Chess Caddy / Guides / Club chess ladder

How a club chess ladder works

A ladder is a rating that follows a player from one club night to the next. Here is the maths, the traps, and the decisions to make before you put one on the internet.

Last updated: 20 August 2026 · Rules verified against the current FIDE and US Chess texts on 25 July 2026.

Most club tournaments end the same way: a final standings sheet, a winner, and then nothing. Next month everybody starts from zero again. A ladder is the thing that joins those evenings together — one number per player that goes up when they beat someone better than expected and down when they do not, carried from event to event for as long as they keep turning up.

It is the single feature that turns a series of club nights into a season. This guide explains how the arithmetic works, what it deliberately refuses to count, and the decisions you should make on purpose rather than by accident — particularly if some of your players are children.

What a club ladder is, and what it is not

A club ladder is an internal rating list. It belongs to your club, it is computed from your own results, and it is not connected to any national or international rating. That distinction matters more than anything else on this page, so it is worth being blunt about it.

Neither replaces the other. A club ladder answers “who has been playing well at our club this season”, which is the question your members actually ask, and it answers it on Tuesday night rather than next month. A federation rating answers “how strong is this player compared with everyone in the country”, which a club cannot compute no matter how good its software is.

The practical uses are ordinary and immediate. It seeds the first round of your next Swiss without anyone guessing at an order of strength. It gives a returning player something to aim at. It lets you band a big event into sections that are defensible. And it gives newcomers a reason to come back a third time.

The maths: expected score, K-factor, and one number

Nearly every club ladder in the world is a version of the Elo system, and Elo is two lines of arithmetic. First, work out what the ratings say should happen:

Expected = 1 / (1 + 10^((opponentRating - yourRating) / 400))

That produces a number between 0 and 1 — your predicted score in that one game. Equal ratings give 0.5. A 200-point gap gives the favourite about 0.76. A 400-point gap gives about 0.91. Then compare the prediction with reality:

New rating = old rating + K x (actual score - expected score)

Actual score is 1 for a win, 0.5 for a draw, 0 for a loss. Beat someone you were expected to beat and the gap is small, so you gain little. Beat someone rated far above you and the gap is close to 1, so you gain a lot. Nothing else is going on.

K decides how fast the number moves. A high K makes ratings responsive and jumpy; a low K makes them stable and slow. Every rating system in existence is a compromise between those two, and they all solve it the same way: new players get a high K so they find their level quickly, and settled players get a lower one so a single bad night does not undo a season.

FIDE also keeps K at 40 for every player until the end of the year of their 18th birthday while their rating is under 2300, on the reasonable grounds that children improve faster than the maths expects. If your club is mostly juniors, that is a rule worth knowing about even if your own ladder does not copy it.

One more detail matters and is easy to get wrong: compute the whole event against the ratings everyone held before it started, then apply the changes at the end. If you update after every single game, the answer depends on which board finished first, and recomputing the event later produces a different number. Freezing the pre-event ratings is what makes a ladder reproducible. FIDE does the same thing across a whole rating period.

A worked example you can check by hand

Ann is rated 1400. Over one four-round Swiss she plays opponents rated 1500, 1300, 1600 and 1350, and scores 2½/4. She has played 30 rated games before tonight, so her K is 20.

Her expected total is 0.36 + 0.64 + 0.24 + 0.57 = 1.81. She actually scored 2.5. The difference is +0.69, so her rating change is 20 x 0.69 = +13.8, and she leaves on 1413.8. Notice where the points came from: the draw against 1600 was worth more than either win, because it was the result the ratings least expected.

That is the entire system. If a member ever asks why they gained so little for winning, the answer is always the same — the ladder already thought they would win.

The first games: provisional ratings

A brand-new player has no rating, so the ladder has to invent one. 1200 is the usual choice. It is a guess, and it can be badly wrong in either direction: the newcomer might be a returning county player or might have learned the moves last week.

Every serious rating system therefore marks early ratings as unreliable. FIDE will not publish a rating at all until it is based on at least five games against rated opponents. US Chess runs a separate “special” formula for players with eight or fewer games and does not call a rating established until it rests on more than 25. A club ladder that shows a number after one game and treats it like everyone else’s is doing something all three of those systems specifically avoid.

The usual club-scale compromise is to call a player provisional until they have five rated games, show the rating in grey, and list provisional players below established ones. That is presentational. The rule that actually protects the ladder is arithmetic:

The exception is what makes a brand-new club work at all. When both players are provisional, the game counts fully for both: neither has a real rating to protect, so nobody is treated unfairly. In a club’s first season everyone is provisional, so every game counts and the ladder gets going normally. The protection then switches itself on player by player, as each one reaches five games and becomes established.

Two things follow that are worth telling your members in advance. A provisional player’s number will swing violently for a few weeks and then settle down — that is the system working, not a bug. And an established player who spends a whole evening playing newcomers may finish on exactly the rating they started with. That is also correct.

