Record · Full record
Every pick made while its strategy was active, since 4 May.
Flat stake · 2,755 picks since 4 May
Match-result picks we've since stopped made more than the whole profit.Past results, not a forecast.
Counted at the best price we found when we posted it. Prices move; check yours before you bet. 780 older picks use the best price of any bookmaker, not a fresh read.
The 95 % range is where the true return per pick likely sits; it crosses zero, so luck isn't ruled out.
The "Match result — old model" strategy (since stopped) made 112 % of the profit.
Picks before 28 Sept were timed by our database, not sealed.
CLV −2.5 % vs the sharp closing price, n 2,205.
2,755 entries: 2,600 distinct picks plus 155 copies (a pick made by several strategies counts once each).
Worst drop counts every copy of a pick; the longest losing run counts each pick once.
| Market | Picks | Won | Per pick |
|---|---|---|---|
| Match result (1X2) | 2,430 | 1,206 | +2.8% |
| Total goals in the match | 325 | 131 | −4.5% |
Ranges and caveats on each market page.
Active strategies count in the headline. Retired ones keep the picks they made while active in the headline. Testing ones have their own record only.
Hit rate is the share that won; on its own it flatters short prices, so it always sits next to the break-even rate those prices need (the average of 1 ÷ odds over the same picks).
Sharp linemeans no prediction model: the price beats the sharpest bookmaker's fair price with its margin removed. Consensus uses several bookmakers agreeing instead.
Since 4 May 2026 · flat €25 per pick (your stake) · ROI at the best price available when each pick was made (all books) · 189 strategies tested, 175 retired
Active rows show each strategy's full record (1,838 settled), including picks from before it became active. The headline (2,755) counts only picks made while a strategy was active — each active row says how many of its picks that is, and the “no longer active” rows hold the rest. A pick two bots both made counts once in each bot's row. In money: the active rows add up to +€1,911 over their full records; the picks of theirs that count in the headline come to −€377, and the no-longer-active rows add +€1,726 — together +€1,350, the headline P&L.
Every pick these strategies made, tested ones included. Only the picks a strategy made while it was active count in the record above; the rest were tests and never entered it.
What our earlier strategies taught us: the same everyday bookmaker prices beat the closing price at the sharpest bookmaker by +4.7% (n 1,463) when they were chosen against that bookmaker's fair price, and −5.9% (n 2,052) when they were chosen against our model. What decides the bet is the yardstick, not the bookmaker.
Sharp line-shopping (a book beats the sharp line)
72 strategies · 16,690 picks · 7,846W–7,594L
ROI +5.0%
15,440
vs close +4.3% (n 14,458)
Model — match result (1X2, double chance, handicap)
40 strategies · 12,008 picks · 5,436W–6,398L
ROI +0.6%
11,834
vs close +1.1% (n 9,599)
Flagged picks: 2,319 in the 1X2 swap window · 2,225 before mid-July (weaker closes)
In-play
20 strategies · 3,089 picks · 1,788W–1,056L
ROI −2.4%
2,844
vs close n/a (in-play)
Flagged picks: 1,237 before mid-July (weaker closes)
New markets (corners, team totals, first half)
3 strategies · 3,073 picks · 1,330W–1,700L
ROI +2.6%
3,030
vs close +0.8% (n 2,959)
Model — goals (over/under, both teams to score)
17 strategies · 2,381 picks · 1,118W–1,214L
ROI −3.4%
2,332
vs close −1.1% (n 1,868)
Flagged picks: 328 in the O/U calibration-bug window · 520 before mid-July (weaker closes)
Soft-book triggers vs OUR MODEL
6 strategies · 2,301 picks · 658W–1,622L
ROI −6.1%
2,280
vs close −5.9% (n 2,052)
Flagged picks: 1,262 in the 1X2 swap window · 939 in the O/U calibration-bug window
Soft-book triggers vs the SHARP line
9 strategies · 1,510 picks · 712W–780L
ROI +10.8%
1,492
vs close +4.7% (n 1,463)
Pre-registered forward test
7 strategies · 1,166 picks · 494W–653L
ROI +0.5%
1,147
vs close +0.1% (n 1,137)
Accumulator legs
1 strategy · 533 picks · 284W–249L
ROI −10.6%
533
vs close −3.9% (n 220)
Flagged picks: 316 before mid-July (weaker closes)
Same basis as every row above: flat 1 unit per pick at the best price available when the pick was made (all books); in-play picks at their own in-play price. vs close = the pick's price against the closing price at the sharpest bookmaker, margin removed (in-play picks have none). Flags: the 1X2 swap window is 10 May–14 Sept, when our served match-result model partly mixed up home and away; the O/U calibration bug ran 3–13 Sept; closing lines before mid-July are weaker evidence. Flagged picks are counted, not removed. Named strategies are the largest per family with a good closing-line record, never picked by result.
189 strategies tested since May; 175 retired. VIP strategies have their own page: VIP details.
Sharp line and consensus picks use no prediction model at all — they beat a fair price, margin stripped out, by at least 3%; the difference is whether one sharp bookmaker sets that fair price or several agreeing ones do. Nothing is capped — a busy Saturday runs long, a thin Tuesday shows none.
Sharp-edge picks are priced directly against the market with the bookmaker's margin removed — against the sharpest single line where one is sharp enough to trust, and otherwise against the consensus of several bookmakers. They start at zero on 14 September 2026 and no past performance is claimed for it.
The picks page is the shortlist: our proven lines in full, newer strategies only at their strongest prices. Every pick, and each strategy's full record, is in the list above — tap a row for its numbers. See also the FAQ.
Every one recorded before the match and settled against official scores.
We are not claiming an edge. Measured against the closing market on the same bets, our model’s picks are not statistically distinguishable from the price they were taken at. The ledger is published in full so you can check that yourself — including the periods where it lost. The totals also count the sharp-edge rule that makes most picks published today; it has been tracked from zero since 14 Sept and makes no claim either.
2,755 bets logged·6 strategies live·175 retired·179 days running
At €25 per pick, the 2,755 settled picks since 4 May would have returned +€1,350 (+2.0% ROI).
Could this be luck? Not ruled out yet. With 2,755 picks, a realistic range is −€1,591 to +€4,291. The more picks we settle, the clearer it gets.
Worst drawdown at €25: −€1,760 · longest losing run, counting each strategy’s copy of a pick: 21 · you’d have needed about €3,600 set aside.
Flat stake, no compounding. Settled picks only, priced at the best price available on a publishable book when the pick was logged. ROI and pick counts don’t change with the stake. Past results, not a forecast.
The work behind it
51,210 picks tested across 189 strategies that made at least one pick, since May 2026 (30,354 distinct selections, 49,081 settled). 175 of them were retired, 8 are still in internal testing, and 6 are on this page.
The ROI at the top is today's proven strategies only. Retired picks count here and in the retired section — never in that number, whether they won or lost.
Streaks
variance is real — even a +5% ROI edge sees double-digit losing runsOver the headline picks, oldest to newest. Current and longest winning runs count each pick once (the recent results list); the longest losing run counts every leg in the headline total, the figure next to the worst drawdown.
Current
5L in a row
Longest winning
10W
Longest losing
21L
The full method, step by step, is on the methodology page. We run the same test on tipster sites: The Price Check. Check any pick: verify one.
Get the picks before kick-off
Free picks on Telegram (opens in new tab)18+ · Betting can cost you money · Get help (opens in new tab)
39 corrections, every one listed