Race For You Difficulty

What the score means

Race For You Difficulty is an objective comparative course-demand score. It helps distinguish runnable, lower-climb events from longer, steeper, more technical, accumulated or exposed challenges. It is not a safety rating, finish probability, coaching recommendation or judgement about whether you should enter.

The final result is clamped to 1–10, rounded to one decimal place and then assigned to the band shown for that displayed score.

The five weighted factors

The five weighted factors form the intrinsic score. Difficulty V2's independent time-limit modifier can only add 0, 0.15, 0.35, 0.55 or 0.70 afterwards and never subtracts from the course's inherent demand. Long continuous races also retain a distance-led intrinsic floor, so a broad time allowance cannot make an expedition look ordinary.

This frozen Difficulty V2 modifier is not the public Cutoff Pressure V2 band below and does not use a personalised finish estimate.

Environmental severity is used only where edition-specific source evidence exists. Weather, surface conditions, route changes and event delivery can vary by year and on race day, so the score is not a prediction of the conditions you will encounter.

Technicality and environmental evidence

Technicality is a Race For You editorial assessment of underfoot and movement complexity. Checked course evidence may include surface and trail quality, steepness, rock, bog, trackless ground, scrambling and route-finding where those directly affect movement. Distance, total ascent, cutoff, price, prestige and the resulting Difficulty score do not determine technicality.

Intermediate decimals sit between these anchors. Technicality remains Unknown when checked evidence cannot support an assessment. It is currently stored as a family-level editorial value rather than a separately versioned, field-level source record, so a material route change must be reviewed rather than assumed to inherit the previous assessment. Navigation, support, night running, mandatory kit and environmental burden are scored separately to avoid counting the same demand twice.

Environmental severity is edition-specific. A known assessment requires a source, verification date and explanatory note; it must not be inferred from an event name, calendar month or dramatic marketing language. Evidence factors identify the nature of the assessment but do not each add points: the single severity value forms one-fifth of the Exposure and support factor.

Environmental evidence confidence changes score-confidence coverage, not the severity value itself. Race-day weather and surface conditions can still differ materially from the published event design.

Unknown and editorial data

Missing elevation, cutoff or price stays visibly unknown. When the score needs an unavailable optional factor, it uses a neutral 5.5 rather than pretending the fact is easy, hard or zero. Technicality and beginner suitability are Race For You editorial assessments; dates, distances, ascent, cutoffs and prices are source-checked facts where available.

Cutoff Pressure V2

Cutoff pressure compares the published time limit with distance, ascent and verified course demands. Lower pressure does not mean the race is easy. It is separate from overall Difficulty V2 and from the personalised Will I Make the Cutoff prediction.

The model first calculates course-equivalent kilometres as distance plus one kilometre for each 100m of ascent. It then applies a bounded 0.85–1.30 course-demand multiplier. Source-backed technicality contributes (score − 5.5) × 0.018; night running contributes −0.012 / +0.012 / +0.040 for none / partial / significant; navigation contributes −0.012 / 0 / +0.018 / +0.035 for fully marked / partially marked / GPS required / self-navigation; support contributes −0.012 / 0 / +0.015 / +0.035 for fully supported / aid stations / semi-supported / self-supported; and evidenced environment contributes −0.008 / +0.012 / +0.030 / +0.050 for ordinary / elevated / severe / extreme. Unknown optional evidence contributes zero adjustment, which is neutral rather than easy.

Continuous distance burden adds up to 0.040 from 80km onward. Multi-stage burden adds 0.008 per additional stored stage, capped at 0.040. When every stored stage has a published distance and active-time cutoff, the tightest raw stage allowance governs—even if there is no single overall cutoff. Overnight recovery is excluded. The governing allowance is rounded to two decimals, then assigned to a fixed band:

When ascent is not verified, the calculation uses a neutral reference density of 20m/km. The public ascent remains Unknown. A race is Unknown only when neither a usable overall cutoff nor a complete set of stage distance/cutoff pairs is available; an organiser-confirmed absence of a fixed overall limit is No fixed limit. Variable-total non-standard formats are Not applicable and are not forced through the conventional model.

