Audit Review · Season 78, day 72 of 90 · Dodge
We told Dodge to do three things. He did most of two of them and went from 856 points to 1,204 with eighteen race days still to run — past the best case the page allowed for. This marks every claim we made against what actually happened, including the ones that were wrong.
The page under review: Dodge — Team Audit & Path to No 1, published season 78 day 12, revised day 20.
The advice and the following of it are different things. Both are marked here, and the advice does not come out of it as well as its own headline number suggests.
Everything it measured was right, and acting on that produced the season. What it modelled mostly was not. And it was not executable as written: it asked Gamarra to lead 28 races while admitting ~25 was his ceiling, without counting the three tours it also endorsed — which took 18 of his 28 starts. The bottleneck it identified, it then over-subscribed threefold.
He was handed a plan that could not all be done and triaged it well. The time-trial rule was executed beyond what was asked, 5 of 18 to 13 of 15. Gamarra's ten discretionary rides — all the tours left him — returned 290 of his 430 points with two wins. What keeps it off an A: Sarajevo, where Gamarra was free and six riders went without him; the richest race type on the chart never entered; and the Ugly section untouched.
The audit gave three cases: a floor of 790, a central case of 990, and a stretch case of 1,200 it said would need a training focus Dodge never applied. He is at 1,204 on day 72.
On 327 starts against 319 for a full previous season, so this is not extra volume — the per-start return went from 2.68 to 3.68. Wins doubled, 3 → 6. Top-nine finishes, 18 → 28.
Against 168 of 319 in S77. He stopped scattering entries across bands and raced where he belonged. That concentration is most of the story, and it is what the promotion rewarded.
Dodge is the only one of the four going up. Every rival is level or falling, so the two places gained are partly earned and partly handed over.
The audit's fix was three moves, all free. Each is marked against what happened.
The whole plan, and worth more than the page claimed. Division-3 time-trial entries went from 5 of 18 to 13 of 15. Starts 30 → 78, points 80 → 544, per start 2.67 → 6.97, top-nine finishes 1 → 13, wins 0 → 2. The page predicted +340 across the season and warned the rate would soften as he entered weaker-fitting races. It rose.
Predicted +340 · delivered +464 by day 72Every named race he actually rode paid. Firenze–Fiesole, predicted 3rd — won. Dresden–Plauen, predicted 7th — won. Ochjor, predicted 5th — 2nd. That is 280 points from three races the page named by title, out of his season total of 430. He skipped Brescia Uphill as instructed. But he did not enter Fill Fill, Moustier, Seven Spices, Tarbela or Hooglede; at Sarajevo History Hillers he entered six riders without Gamarra, the best of them 36th; and Lyon GP he could not have entered at all — see below.
Predicted +150 · the three races he did take returned 280The claim was that running every time trial at Risky costs the best riders, and Steady should be used for anyone projecting top 10–15. The spread went the wrong way: podiums rose 4 → 7, but finishes outside 40th went from 49% of starts to 59%, and the mean placing worsened from 43.5 to 55.0. Both tails fattened. That is what Risky-everywhere looks like — but it is also what entering far more weak-fit time trials looks like, and the two cannot be separated from results. The rule was written so it cannot be marked. That is our fault, not Dodge's.
It also assigned on an untested ordering. The move named Soisalo, Arends and Texier as "your best three" to protect with Steady and said to keep Risky for Florimo — who went on to top the four on points. That is not proof the order was wrong, since the four rode different programmes (see below), but it is a rule that hands the high-variance setting to a named rider on a ranking nothing had yet verified. A composite used to describe is a curiosity; one used to assign is a liability.
Check the signup screen — and re-order the riders before re-issuing itWithdrawing this was the best thing the process did. Xia Jinxing raced twelve times in season 78 for zero points, exactly as the withdrawal predicted. Had the training camp been spent on him it would have been wasted.
