ProSoccer review: 3-day test, too short to call

Margus Sellin2 Nov 2026Grade C · 31 Oct to 2 Nov 2026

Our verdict

+8.9%

priced 1X2 tips · 180 picks · 31 Oct – 2 Nov · at its own odds, unverified Why

We sell picks too. Its record was read after the matches.

OddsIntel sells a paid VIP Telegram channel of its own picks, so we compete with ProSoccer. Our own numbers face the same test below. ProSoccer's history was read from its public pages after the matches (proof grade C).

picks tested: 180; picks tested
picks read: 478; picks read
at the sharp close: +11.9%; at the sharp close

In our test, €10 flat on every ProSoccer 1X2 pick would have returned €161 in profit, which 3 days cannot tell from luck (180 tips is too few to read; 95% range of the return −20.0% to +19.0%; past results, illustration only). This is a 3-day test, far shorter than the 30 days our format asks for, because the site publishes no dated history we can read.

What is ProSoccer? Price, tiers and claims

In short: a free predictions site with a tip and an average-odds figure for every match, and no profit or loss shown.

ProSoccer publishes free football predictions. From 31 Oct (capture began 30 Oct, before the test window), each match carries its own 1X2 probabilities, a tip, the average bookmaker odds, two predicted scores and under/over 2.5 goals probabilities, across more than 150 leagues and cups, from a model it says uses artificial neural networks. The page does not name an operator, and we saw no paid tier.

We tested one pick per match, its tip. A single-result tip is graded at the average odds printed beside that outcome. A two-outcome tip is graded as a double chance; the site prints no price for it, so we judged it only at the sharp close. These are its claims:

What the site says, next to what we measured

ProSoccer says	What we measured
“Since 2002 we have been publishing our free predictions on the web, which are generated by a computational intelligence software”	Match result (1X2), 31 Oct to 2 Nov 2026: 180 picks returned +8.9% at its own odds (95% range −20.0% to +19.0%).
“Best of successful predictions for yesterday”	We can only test what it publishes: 478 picks in our window. Alerts it left out of the record are not visible to us.
“The predictions are based on sports data entered in our system, which uses artificial neural networks.”	Match result (1X2), 31 Oct to 2 Nov 2026: 180 picks returned +8.9% at its own odds (95% range −20.0% to +19.0%).

“Since 2002 we have been publishing our free predictions on the web, which are generated by a computational intelligence software”
What we measured
Match result (1X2), 31 Oct to 2 Nov 2026: 180 picks returned +8.9% at its own odds (95% range −20.0% to +19.0%).
“Best of successful predictions for yesterday”
What we measured
We can only test what it publishes: 478 picks in our window. Alerts it left out of the record are not visible to us.
“The predictions are based on sports data entered in our system, which uses artificial neural networks.”
What we measured
Match result (1X2), 31 Oct to 2 Nov 2026: 180 picks returned +8.9% at its own odds (95% range −20.0% to +19.0%).

How we checked ProSoccer's picks

Every pick on ProSoccer's public results pages, 31 Oct to 2 Nov 2026: 180 graded picks in 1 market, checked against the scores and closing odds we record.

The method in full

What we collected: Every pick on ProSoccer's public results pages in the window, with the odds printed beside it and its result (478 rows read)
Period: 31 Oct to 2 Nov 2026 (3 days, by kick-off date)
Sample: 180 graded picks in 1 market; 95 picks we could not match to a fixture are excluded; 6 picks in markets we could not test are counted and excluded; ProSoccer prints no odds for Double chance (137 picks), so it is graded only at the sharp closing price, never at a price ProSoccer did not print
Prices: the 'Avg. Odds' the site prints next to each match (average of bookmakers, no book named). ONE pick per match: its TIP. A single result tip (a1 / a2) is graded at the printed average odds of that outcome. A two-outcome tip (a12 / a21 / a1X / aX1 / a2X / aX2) is graded as that double chance; the site prints no price for it, so it has NO own-odds row and is judged only at the Pinnacle close (said so on the page, never priced from its 1X2 numbers). c-tips (rows with two odds columns) are not a market we test; as the second basis, the sharp close: the sharp closing line, with its margin removed
Proof grade: C: grade C because 2 of 478 tips were first seen after kick-off and 0 were edited later; the other 476 were first seen before kick-off by our own capture
Pre-committed: Site, window, markets and tests logged 30 Oct 2026, before the audit ran

ProSoccer's history was read after the matches from its results pages; we did not capture it live, so the pages could have changed before we read them. Paid tiers were not tested. Break-even is computed pick by pick (the average of 1 ÷ odds). Every 95% range comes from a bootstrap that resamples whole days, because picks on the same day are linked. Voids and pushes return the stake and count as no bet; results are our fixture result over 90 minutes plus stoppage time.