Evidence coverage totals 100 points: elevation provenance contributes up to 25, format up to 20, technicality 15, cutoff evidence 10, and navigation, support, night, kit and environment up to 6 each. High confidence means 80–100, Moderate 55–79 and Limited below 55. Approximate or conflicting ascent and missing evidence reduce the total. Confidence describes score-input coverage, not the runner's chance of finishing.

Read the race data disclaimer or see the latest recorded successful source check on any race page.

Race DNA and Smart Alternatives

Race DNA 1.0.0 is a deterministic race-to-race catalogue comparison. It is not a personal recommendation, safety judgement, finish prediction or organiser ranking. Sponsor relationships, featured placement, Race Pulse activity and member behaviour do not affect its order.

Numeric facts are compared proportionally; terrain and format use shared recorded sets; navigation, support, night running and fixed Cutoff Pressure bands use fixed published orderings. Country, region and place are used only when a person explicitly selects them as filters or when the page states that context. They do not create a hidden proximity score.

A missing, unknown, unverified or conflicting value is excluded rather than changed to zero. The method reports which side is missing evidence and lowers comparison coverage. A pair needs at least 35% comparable weight and two substantive shared dimensions. Approximate ascent and a neutral-ascent Cutoff Pressure calculation remain labelled. Results describe specific similarities and differences; they do not call one race globally easier or harder.

Permanent family identity is the exclusion key, so another edition of the source race cannot be returned as a match. Ultra and Marathon categories never cross. Conventional fixed-distance races never cross with variable-total events. Fixed-time, backyard and vertical events compare only with the same exact format kind, using their recorded event window, loop, interval, base-camp or vertical-objective facts instead of an invented headline distance.

Smart Alternatives first preserves the active category, identity, country, date, distance, format and other explicit filters, then reuses the same Race DNA order. It can respond to sold-out, closed, cancelled, completed, no-next-edition and predictor-incompatible contexts. No filter is relaxed silently, and an empty compatible set is a valid result. Alternatives shown for a cancelled event are independent catalogue comparisons, not organiser-designated replacements.

Equal results are resolved by evidence coverage and then stable family ID. The same inputs and Race DNA version therefore produce the same order.

Source-check freshness

Public freshness starts only from a recorded successful official-source check. An attempted check, page view, data import or deployment is not treated as successful verification. Fresh, Due and Overdue describe the configured checking cadence for that edition context; they do not assert that every event fact is unchanged or wrong.

Dates TBC, near-term races, distant races, completed editions without successors and cancelled editions can have different review cadences. Recorded current-source conflicts remain visible even when the successful-check cadence is current. The organiser remains the final authority before entry, travel or race-day decisions.

Personal cutoff predictor

The cutoff predictor is a separate personalised heuristic with separate inputs and versioning. Difficulty Model V2 does not change its coefficients; predictor continuity retains its frozen V1 difficulty baseline where a personalised difficulty comparison is needed.

The predictor starts from one exact completed edition—or up to five usable saved results. Each result is translated independently: both courses become an effort-distance that includes climbing, conservative ultra-distance scaling is applied, then the estimate adjusts for terrain, technicality, race format, navigation, support and night exposure.

With exactly one comparable result, the existing one-result model is used unchanged. With several results, course relevance, evidence quality and agreement determine influence. The output remains a finish range, not an assertion that several saved races guarantee a more accurate answer.

Multi-stage comparisons use the official combined stage time, excluding overnight rest. Translating a staged result to a continuous target receives a conservative 10% format penalty; translating a continuous result to a staged target receives only a 4% recovery allowance. The range is widened because aggregate distance and ascent cannot model individual stage cutoffs, stage layout or day-to-day recovery.

