Method

Palawizard ranks every racket by a similarity score from 0 to 100 against a reference racket you choose (the default is the Adidas Adipower Multiweight CTRL 3.4). A higher score means closer to your reference across the measured characteristics below. The distance is symmetric on purpose: comparing A to B gives the same number as B to A, so the tool stays a general reference-comparator. Pick a different reference and the whole table re-ranks around it.

Five numeric metrics carry most of the score, plus three categorical ones that nudge it up or down.

• Swingweight (SW): how heavy the racket feels through the swing, and the biggest driver of stability versus manoeuvrability.
• Balance: where the mass sits, in cm from the grip. Higher is more head-heavy.
• Twistweight (TW): resistance to twisting on off-centre hits. Higher is more forgiving.
• Face / screen stiffness: how stiff the hitting face is (Rigidité du tamis). Higher is more direct.
• Frame stiffness (RA): how stiff the frame is. Higher is more direct, with less flex.
• Head shape: classified from the product-image silhouette on a 5-step spectrum (Round → Hybrid R/T → Teardrop → Hybrid T/D → Diamond).
• Core foam: firmness on a 4-step spectrum (Firm → Medium → Soft → Lively).
• Trampoline: rebound or spring-back level, from Very low to High.

The numeric part produces the base score from a weighted, normalised distance. For each metric the two rackets share, the absolute difference is divided by a per-metric scale (the gap that counts as one full step apart), then combined using the weights below:

The base is then scaled by how much data the two rackets actually share. If they share less than 60% of the numeric weight, no score is shown at all. A racket with only a spec or two is an interesting lead, not a false-precise match.

Three categorical adjustments are then added to the base:

• Shape: a compatibility band on the 5-step spectrum. Same or one bucket off is free (+2), two apart costs 2 points, three apart 5, opposite (Round to Diamond) 8. One bucket off being free is deliberate, because shape is a coarse 5-way chop of a continuum and a borderline call shouldn't swing the score.
• Foam: same firmness adds 5, one step apart costs 3, two steps 5, opposite (Firm to Lively) 8.
• Trampoline: 2 points for every level of difference.

The result is clamped to 0 to 100. A racket compared to itself scores 100. A metric only counts when both rackets have a value for it, and missing categorical fields simply skip their adjustment.

The score is also shown as a plain-language band:

• Very close: 80 and up
• Close: 65 to 79
• Comparable: 50 to 64
• Different: below 50

The scales are set to each metric's meaningful spread across the catalogue, and the relevance thresholds are tuned against the reference. The shape term uses a compatibility band rather than a strict matrix because a five-bucket chop over many borderline rackets would otherwise over-penalise a one-bucket judgment call. A borderline shape is shown as a marker (⛑) but is never a scoring input.

One source caveat worth flagging here: PadelShop's social readings report a figure labelled “RA” that behaves like face/screen stiffness, not frame RA. We store and label it that way and keep it display-only, so it does not feed the score.

Measurements are third-party lab and social readings retained with their provenance, and per-sample and per-method variance is real. A score is a heuristic comparison tool, not an authoritative ranking. CSV and PNG exports carry those third-party measurements and product photos. See Sources.

Where a price is shown it is informational, not authoritative, and it never enters the score or the ranking. It is the lowest listing we currently have for that racket, with the retailer and an outbound link, fetched separately from the comparison data so a price refresh never re-ranks anything. Treat every price as a snapshot from when it was captured and confirm the current price on the retailer's own page before buying. Outbound retailer links may become affiliate links in future; that would never influence the score or the ordering.