The window is 3 days (31 Oct to 2 Nov 2026), far under the 30 days our format asks for. The site has no usable history: its archive page lists days only up to 2018, and a past date returns today's table, so we could not backfill and this 3-day test uses what we captured live. We did not contact ProSoccer. Re-check due on or after 1 Dec 2026, when the full 30 days are in.

ProSoccer's pages were read twice under two dates in our first capture; we kept the earliest sighting of each match (955 → 478 rows). Of 478 picks read, 2 were first seen after kick-off and 0 were seen edited later in the one row we kept per match (not seeing an edit is not proof there was none), so we grade the issue C.

6 c-tips (rows with two odds columns) were left out as untested, and 1 double-chance picks were voided. About 19.8% of picks could not be matched to a fixture we record; the site shows final scores but no won or lost mark, so we cannot check its results against ours, and we leave those picks out. The tip codes are not explained on the site; reading a1 and a2 as a home or away win and the two-outcome codes as double chances is our inference. 51 tips (27 of them 1X2) had no result yet when we ran the audit and are left out.

The full method is on our methodology page.

ProSoccer results at its own odds

ProSoccer's tested markets, 31 Oct to 2 Nov 2026

Market	Picks	Win rate	Avg odds	Needed to break even	Return at its odds	95% range	At the sharp close
Match result (1X2)too few to read	180	60.0%	1.90	56.4%	+8.9%too few to read	−20.0% to +19.0%	+12.2%180 of 180 picks: too few to read

Market	Won	Needed	Return
Match result (1X2)too few to read	60.0%	56.4%	+8.9%too few to read

Source: ProSoccer results pages, graded by us. Rows under 300 picks greyed.

The 1X2 row is greyed because more than 5% of its picks were unmatched. The mean CLV against the fair close was −6.1% (95% range −13.4% to −5.2%), with only 22.8% of picks beating it; under grade C we print it but call it unreliable. With CLV negative and unreliable, the +12.2% at the sharp close (95% range −14.0% to +20.9%) is not a result, and 180 picks is too few to read (+11.9% across 317 tips including double chance; CLV −6.1% makes it unreliable). A win rate of 60.0% against a break-even of 56.4% is a small gap on 180 tips, too few to read. Double chance has 137 picks at the sharp close, too few to read.

No market reached 300 picks in this window, so there is no profit and loss chart: a line drawn from fewer picks would mostly show luck.

Did ProSoccer beat the closing odds?

ProSoccer's odds: capture time unknown, read after the matches (proof grade C). A tipster who beats the close is finding prices; one who doesn't is following them.

ProSoccer against the de-vigged sharp close (capture time unknown: unreliable)

Market	Picks matched to a sharp close	Mean CLV	95% range	Picks beating the fair close
Match result (1X2)too few to read	180	−6.1%	−13.4% to −5.2%	22.8%

Match result (1X2) · 180 matched · too few to read

Mean CLV = the expected return at the fair close. CLV per pick = its odds × the sharp close's fair probability − 1 (margin removed by the Shin method for three outcomes, the power method for two). The share beating the close is descriptive only. Source: ProSoccer results pages, graded by us. Rows under 300 picks greyed.

Could you get its printed price?

ProSoccer's printed odds against the prices bookmakers offered before kick-off

Market	Picks checked	Within the tolerance of the best price offered	More than 1.5× every book	Median against the best price	Against its reference book
Match result (1X2)	183	95.1%	0.0%	−6.7%	n/a

Match result (1X2)
Within the tolerance of the best price offered
95.1%
Picks checked
183
More than 1.5× every book
0.0%
Median against the best price
−6.7%
Against its reference book
n/a

Best price = the highest price any of the bookmakers we record offered on the same pick in the 48 hours before kick-off (median 9 books recorded per fixture, so where it is 1 this check is weak; exchanges excluded), with a 2% tolerance.