“Estimate confidence” describes the relevance, quality and agreement of the available evidence—not your chance of finishing. “Cutoff outlook” only states whether the estimated range falls inside, crosses or exceeds the currently stored cutoff. An organiser-confirmed no-limit race is labelled “No fixed limit”, with cutoff outlook and buffer marked not applicable; a cutoff that was not verified remains “Unknown”. The output is always a range and is never a safety or medical guarantee.

Read the prediction disclaimer.

How saved results are weighted

Course relevance is target-specific and never uses finish time. It asks whether a previous course resembles the target across these dimensions:

Unknown optional fields are omitted and the remaining known weights are renormalised, so unknown never becomes a false perfect match or a zero. Distance, ascent and format remain required because a finish cannot be translated honestly without those core course facts.

Evidence quality is assessed separately from course relevance. Approximate or conflicting ascent, incomplete multi-stage evidence and manually entered course facts can reduce influence without pretending the course itself is less comparable. Perceived effort and unusual race-day conditions only reduce influence slightly; they never rewrite the finish time or guess how much faster the runner “should” have been.

Robust multi-result estimate

Every usable result produces an independent target estimate. Several estimates are combined in the logarithmic time domain, which handles proportional differences more consistently than a naive average. A relevance-and-quality-weighted median provides a robust anchor, and bounded differences allow aligned results to move the final centre without giving weak evidence an artificial minimum influence.

With at least three results, an estimate more than 35% from the robust anchor is treated as discordant. Its influence is capped rather than silently discarded, and the final range widens when the contributing estimates disagree. Agreement, effective evidence and average course relevance all affect estimate confidence.

A staged-versus-continuous comparison is capped at 50% course relevance and retains the conservative format adjustment. Two staged races can be compared using published stage loads, but the model still uses combined active stage time and cannot assess sleep, overnight recovery, cumulative fatigue or each stage cutoff.

This remains a transparent, non-scientific performance heuristic. Training, health, weather, nutrition, navigation and race-day decisions can move the real result outside the estimate.

Runner evidence profile

Runners can save up to five completed results on their own device. Catalogue results store the exact historical edition ID and finish time; course facts are resolved again from the maintained edition record. A manual result stores only the course facts the runner entered. A missing or removed edition is never replaced with another year.

The profile describes evidence across long-distance completion, climbing exposure, technical terrain, runnable-course pace, night running and multi-stage experience. Each trait names the result behind its strongest signal and reports how many stored results contain the facts needed for that trait. Missing ascent, format or night data is excluded from that trait rather than treated as zero.

Developing, moderate and strong are deterministic evidence bands—not physiological ratings. Profile confidence reflects only the number and completeness of saved results. Optional perceived effort and race-day conditions do not change this target-agnostic profile; in saved-history prediction they only reduce a result's influence slightly and never directionally adjust its finish.

The profile and target prediction answer different questions. The profile summarises the experience present in all saved facts, while the predictor weights only the results sufficiently comparable with the selected target.

Find My Race matches

Match percentages are a weighted compatibility score: distance 22%, travel 12%, season 8%, climbing 11%, terrain 11%, budget 8%, experience 12%, main priority 10% and cutoff preference 6%. Every result exposes its factor fit, factor weight, positive reasons and material cautions. If Don’t mind is chosen for main priority, that factor is omitted and the remaining known weights are renormalised to 100%, so neutral priority adds neither a bonus nor a penalty.

A high match means “fits these answers better within the current catalogue,” not “you will finish” or “this is objectively the best race.”

Versioning

Race For You Difficulty is Model V2. Cutoff Pressure is Model V2. Race DNA is 1.0.0 and Smart Alternatives reuses that same ordering. The personal cutoff predictor remains v1.3 and Find My Race remains v1 with the V4 neutral-priority refinement. Meaningful formula changes must be versioned and documented rather than silently rewriting historical comparisons. Last reviewed 5 September 2026.

These tools are not scientifically validated and cannot account for your health, training, experience, equipment, navigation decisions or changing race-day conditions. Read the organiser's current rules and mandatory kit, make your own safety decisions and seek qualified medical or coaching advice where appropriate.