The hard part is names, not arithmetic

The maths above is a few lines of code. The genuinely difficult part of running a ladder is deciding that the J. Smith somebody typed in tonight is the John Smith who has been on the ladder for two seasons.

Get this wrong in one direction and a regular quietly acquires a second entry, splitting their record in half. Get it wrong in the other and two different people are welded into one, which is much worse and much harder to unpick — and if they are two children with the same surname, worse again.

So the rule is simple and non-negotiable: software may suggest a match, a human must confirm it. Good matching catches the cases you would expect — an initial standing in for a first name, a middle name that appears some weeks and not others, a one-letter typo in a surname, a swapped name order — and offers them to the director as a list to approve. It never merges anybody silently. And whatever it does must be reversible, because the merge you regret is discovered three weeks later.

Two habits make this much easier. Ask members to enter their name the same way every time, and put that on the sign-in sheet. And when you do get a duplicate, fix it that night while you still remember which one is which.

What never counts: byes, forfeits and bughouse

A ladder is only trusted if it refuses to rate things it cannot honestly rate. Three cases come up in every club season.

• Byes. A bye is a point awarded because there was nobody to play. It counts towards the tournament standings and it must never touch the ladder — no game happened, and there is no opponent whose rating could be part of the calculation.
• Forfeits and no-shows. Same reasoning. A player who wins because their opponent did not arrive has demonstrated nothing about their chess. FIDE’s rating regulations say it plainly: unplayed games, whether by forfeit or any other reason, are not counted.
• Bughouse. Two-against-two on two boards is a real and excellent format, and standard Elo does not model it. Rating it would produce numbers that look official and are not. Run bughouse, publish its standings, and keep it off the ladder.

The general principle behind all three: rate games that were played, by the people who played them, against opponents who have ratings. Everything else stays in the tournament result where it belongs.

Correcting a result three events later

Somebody will write down the wrong result. A director will publish the standings at round two and finish the event later. An event will be deleted because it was created by mistake. All three are ordinary, and all three break a ladder that stores only the current number.

The fix is to keep the whole history and recompute rather than patch. Store each event’s rating changes alongside the event itself, and when anything changes, replay every event in date order from the beginning. It is more work per correction and it is the only approach that survives contact with a real club season.

The failure mode to watch for is subtle and worth stating, because it is invisible: if a correction is saved but the ladder is not replayed, every page still loads, every number still looks plausible, and the ratings are quietly wrong. Nothing on screen tells you. When you correct a result, make sure the ladder is actually rebuilt — and if your software has a manual rebuild button, that is what it is for.

Privacy: names, children, and who can see the page

A ladder is a list of named people and how good they are at something, kept over time. Publishing one is a real decision, and the answer is different for an adult club and a junior section.

Three questions, in order.

• Who should be able to see it? Most clubs want one of three things: visible only to the director, visible to anyone with the link but not listed in search results, or fully public and indexable. A link that is unguessable and marked noindex covers the common case — members can find it, search engines do not surface it.
• Full names, or initials? Showing “J. B.” instead of “Jason Becher” keeps the ladder perfectly usable — members recognise themselves and each other — while removing the part that a stranger could act on. For any event involving under-18s, initials should be the default rather than the option.
• Have the parents agreed? Publishing a child’s full name, alongside their results and the fact that they are at a named venue on a known weeknight, is a safeguarding decision and not a technical one. Get consent first, ask the child as well as the parent, and check whether your school or club already has a policy — it will outrank anything on this page.

Beyond that, collect less. A ladder needs a name and results. It does not need a date of birth, an address, a phone number or an email. Data-protection rules differ substantially between countries, so treat this as a prompt to check your own rather than as legal advice — but “we only ever stored names and results” is a good position to be in under any of them.

Doing this in Chess Caddy

Chess Caddy runs tournaments free, offline, with no account, and that does not change. The ladder is part of a club page — a permanent home on the web for a club or school, at chesscaddy.com/club/your-club, holding the ladder, the roster, per-player profiles and every event you have run. That part is a paid subscription, because unlike the app it runs on servers that cost money every month.

What it implements is exactly what this guide describes. Players start at 1200. K is 40 for the first 20 games and 20 afterwards. Ratings are provisional under five games, shown in grey, and a provisional opponent cannot move an established player’s rating unless both are provisional. Every event is computed against pre-event ratings and applied at the end. Byes, forfeits and bughouse are never rated. Name matching suggests links for the director to confirm, merges can be undone, and correcting or deleting a published result replays the whole ladder automatically.

Practically, the loop is short: run the event in the app as usual, press Publish, and file it under your club. Confirm which of tonight’s names are which existing members, and the ladder updates. The per-player pages then carry a lifetime record and a rating history, which is the part members actually come back to look at.

Pricing is $2.99 a month or $24.99 a year, one price for clubs and schools alike, cancel whenever you like. If you are not sure a ladder is for you, run two or three events with the free Swiss tools first — the published result links are free too, and a club that has never published anything has nothing to put on a ladder yet.

Frequently asked questions