Hockenheim, Germany
A strong driver recovering from a poor qualifying position isn't automatically a surprise. This model estimates a finish from both pre-race Elo and qualifying, then compares it with what happened.
Expected finish assumes a classified finish. DNFs stay in the full grid but receive no performance verdict. Qualifying is not the starting grid; grid differences are flagged, not silently substituted.
Fit on 230 earlier finishes across 22 races, 1996–1998. No results from this race enter its expectation.
Missing Elo or qualifying stays visible in the full grid. A grid position is not used as a qualifying result.
Expected P2.1 Finished P1
Expected P2.8 Finished P2
Expected P4.4 Finished P4
Expected P4.5 Finished P11
Expected P8.8 Finished P15
Expected P10.8 Finished P16
Up to three per side, ranked by expected minus actual finish. DNFs, missing estimates and gaps rounding to 0.0 are excluded. “Flop” means below the model's estimate—not poor driving. A ranked gap can still be inside the usual error range.
| Driver | Pre-race Elo | Qualifying | Expected | Result | P1 ← position → P21 | vs expected | Reading |
|---|---|---|---|---|---|---|---|
| M. HäkkinenMcLaren | #22378 | P1 | P2.1No error range | P1 | +1.1places | Range unavailable | |
| D. CoulthardMcLaren | #62256 | P2 | P2.8No error range | P2 | +0.8places | Range unavailable | |
| J. VilleneuveWilliams | #42314 | P3 | P3.0No error range | P3 | 0.0places | Range unavailable | |
| D. HillJordan | #122050 | P5 | P4.4No error range | P4 | +0.4places | Range unavailable | |
| M. SchumacherFerrari | #12441 | P9 | P4.9No error range | P5 | −0.1places | Range unavailable | |
| R. SchumacherJordan | #102116 | P4 | P3.8No error range | P6 | −2.2places | Range unavailable | |
| G. FisichellaBenetton | #72202 | P8 | P5.1No error range | P7 | −1.9places | Range unavailable | |
| E. IrvineFerrari | #32376 | P6 | P4.0No error range | P8 | −4.0places | Range unavailable | |
| H. FrentzenWilliams | #152030 | P10 | P6.5No error range | P9 | −2.5places | Range unavailable | |
| J. AlesiSauber | #82123 | P11 | P6.3No error range | P10 | −3.7places | Range unavailable | |
| A. WurzBenetton | #52261 | P7 | P4.5No error range | P11 | −6.5places | Range unavailable | |
| J. TrulliProst | #172018 | P14 | P8.1No error range | P12 | −3.9places | Range unavailable | |
| T. TakagiTyrrell | #211948 | P15 | P8.8No error range | P13 | −4.2places | Range unavailable | |
| M. SaloArrows | #201949 | P17 | P9.4No error range | P14 | −4.6places | Range unavailable | |
| O. PanisProst | #162028 | P16 | P8.8No error range | P15 | −6.2places | Range unavailable | |
| E. TueroMinardi | #191963 | P21 | P10.8No error range | P16 | −5.2places | Range unavailable | |
| J. HerbertSauber | #92117 | P12 | P6.7No error range | DNFGearbox | — | Not assessedDNF: Gearbox | |
| S. NakanoMinardi | #112106 | P20 | P9.8No error range | DNFGearbox | — | Not assessedDNF: Gearbox | |
| R. BarrichelloStewart | #142035 | P13 | P7.5No error range | DNFGearbox | — | Not assessedDNF: Gearbox | |
| J. VerstappenStewart | #132045 | P19 | P9.6No error range | DNFGearbox | — | Not assessedDNF: Gearbox | |
| P. DinizArrows | #182006 | P18 | P9.7No error range | DNFThrottle | — | Not assessedDNF: Throttle |
Average absolute error on 21 classified finishes in 2 earlier races within the last ten years. Each race was held out and predicted using only older races; all three models use the same samples.
The range scales the 80th percentile of earlier held-out absolute errors to this field size, rounding outward. It is an empirical error guide, not a guaranteed 80% chance for a particular driver. We require at least five checked races and 100 checked finishes before showing it.
Inputs are field-normalised Elo rank and qualifying position. The target is position among all race entries, trained only on classified finishers. Estimates are averages, not a unique predicted finishing order. Penalties, the circuit, weather, incidents and strategy aren't model inputs. Historical qualifying coverage is patchy; this is an exploratory backtest on corrected records, not proof of live forecasting accuracy.
These are actual rating changes. The expectation model above does not change Elo or fantasy scoring.
Top five in the stored standings. Points include sprint results where applicable.
Movement since round 10. A dash means the earlier standing is unavailable.
Explore the season →Two-driver constructor entries from this race. A result comparison, not an adjustment for strategy or reliability.
1 place apart at the finish.
Q1 · Häkkinen 0.509s faster
6 places apart at the finish.
Q1 · Villeneuve 1.102s faster
2 places apart at the finish.
Q1 · Schumacher 0.189s faster
3 places apart at the finish.
Q1 · Irvine 0.189s faster
4 places apart at the finish.
Q1 · Wurz 0.028s faster
Non-finish recorded; no finishing gap comparison.
Q1 · Alesi 0.936s faster
3 places apart at the finish.
Q1 · Trulli 0.353s faster
Non-finish recorded; no finishing gap comparison.
Q1 · Salo 0.312s faster
Non-finish recorded; no finishing gap comparison.
Q1 · Nakano 0.552s faster
Non-finish recorded; no finishing gap comparison.
Q1 · Barrichello 0.847s faster
This debrief is calculated from stored results and rating snapshots. The experimental expectation model learns from earlier races only, using Elo rank and qualifying together; it never feeds back into Elo. It is not a separately reviewed race report.
Elo uses the result order, including retirements. It does not isolate driver skill from the car, reliability, penalties or strategy. Elo seed comparisons use pre-race ratings, never today's ratings. Equal ratings share the midpoint of their rank range (for example, tied first and second become seed 1.5).
Grid gains use numbered starting slots and classified finishes only. Zero or missing grid values are excluded; a position gained is not necessarily an overtake. Qualifying gaps use the latest segment both teammates have a valid time for, never Q1 against Q3.
Lap-by-lap pace, tyre stints, pit-stop execution and safety-car effects are not available in this report. Missing data is omitted, not inferred.
Read the Elo methodology →