other players that were figurating in the top 10 of both, United and Nations leaderboards. View part I.

Note: not all players presented on this list show suspicious behaviour. They are included because we analyse all entries from the top 10.

roman9898
Roman’s inputs are extremely frequent and very unnatural, compared to any run covered in section 3.3. In the presented run, roman achieved a maximum of 25 spikes per second, and his average totals 10. However, given the amount of replays uploaded by this player, we are unable to determine if they used external programs to achieve this run.

twstw
twstw’s inputs greatly differ from the norm. The footage shows that twstw plays very differently, tapping at extremely high speeds. However, this playstyle is possible by setting the analog sensitivity in-game to very high values. Setting the controller sensitivity to 2.5 ingame, we were able to achieve this behaviour. Therefore, we argue this player did not cheat.

jea
Even though Roa’s replay reached a high peak of 26 spikes/second and an average of 8.52, the replay itself showcases legitimate stick movement. The scores only demonstrate very frequent spiking in the small range. We argue however, that the effect is caused by the device Roa plays with and is not in any way related to unfair play. We argue this is one of the false positives that the script identified and this behaviour is also present in other player’s replays.

arti_show
All Arti’s most suspicious replays are done on one United map — Bay A4. This replay is one of them showcasing a time of 8.87. During the run, Arti achieves a maximum spikes/second of 16 and an average of 7.62 spikes/second. His inputs feature unnatural movement and unrealistic line adjustments. His runs on other Stadium maps do not contain such behaviour at all, nor do they reach the same numerical values. As an example, Arti’s Stadium C3 run has a maximum of 5 spikes/second and 1.29 average. Therefore we argue that Arti’s Bay A4 run was done with external tools to slow-down the game.

overninja
This is the only suspicious run done by overninja. In the run we see quite the same behaviour as in Arti’s and riolu’s replays. However, given the amount of replays uploaded by this player, we are unable to determine if they used external programs to achieve this run.

alzimo.rtg
alexx’s runs are another example of frequent spikes that we believe happen on hardware level. In the clip, the replay features similar behaviour to the one of Roa’s replays but the movement dynamics are completely natural. In this replay, alexx reached a maximum of 16 spikes/second and an average of 4.15 spikes/second. We also argue this is one of the false positives, that was ranked high due to usage of specific controllers/joysticks.

grievous97
Lanz’s run on TMF Bay A5 contains some of the fastest stick movement we have seen in this dataset. Lanz reached a maximum of 26 spikes/second and had a 13.19 average. We argue such an average and the movement presented is simply not achievable with normal hardware. Even though Lanz’s suspicious runs exist only on TMF Bay A5, we argue that Lanz used external tools to achieve them, at least on this map.

Lanz recently reached out to us admitting his mistakes and playing unfairly. He used Cheat Engine to set his records. Lanz said that he used to play Bay and Snow environments at 40% of the game speed.

We argue this should be taken into account when judging this player.

_00fri_fffco_f00u71
Ricou’s inputs look completely natural in the presented example and in other replays as well. The reason that the script ranked this replay high is because of a small number of uploaded replays by this player and also because his joystick seems to generate noisy analog output, which translates to high spikes/second. Therefore we argue ricou did not use any external tools to drive his runs.

aymeric_98
Many Neko’s Rally runs feature behaviour similar to one exhibited by Techno, often reaching a peak of 17 spikes/second. We argue it’s not practical to play Rally with this kind of tapping and as such we think that Neko used external tools to slow down the game. Check out other runs done by Neko in the spreadsheet.

Neko recently reached out to us admitting his mistakes and playing unfairly. He used Cheat Engine to set his records. We argue this should be taken into account when judging this player.

vinsgorb
Presented run achieved a peak of 15 spikes/second with 4.86 being the average. We observe again that Vins inputs extremely short bursts of steering and no smooth steering is present in Vins’ replays. He have presented four runs from this player but we detected 14 more runs with a higher peak than 15 spikes/second. We believe further analysis of these replays is redundant as the 4 runs presented are enough to argue that they slowed down the game. The remaining records will be linked in the spreadsheet. Check out other runs done by this player in the spreadsheet.

Sunspinx
This is the only suspicious replay by Sunspinx, which is the same situation as Lanz, with the difference that this replay was driven on Bay A4. The replay achieves a peak of 23 spikes/second and 11.62 average. However, given the amount of replays uploaded by this player, we are unable to determine if they used external programs to achieve this run.

william_h_bonney
William’s records on B12-Race, B09-Acrobatic and A02-Race exhibit the same behaviour as the replays from Techno. They feature extremely quick spikes on both sides of the steering range and do not make sense given the tracks the replays were driven on. We also argue that they cheated in Official mode on at least 17 tracks, but we think further analysis is not necessary to conclude they slowed down the game. These 17 replays are available in the spreadsheet.

We have also discovered players outside of the 10 largest spike/second deviations that we argue have cheated. These can be found in the spreadsheet. Note however, that the “other players” section within the spreadsheet shows players who are not suspicious, but are still analysed because of their appearance in the top 10.