The methodology we use
Every product follows the same path: catalogued from the manufacturer’s specification, checked against a published source we link to, then scored on six criteria calculated from those specifications with a formula printed on this site. Where we have measured or played a product ourselves, that number replaces the calculated one and the page says which is which.
A product enters the database
Frames, strings, balls and grips sold in Europe are added from the manufacturer’s own specification. Nothing is published before its specification is entered — an empty field stays empty on the page rather than being filled with an estimate.
The specification is checked against a source
Head size, unstrung weight, balance, string pattern and, where published, stiffness and swing weight are checked against a source we can link to — the manufacturer’s page or a European retailer’s listing. The source is stored with the product, and the page shows whether the check has been done.
The six criteria are calculated
Power, control, spin, comfort, stability and manoeuvrability are calculated from those verified figures using the formula published on the How we score page. The same specification always produces the same score, and every criterion on a product page names the figures it was calculated from, so a reader can check the arithmetic.
Missing inputs stay missing
Comfort is calculated from frame stiffness, so a racquet with no RA figure on file carries no comfort score — the page says the stiffness is not on file and the overall score is worked out from the five criteria that are real. We would rather show five honest numbers than six with a guess among them.
Measurement and play override the formula
When a product is measured on our bench or hit with by a reviewer, the figure recorded replaces the calculated one for that criterion and is marked as measured. A calculated score is never presented as a test result, and a test result is never overwritten by the formula.
Corrections stay visible
Any reader correction with a source is checked and, if it holds, published as a dated entry on the product page. Because the six criteria follow from the specification, correcting a specification recalculates the scores — and the correction says so.
How the calculator reaches a tension
The tension calculator is arithmetic, not a test result. It starts from a figure the maker published for that frame, applies corrections we print on the page as it applies them, and ends with a number tied to a named string — because a tension without a string is half an answer. Nothing below is measured by us unless it says so.
- It starts from the maker’s own printed range Where the manufacturer publishes a recommended range for a frame, the middle of THAT frame’s range is the starting point — 154 of the 259 frames on file carry one. For the rest the starting point is 23.5 kg, which is the median of those printed middles: our own figure, counted from this catalogue, not a number borrowed from anywhere. The page says which of the two a proposal used.
- The string family moves the number before anything else A printed range assumes a nylon-based string. A polyester is the stiffest material in the shop, so two kilos come off before any other correction; a hybrid takes half of that, natural gut carries half a kilo more, and multifilament and synthetic gut leave the range as it stands. This is the correction most club players miss, and it is why the calculator will not quote a tension without naming the string it is for.
- Then the frame’s own numbers A dense 18×20 pattern is stiff on its own, so half a kilo comes off. A large head hands out free depth and can take up to a kilo more; a small one is tight already and gives half a kilo back. Frame stiffness moves it again where we hold an RA figure — 91 of 259 frames — and where we do not, no stiffness correction is applied and the page says the figure is not on file rather than assuming one.
- Then how you play, and the comfort setting Each playing style carries its own adjustment, set as a configuration row rather than buried in code. The arm-sensitive setting is separate and is applied as well as the style, never instead of it: it lowers the proposal and moves the recommendation to a softer string. A proposal with it marked always comes back below the same frame and style without it — a check runs across the whole catalogue and fails if it ever does not.
- Floors, and when the printed range is the wrong advice A polyester proposal landing under the maker’s printed minimum is correct rather than broken, because the printed range is advice for nylon: the engine allows 2.0 kg below it. With the arm-sensitive setting marked that limit is removed entirely — an elbow that is already complaining does not care what the sticker says. Under everything sits 18.0 kg, below which a bed stops behaving predictably and the strings begin to move on their own.
- The string is scored, not picked Every string with a specification on file — 120 today — is scored against the frame on family, shape, gauge and price, and the top of that list is what the number is for. The page says how many were scored and gives the reasons that earned each card its place.
- Two setups, and where the knots come from The proposal names two ways to string the frame: the whole bed in one string, and a hybrid with that string in the mains and a softer one in the crosses half a kilo higher. Each carries its own tension, because the family correction above applies to each part separately. The knot counts are arithmetic, not a measurement: a knot is tied at each end of each piece of string, so one piece is two knots and a hybrid — two different strings, so two pieces — is four. Some frames and some stringers take two pieces either way, and then it is four; we hold no grommet layouts, so we say that rather than pretending to know.
- Nothing is proposed that we cannot name If the catalogue holds no softer string suited to a frame, the hybrid setup is left out rather than described with a blank in it. If a frame has no specification, it is not in the calculator at all. 0 of the 120 strings on file carry a tension loss we have recorded ourselves, and that figure is the only loss number printed anywhere on the tool.
What the calculator does not claim
The rules we hold to
Scores follow from these measurements. The weighting behind each number is set out separately.
How we score →