Correctly withdrawn · Xia: 12 starts, 0 points| Race | Sign day | Race day | We said | What happened |
|---|---|---|---|---|
| Firenze – Fiesole | 62 | 70 | Enter | Gamarra won it — 100 pts |
| Brescia Uphill | 48 | 56 | Skip | Skipped, as advised |
| Sarajevo History Hillers | 46 | 54 | Enter | Entered without Gamarra — best 36th |
| Fill Fill – Nefasit | 32 | 40 | Enter | Not entered |
| Boat of Garten | 80 | 88 | Skip | Has not run yet |
| Lyon GP | 3 | 11 | Enter | Impossible — signup closed on day 3, we published on day 12 |
Of the four races Move 2 told him to enter, one had already been run, two he skipped, one he entered without the rider the move was about — and the one he executed as written, he won. Of the two it told him to skip, he skipped the one that has come up. That is one clean execution out of five actionable instructions, and the clean one was worth 100 points.
Four claims about what Dodge already owned. Every one of them was tagged Modelled — and this is where the page falls down.
This review claimed the four riders rode "exactly nine time trials each — as clean a comparison as this game offers", and concluded our ranking came out backwards. That was wrong, and Zee caught it. They rode nine each but not the same nine: Florimo rode nine single-day time trials and no tour stages at all, while the other three rode four tour stages and five singles apiece. Different programmes, different fields. Compared on single-day races alone, Florimo and Soisalo are level.
| Rider | Our rank | Single-day TTs | Pts/start | Tour-stage TTs | Pts/start |
|---|---|---|---|---|---|
| Nicholas Florimo | 362nd | 9 | 20.9 | 0 | — |
| Leo Soisalo | 32nd | 5 | 20.0 | 4 | 14.5 |
| Leonard Texier | 141st | 5 | 6.0 | 4 | 37.0 |
| Wilfried Arends | 99th | 5 | 0.0 | 4 | 27.5 |
On single-day races Florimo and Soisalo are level, 20.9 against 20.0 per start — and Soisalo was the rider the audit named best. On tour stages the order is different again, with Texier first. There is no consistent ordering across the two programmes, which means the season did not test the composite at all. The honest verdict on the time-trial ranking is not "inverted" but "still unproven" — and the raw-versus-adjusted argument this review built on top of it does not stand either. What survives is the original claim: it is a usable group, and all four scored.
The page named five riders at hill grade — Ruey 116, Woering 111, Gamarra 109, Atila 108, Jans 101 — by benchmarking them against the average hh of division-3 top-ten finishers. That compares support riders against race winners. In a road race only the captain scores, so a helper returning nothing is expected, not a failure — which makes the benchmark meaningless in both directions. What actually happened: Gamarra 350 points from 10 hill starts, Ruey 60 from 8, and Woering, Atila and Jans nothing at all from 27 starts between them.
The claim was that he rates top-9 in 23 of 64 hill, climb and all-rounder races, a 36% scoring rate, and that at 29 he was at his peak. He went 9 top-nine finishes from 22 non-time-trial starts, 41%, with two wins, and finished the season as the team's top scorer by a factor of 2.4. This is the one modelled claim that landed clean.
Xia: 12 starts, no points. The withdrawal note said he would rank last or near-last in every remaining division-3 downhill race. It was correct, and it was published before Dodge acted, so it cost nothing.
Every finding in The Bad was tagged Observed — measured from Dodge's own results rather than modelled. Re-measured on the same definitions, the record is five out of five.
The single most valuable sentence on the audit. He now enters almost all of them: starts 30 → 78, points 80 → 544, per start 2.67 → 6.97, wins 0 → 2. Everything else on this page is a footnote to this one line.
HL went from nothing to 100 points, SP to 74, CB to 10, and ALL was dropped entirely. But CLDH is still 24 starts for nothing, and with 17 untyped and 6 CLHL that is 47 starts returning zero — of which 18 are compulsory tour stages, leaving 29 he actually chose. The finding held; the fix is half done.
The diagnosis was right and is the best thing the page did: pointed at his best races, Gamarra won two and took a second. But the specific races this finding named never got a fair test — Lyon GP had already run, Fill Fill was skipped, Sarajevo was entered without him. Only Brescia Uphill, the one it said to avoid, was followed. The claim stands on other evidence than the evidence it offered.
Finishes in 10th–13th fell from 15 to 9 while top-nine finishes rose from 18 to 28. The cheapest points on the page, and he took them.