Is it skill? ProSoccer against a no-analysis rule

On each day ProSoccer published picks, the rule takes the same number of picks, choosing whichever outcome the sharp closing odds made most likely. No form, no team news, no model. Both sides are priced at the sharp close, so each side counts only the picks with a sharp closing price, and the two counts can differ.

Same days, picks with a sharp close: the site against a no-analysis rule

Market	Its picks	Its win rate	Its odds	Its return	Rule picks	Rule win rate	Rule odds	Rule return
Match result (1X2)too few to read	180	60.0%	2.00	+12.2%too few to read	0	n/a	n/a	n/a

Match result (1X2) · 180 vs 0 picks · too few to read

Returns at the sharp close. Caveat: the rule chooses with kick-off prices, so it shows what a win rate is made of, not a strategy, and nobody has kick-off prices when a tip is published.

One real day of ProSoccer picks

ProSoccer's card, 31 Oct 2026 (selected rows)

Match	League	Pick	Odds	Sharp close	Score	Result
BAYERN MUNCHEN v BORUSSIA DORTMU	DE1 (Germany)	Match result (1X2): a1	1.30	1.32	2–2	Lost
SCHALKE 04 v RB LEIPZIG	DE1 (Germany)	Match result (1X2): a2	1.85	1.87	0–2	Won
LEVERKUSEN v STUTTGART	DE1 (Germany)	Match result (1X2): a1	1.70	1.78	2–4	Lost
MONCHENGLADBACH v PADERBORN	DE1 (Germany)	Match result (1X2): a1	1.80	2.19	0–0	Lost

BAYERN MUNCHEN v BORUSSIA DORTMU
Result
Lost
League
DE1 (Germany)
Pick
Match result (1X2): a1
Odds
1.30
Sharp close
1.32
Score
2–2
SCHALKE 04 v RB LEIPZIG
Result
Won
League
DE1 (Germany)
Pick
Match result (1X2): a2
Odds
1.85
Sharp close
1.87
Score
0–2
LEVERKUSEN v STUTTGART
Result
Lost
League
DE1 (Germany)
Pick
Match result (1X2): a1
Odds
1.70
Sharp close
1.78
Score
2–4
MONCHENGLADBACH v PADERBORN
Result
Lost
League
DE1 (Germany)
Pick
Match result (1X2): a1
Odds
1.80
Sharp close
2.19
Score
0–0

The full card that day held 194 graded picks, 134 of them won. Source: ProSoccer results.

ProSoccer scorecard: who it's for

Shows the odds for every pick

Odds printed beside 100.0% of the picks we graded, in every tested market.

Shows profit or loss

It shows final scores and a selection of past wins, but no won or lost record and no profit or loss.

Profitable at its own odds

Not tested

One priced market at its printed odds (+8.9%, 95% range −20.0% to +19.0%); the row is greyed (19.8% unmatched), the range includes losing and the window is 3 days, so this is noise on too few tips to read (180, under 300). Tips were read live; 2 of 478 first appeared after kick-off, none were edited, so we grade C.

Beats the closing odds

Unreliable

ProSoccer's odds were read after the matches (proof grade C), so the capture time is unknown and this row cannot pass. The CLV table above is printed for reference only.

Beats a no-analysis rule

Not tested

No busy market to compare; the smaller markets are shown above, greyed.

Published record matches its picks

Not tested

Not compared.

Proof grade

Grade C

grade C because 2 of 478 tips were first seen after kick-off and 0 were edited later; the other 476 were first seen before kick-off by our own capture

What it does well

It is free, shows an average-odds figure beside every tip and lists many leagues and cups.
Its printed 1X2 odds were not more than 2% above the best price we hold on 95.1% of the 183 picks we could check; the median printed price was 6.7% below our best price.
Beyond a list of past wins, it does not claim a profit figure; it says what its model is and shows the final scores.

Who it's for: Readers who want a free daily list of model tips with average odds, and who will check their own bookmaker's price first.

