TTLossbeater

Measured against the market · last 30 days

Most prediction numbers can't be checked. Ours can.

Every read is published with its denominator and graded when it settles — including the ones we get wrong. A losing match is never removed from the record.

Our signals

61.1%

named the winner across 29,121 matches

The favourite

61.0%

won the same ones

±0.0 pp

A statistical tie

on the same 29,121 matches

That is a statistical tie with the market — and we publish it anyway. Nobody reliably out-predicts these odds; the value here is the analysis underneath, not a promise to beat the bookmaker.

Where these two figures come from: both read the signal call stored on each match, and that call is rebuilt whenever a finished match is re-processed — so neither is a frozen pre-match snapshot. What does not change is which matches are counted: nothing is dropped for losing.

About one in three of what we send to the Pro group also goes out free on Telegram, before first serve.

Open the channel

Figures refresh every 30 minutes. Judge the 30- and 90-day windows; shorter ones are noise. Past results do not indicate future results. Information only, not betting advice. 18+, where betting is legal for you.

Signal Performance

2026-08-12 → 2026-09-11 · 31,815 resolved matches

31,815Resolved matchesSettled in the window
60.6%Match winner19,288 of 31,811
56.9%First set18,004 of 31,654
+0.1 ppVersus the market61.1% ours · 61.0% the favourite

Daily hit rate

match winner, per settled day
Above the market on 12 of 31 days

The dashed line is what the bookmaker's favourite did over the same window. Crossing it on a given day is sampling noise at roughly 1,026 matches a day — the 31-day figure is the one that means something. The two are not drawn on identical sets either: the daily line covers every resolved match, the dashed one only those carrying a price.

A different population

Shipped to the Pro group

Everything above covers every resolved match, sent or not. This covers only the 1,349 we sent, summed across every tier the record carries.

1,349

Sent

56.0%

Hit rate

1.70

Avg odds

756 of 1,349 won

A hit rate only means something next to the price. 56.0% on 1,349 picks at average odds of 1.70 — breaking even at that price needs 58.9%. What you make of it depends on which picks you take and how you stake.

By market

what we predicted on every resolved match, against what the price implied

Match winner

Who takes the match

Ours

61.1%

The market

61.0%

±0.0pp

Both figures on the same 29,121 matches — those carrying a signal call and a price with a favourite in it. Matches quoted at identical odds on both sides are left out, so this is not the same count as the odds bands below.

First set

Who takes set one

Ours

56.9%

The market

No line

Correct 18,004 of 31,654 resolved

Over / under

Total sets played

Ours

56.0%

The market

No line

Correct 12,588 of 22,489 resolved

Only match winner has a published market counterpart. First set and over/under are hit rates with no reference, so no delta is shown for them.

Where we differ from the price

Loading the disagreement ledger…

By the favourite's odds

bands set by the favourite's price · our hit rate against what the price gave the side we backed

Banded by the favourite's price · 31,470 matches

Heavy favourite

Favourite under 1.40

A favourite in this band was priced at 74.3% and won 75.6% of 5,412 — that pair grades the market, not us.

5,412
74.2%

+1.3 pp

Vs our side's price

Value

Favourite 1.40 – 1.75

A favourite in this band was priced at 58.6% and won 58.9% of 19,514 — that pair grades the market, not us.

19,514
58.3%

+0.6 pp

Vs our side's price

Close

Favourite 1.75 – 2.30

A favourite in this band was priced at 50.9% and won 51.8% of 4,195 — that pair grades the market, not us.

6,544
50.5%

+2.9 pp

Vs our side's price

The mark is what the price gave the side WE backed — not the favourite's, because we do not always back the favourite. A hit rate above it is not money: it is an accuracy edge, and an accuracy edge still has to clear the bookmaker's margin before it pays.

By league

match winner only

21 leagues over the sample floor

Setka Cup

Core league

9,763

TT Elite Series

Core league

9,199

Czech Liga Pro

Core league

5,944

TT Cup

Core league

5,262

Setka Cup Women

Core league

298

32 low-sample leagues

Under 30 matches each — grouped, so a one-match league can't render as a 0% or a 100%

406

Matches

No per-league market rate is published yet, so no league carries a delta. A hit rate on its own does not say whether that league beat its own prices.

Where these figures come from: the accuracy numbers read the signal call stored on each match, and that call is rebuilt whenever a finished match is re-processed — so we do not claim it is a frozen pre-match snapshot. The Pro-group card is the frozen record: those picks are graded from what was actually sent, at the odds sent.

Track Record — Prediction Performance & Hit Rate | TT Lossbeater