Ruey 76 → 86, Brigham 76 → 92, Degn 76 → 85, Xia 76 → 84. Only Vandommele is still sitting on 76. Xia's recovery did not help — he raced twelve times for nothing — but that is the withdrawn downhill call's fault, not this one's.
| Rider | S77 | S78 | Verdict |
|---|---|---|---|
| Jochen Ruey | DP 76 | DP 86 | Recovered |
| Jin Brigham | DP 76 | DP 92 | Recovered |
| Alan Degn | DP 76 | DP 85 | Recovered |
| Xia Jinxing | DP 76 | DP 84 | Recovered |
| George Vandommele | DP 76 | DP 76 | Unchanged |
| Ian Defley (94) | 0 starts | 0 starts | Still on roster |
| Fenton Sommers (90) | 0 starts | 0 starts | Still on roster |
| Mehdi Lamy | on roster | 15 starts, 1 pt | Still on roster |
The Ugly was ignored in full. Defley and Sommers are ninety-four and ninety, have never been entered in a race, and are still occupying two of twenty-two squad places — now in a harder division.
The audit's central chart, re-measured. It is the clearest single picture of what changed: the two types it called out as paying went up, the two it called dead went in opposite directions, and the richest type on the whole chart was abandoned entirely.
| Type | Published | S78 | S78 starts | S78 pts | Change |
|---|---|---|---|---|---|
| DH | 5.83 | 11.67 | 6 | 70 | Doubled — 1 win |
| TT | 6.79 | 6.97 | 78 | 544 | Held, at 2.6× the volume |
| HLSP | 3.06 | 5.71 | 42 | 240 | Nearly doubled |
| HL | 0.00 | 4.17 | 24 | 100 | From nothing |
| CL | 2.00 | 3.33 | 18 | 60 | Up |
| SP | 0.00 | 1.85 | 40 | 74 | Scoring — 30 of 40 compulsory |
| FRSP | — | 0.50 | 12 | 6 | All 12 compulsory |
| CB | 0.00 | 0.42 | 24 | 10 | Barely moved |
| HLCL | 1.63 | 0.42 | 24 | 10 | Collapsed |
| untyped | 4.60 | 0.00 | 17 | 0 | Collapsed to nothing |
| CLDH | 0.00 | 0.00 | 24 | 0 | Unchanged — 18 compulsory |
| CLHL | — | 0.00 | 6 | 0 | New, and dead |
| ODD | 10.00 | — | 0 | 0 | Not entered once |
| ALL | 0.00 | — | 0 | 0 | Dropped, correctly |
The ODD row is the expensive one. It sat at the top of the audit's chart at 10.00 points per start, the richest type on it — and Dodge entered neither of the two that ran. One of them was Seven Spices of Halmahera, which the calendar predicted him third in. The chart was printed; the row was ignored.
You sign up for a tour, not for its stages — every stage then runs whether it suits you or not. That covers 21 of Dodge's 57 races and 126 of his 327 starts, and it changes how the wasted-start count should be read, including in the first version of this review.
| Tour | Days | Stages | Starts | Points | Per start |
|---|---|---|---|---|---|
| Perm Tour | 47–48 | 4 | 24 | 258 | 10.75 |
| Norway Tour | 41 | 9 | 54 | 158 | 2.93 |
| Copa de la Historia de los Andes | 5–6 | 3 | 18 | 70 | 3.89 |
| Kangaroo Tour | 59–61 | 5 | 30 | 40 | 1.33 |
| All tours | — | 21 | 126 | 526 | 4.17 |
| Everything else (single races) | — | 36 | 201 | 678 | 3.37 |
Tours pay better than single races — 4.17 points per start against 3.37 — so the compulsory block is not a drag. But the stage results inside it should never be read as entry decisions. Three of the poorest-looking rows on Dodge's season (30 of 40 SP starts, all 12 FRSP, 18 of 24 CLDH) are Norway and Kangaroo stages he had no way to decline. Judge him on the tour he signed for, not on the stage he was dragged to.
