PickleLog · Ratings

How your rating works

Your rating is a single number that estimates your current playing strength — measured, not guessed. It's built on the Glicko family of rating systems, the same statistical approach used to rank players in competitive chess and online games. Here's how it thinks.

We keep the exact formula private to protect the system's integrity, but nothing about how it works is a secret. This page lays out every principle behind the number.

The idea

Every rated game updates a live estimate of your skill

A rating isn't a running total of wins. It's a probability model: given everyone's ratings, the system forms an expectation for how a game should go, then compares that expectation to what actually happened. Beat expectations and your rating rises; fall short and it dips. Do exactly what was expected, and it barely moves.

Three things shape each update: who you played, how the game went, and how confident the system already is in your rating. The rest of this page is those three things.

01Opponent rating

Who you beat matters as much as whether you won

Before a game, the system estimates each side's chance of winning from the rating gap. Beating a stronger opponent was unlikely, so it moves your rating up more. Beating someone far below you was expected, so it barely nudges you — and losing to them costs the most.

This is what makes ratings comparable across groups. A 4-0 week against beginners and a 4-0 week against strong regulars are not the same result, and your rating reflects that difference automatically.

E = 11 + 10(Ropp − Ryou)/400

Your expected chance of winning is a smooth curve of the rating gap: an even match sits at 50%, and a ~400-point edge makes you roughly a 10-to-1 favorite. This is the shared, public backbone of every Elo- and Glicko-based rating — what stays private is only how far each result then moves you.

Win probability vs. rating gap — the expectation each game is measured against.
Team-aware In doubles, the system uses the combined strength of each side, then attributes the result to the players on court — so partnering up or down is accounted for, not ignored.
02Point differential

Margin refines the signal — within limits

An 11–2 win says more than an 11–9 win, and the rating listens. Point differential is a secondary input that sharpens each update beyond the bare win or loss.

But it's deliberately bounded. A single blowout can't rocket you up the ladder, and one bad beating won't sink you. The system trusts a pattern of results over many games far more than any one scoreline — which is exactly what keeps the number stable and hard to game.

03Reliability

Your rating carries a confidence level

Alongside your rating, the system tracks how sure it is about that number — we call it reliability (statistically, a rating deviation). A brand-new player's rating is a rough guess, so it's held with low reliability and allowed to move quickly. As you play, the estimate tightens, reliability rises, and your rating settles down.

new rating = old rating + w × ( Sactual − Eexpected )

S is what actually happened (1 for a win, 0 for a loss), E is what was expected from the formula above, and the weight w is larger while your reliability is low. That single term is why a new player's rating moves fast and finds its level, while an established player's barely budges from one game.

Early games swing widely as the system finds your level; the band narrows as reliability builds.
≈ 25–35 games Roughly how many rated games it takes for reliability to become stable. The exact number depends on how consistent your results are — steady, predictable outcomes converge faster; erratic ones take a little longer.

While reliability is still low

The system is intentionally cautious. Your rating is shown but treated as provisional: it moves faster so it can find your true level sooner, and it's weighted more lightly when you're compared against established players. That protects both you and your opponents from an early, misleading number.

What the red asterisk means

While your reliability is still low, PickleLog marks your rating with a red asterisk (*) everywhere it's shown — your profile, standings, and match history. It isn't a penalty or a warning about your play; it's a flag that the number is still provisional and moving quickly as the system learns your level. It clears on its own, typically after around 30–40 approved matches, once your reliability has built up enough for the rating to be treated as established.

04Recency

It reflects your form now, not your best week ever

Skill drifts, so ratings shouldn't be frozen. Recent games carry more weight than old ones, and if you stop playing, the system slowly grows less certain your rating still describes you — your reliability decays over time away from the court.

Come back after a long break and your rating will move a bit more freely again for a few games while it re-confirms your level, then re-stabilize. Nobody's rating quietly rots or freezes in place — it stays honest to how you're actually playing.

05Comparing across groups

Your rating is built from who you've actually played

Every update to your rating comes from games against people already in the system. That makes it very reliable for comparing yourself to people you've played, or who've played people you've played — your regular group, club, or league.

It's less precise the further you get from that web. Two players who've never shared a court, in two groups that have never played each other, may show similar numbers without there being much real evidence connecting them. The rating is honest about your results — it just can't vouch for a comparison it has no data on.

06DUPR conversion

Translating your rating to a DUPR-style number

If you're used to thinking in DUPR terms, PickleLog's rating converts over with simple math: multiply by 10. A PickleLog rating of .330 lines up with roughly a 3.3 on the DUPR scale.

.330 → 3.3 PickleLog rating × 10 ≈ DUPR-equivalent. Use it as a quick reference point, not an exact match.

DUPR is a separate organization with its own algorithm, player pool, and match data, so this is a rough translation for orientation — not an official DUPR rating. Two players with matching converted numbers haven't necessarily been measured the same way.

What doesn't count

Not everything you log touches your rating

Plenty of PickleLog is there for motivation and record-keeping. Those things are kept completely separate from the skill rating on purpose.

Affects your rating

  • Rated games with an approved final score
  • Opponent (and partner) ratings
  • Point differential, within bounds
  • How recently you've played

Does not affect your rating

  • Unrated games — logged for your own stats and trophies only
  • XP & levels — a measure of activity, not skill
  • Trophies & streaks — earned for milestones, never for rating
  • Games still awaiting approval, or placeholder players
Good to know

The questions that come up most

My rating dropped after a win. Is that a bug?

Almost never. If you were heavily favored and the game was close, you may have underperformed the expectation even in a win — especially by a slim margin. It works both ways: you can gain rating from a hard-fought loss to a much stronger opponent.

Why does my rating still move a lot?

You're likely still in the provisional window (under ~25–35 games) or returning from a break. Larger moves early are a feature — the system is homing in on your level. They shrink as reliability builds.

Can I inflate my rating by cherry-picking games?

Not meaningfully. Expectations are set by the rating gap, margins are capped, and the system weights the long-run pattern over any single result. Both sides also confirm the final score before it counts.

Is my rating public?

Your history is yours. Where ratings appear alongside others (like standings), they're shown to give context — the same number, computed the same way for everyone.

Can a score be corrected after it's approved?

No. Once everyone approves a match, the score — and the rating change it caused — is final. That finality is what makes ratings trustworthy: there's no re-litigating a result after the fact, so it's worth double-checking before you approve.

Is there a highest or lowest possible rating?

No fixed floor or ceiling. Your rating is simply a running estimate from your results over time, and it can move as far up or down as your match history supports.