The audit printed a 43-race calendar. Twenty-six have now signed and run — the rest fall after day 72 or belong to next season. Seventeen were entered, but that number flatters him: seven were tour stages that came free with three signups. Three more of the nineteen singles fell on days a tour already had twelve riders out. Counted as decisions genuinely open to him, it is 10 of 16.
| Race | Day | Pred. | Best actual | Pts | What happened |
|---|---|---|---|---|---|
| Firenze – Fiesole | 70 | 3 | 1 | 100 | Gamarra won it |
| Dresden – Plauen | 29 | 7 | 1 | 100 | Gamarra won it |
| TT Zyte | 59 | 7 | 4 | 50 | Florimo 4th — none of the three named riders went |
| Mackay Time Trial | 39 | 8 | 5 | 40 | Florimo 5th |
| Salamanca ITT | 64 | 7 | 6 | 30 | Texier 6th |
| Confronto no Largo do Senado | 65 | 7 | 8 | 10 | Tamas 8th, as called |
| Kesch Peak Climb | 44 | 11 | 11 | 0 | Florimo 11th — prediction exact, still outside the money |
| Fort William ITT | 49 | 4 | 10 | 0 | Entered without Soisalo, Arends or Gamarra |
| Sarajevo History Hillers | 54 | 4 | 36 | 0 | Entered without Gamarra |
| D´aguilar Downhill | 54 | 7 | 34 | 0 | Entered without Soisalo |
| Perm Tour · day 47–48 · 2 planned stages · one signup | 208 | both planned stages paid | |||
| Perm – Perm (ITT) | 48 | 3 | 2 | 128 | Texier 2nd, Arends 4th, Soisalo 9th |
| Ochjor – Ochjor | 47 | 5 | 2 | 80 | Gamarra 2nd |
| Norway Tour · day 41 · 3 planned stages · one signup | 148 | 2 of 3 paid | |||
| Prologue (Norway Tour) | 41 | 1 | 2 | 78 | Soisalo 2nd, Arends 5th, Texier 7th |
| Bergen (ITT) | 41 | 1 | 1 | 70 | Texier won it |
| Bodø-Mosjøen | 41 | 4 | 10 | 0 | Gamarra 10th — we missed by six |
| Kangaroo Tour · day 59–61 · 2 planned stages · one signup | 40 | 1 of 2 paid | |||
| Gold Coast – Gold Coast | 59 | 8 | 3 | 40 | Arends 3rd |
| Sydney – Sydney | 61 | 6 | 15 | 0 | Gamarra 15th — we missed by nine |
| Seven Spices of Halmahera | 48 | 3 | — | — | Not entered — the most expensive skip of the season |
| Moustier GP | 47 | 6 | — | — | Not entered |
| Tarbela Dam Dash | 53 | 5 | — | — | Not entered |
| Hooglede GP | 71 | 5 | — | — | Not entered |
| Fill Fill – Nefasit | 40 | 7 | — | — | Not entered |
| Ovruch – Ovruch | 60 | 8 | — | — | Not entered |
| Tripoli – Tajura | 30 | 9 | — | — | Not entered |
| Tour of Hawaii | 37 | 9 | — | — | Not entered |
| Coustouges ITT | 31 | 10 | — | — | Not entered |
Sixty per cent of the season's points came out of races printed by name on day 12 — still the strongest number on either page. But the split underneath it is the sharper story: 396 of those 726 points came from just three tour signups, and only 330 from the ten single races he picked out one at a time. The plan was right about the tours and he took all three; of the sixteen single races genuinely open to him, he took ten. Of the six he passed up, two — Tarbela and Hooglede — were predicted in the money and carry no excuse; Seven Spices and Moustier fell on a fully committed tour day.
Dodge entered three planned races with six riders but without the rider we named. Right race, wrong squad — as close to a controlled test as this game allows. All three returned nothing.
Mendiluce led them home in 36th; the rest went 54th, 69th, 72nd, 101st and 117th. Gamarra raced nothing that day.
Florimo best in 34th. The named captain sat it out.
We listed Soisalo 4th, Arends 10th, Gamarra 12th. The six who went were Florimo, Lamy, Brigham, Deyhle, Gerrard and Degn. Florimo's 10th was the best of them — one place outside the money, in a race we rated their fourth-best card of the season.
Against that: in the twelve planned races where the named rider did ride, the median gap between predicted and actual place was two places, and six of the twelve beat their prediction. The race and the rider are one instruction.
Seven faults, in order of size. The largest is not a wrong claim at all — it is that the plan asked for more racing than the squad had to give, and never did the arithmetic.
The calendar named Gamarra to lead 28 of its 43 races, and the page itself wrote that "a realistic ceiling is around 25 starts". It never subtracted the tours. Gamarra made 28 starts all season, but 18 of them were compulsory tour stages — from the three tours this same plan endorsed — leaving ten discretionary rides for a calendar that wanted twenty-eight. Every hill race on that list was competing with the others, not just with the tours. A plan that names its own bottleneck and then over-subscribes it threefold is not a plan, it is a wish list, and the triage it forced was left entirely to Dodge.
Move 2 opened with "enter Lyon GP". That race ran on day 11, its signup closed on day 3, and the audit was published on day 12 — so the first instruction in the move was dead before anyone read it. It came from carrying a season-77 observation straight into season-78 advice without checking the calendar. Nothing flagged it at the time, and no earlier pass of this review caught it.
990 was the number; he passed it around day 60 and is on 1,204 with eighteen days left. Understating is the safer direction, but the page presented 1,200 as requiring a training focus that was never applied, and that was simply not true.
Moustier (day 47), Seven Spices (day 48) and Ovruch (day 60) all fall on days Dodge already had twelve riders out in the Perm and Kangaroo tours — tours this same plan told him to ride. The game does not forbid racing both: a rider entered in two races on one day rides both, and nothing here should be read as a scheduling rule. But the manager's account is that "it felt like I could do one or the other most of the time, which led to a bunch of missed signups". The results show the pattern cleanly, and it is not carelessness: on all three of those days the tour took twelve starts, two stages apiece. On day 59 the Kangaroo Tour took only six — and he entered TT Zyte alongside it. Three days where the squad was fully committed, three skips; one day where it was half committed, and he did both. A calendar that lists races by sign day, with no sense of which weeks are already full, invites exactly this. Seven Spices — the skip this page called the most expensive of the season — is one of the three.
Effort is a signup setting that leaves no trace in the results table, and the spread it is meant to narrow is confounded with entering more weak-fit races. A rule that cannot be checked should not have carried a number in the fix list.
Bodø-Mosjøen predicted 4th, finished 10th; Sydney predicted 6th, finished 15th. The page said why — the scorer rates each rider alone and cannot see lead-out or Team Spirit — but the errors are big enough that road predictions should be printed as bands, not places. Time trials, where the rider genuinely is alone, landed within a place or two nearly every time.
Dodge is in division 2 now. The nine skipped races and the remaining fourteen on the audit's calendar were all division-3 races — they cannot be entered any more. What was a recoverable gap is now a closed one, and only the roster items carry forward.
Seven Spices, Moustier, Tarbela, Hooglede and the rest are division-3 races. The promotion that the plan earned is also what put them out of reach. The lesson survives — every named race he rode paid, every one he skipped scored nothing — but the points do not.
17 untyped, 6 CLHL and 6 of the 24 CLDH starts were single races he chose, and they paid nothing. The other 18 CLDH starts came with the tours and are not his to cut.
Ninety-four and ninety, zero starts in either season measured here. Two of twenty-two squad places that cannot be filled — and he has just been promoted into a division where they will cost more.
A season of racing did not test it: the four rode different programmes, so their returns are not comparable. Until it is either verified or rebuilt, stop using it to assign — enter all four and let the results order them. The division-2 plan should carry this caveat.
856 to 1,204 with eighteen days to run, fourth to third, and promoted. The plan's one great call — enter your own division's time trials — carried it, and Dodge executed that one past what was asked. Most of what the plan got wrong, he had to work around unaided.
How this was measured. Season-78 race results through race day 72 of 90 against complete season-77 results, for team Dodge only. "Planned races" are the forty-three rows of the audit's own season-78 calendar; twenty-six have signed and run, and the rest fall after day 72 or, in three cases, run next season. Points are the game's own award points, not a model. Predicted places are the audit's published numbers, unaltered — nothing was re-predicted after the fact. Tour stages are identified from the race table's own stage-to-tour link, and stage results are never read as entry decisions. The time-trial comparison is unusually clean because all four riders made exactly nine starts; the hill comparison is not, because in a road race only the captain scores, which is itself the flaw in the claim being marked. Season 78 is not finished, so every S78 figure will grow: the S77 column is a full season and the S78 column is 80% of one.