Who should skip it: Anyone who wants a record they can check: it shows no won or lost marks and no history, and three days of our data prove little either way.

Is ProSoccer legit? Our verdict

Yes, a real, free site with no operator named.

180 tips over 3 days, graded at its printed odds; the range (−20.0% to +19.0%) includes losing, the row is greyed, 180 is too few to read, and the figure is noise.

Far too short a window to rate it; the re-check on 1 Dec 2026 covers 30 days.

Appendix

Our own picks, same test

The same window and the same columns, for two lines of ours: every settled pre-match pick of every bot on /performance at the price we published, and the subset posted to our public Telegram channel before kick-off. Our picks are in different markets at different odds, so this compares what a reader of each got, not two models on the same games.

Ours, same window: +12.0% (115, all) and +22.3% (49, Telegram).

Our numbers, the full table

OddsIntel picks, 31 Oct to 2 Nov 2026

Line	Proof	Picks	Win rate	Avg odds	Break-even	Return	95% range	At the sharp close	CLV
Every pick on /performance	C	115	57.4%	2.08	50.7%	+12.0%too few to read	+0.1% to +34.4%	n/a	n/a
Posted to Telegram before kick-off	B	49	63.3%	2.00	52.6%	+22.3%too few to read	+11.1% to +33.9%	n/a	n/a

Every pick on /performance
Return
+12.0%too few to read
+0.1% to +34.4%
Proof
C
Picks
115
Win rate
57.4%
Avg odds
2.08
Break-even
50.7%
At the sharp close
n/a
CLV
n/a
Posted to Telegram before kick-off
Return
+22.3%too few to read
+11.1% to +33.9%
Proof
B
Picks
49
Win rate
63.3%
Avg odds
2.00
Break-even
52.6%
At the sharp close
n/a
CLV
n/a

Every settled pre-match pick of every bot on /performance, at the published price; the subset posted to our public Telegram channel before kick-off. Flat stake, break-even per pick, CLV against the de-vigged sharp close by the same method as above. 95.7% of the /performance picks were sealed before kick-off.

How to check any tipster in 30 seconds

The five checks

1
Ask for the odds
No odds, no verdict. A win rate on its own can't be checked.
2
Work out the break-even
1 ÷ odds for each pick, then the average. At odds of 1.25 you need to win 80% of the time. Is the win rate above it?
3
Check the sample
A few hundred picks prove little. Thousands start to.
4
Check the price you would get
Quoted odds you can't find at your own bookmaker don't count.
5
Check it against the close
Did the picks get better odds than the market settled on at kick-off? That is the sign of finding prices, not following them.

ProSoccer FAQ

FAQ (6)

Is ProSoccer legit?

Yes, as a free predictions site: it publishes a tip and an average-odds figure for every match, from a model it says uses neural networks. In our 3-day test its 180 single-result tips (too few to read) gave a return whose 95% range runs from −20.0% to +19.0%, so it includes losing. Three days cannot tell skill from luck.

Is ProSoccer a scam?

We found no evidence that it is a scam: it is free, asks for nothing, and 95.1% of its printed 1X2 odds were not more than 2% above the best price we hold. It names no operator and shows no profit figure. We did not contact ProSoccer, and we could not check its tips over a longer period.

What are ProSoccer's real results?

Over 31 Oct to 2 Nov 2026 (3 days), its 180 single-result tips returned +8.9% at its printed average odds (95% range −20.0% to +19.0%) and +12.2% at the sharp close (95% range −14.0% to +20.9%). That is too few tips to read, and CLV was −6.11% and unreliable, so neither is a result (across 317 tips including double chance the sharp-close return was +11.9%, equally unreliable). Its two-outcome tips have no printed price.

Is ProSoccer's paid tier worth it?

We did not test it; we tested only the free predictions on its public pages. We saw no paid tier.

How did we measure this?

We read its daily prediction pages every three hours, matched tips to the fixtures and closing odds we record, and graded them at its printed average odds and at the sharp close (a sharp benchmark's closing line). It does not explain its tip codes; reading a1/a2 and two-outcome codes as results and double chances is our inference.

How accurate is ProSoccer?

Its 180 single-result tips (too few to read) won 60.0% of the time against a break-even of 56.4% at its printed odds, a small gap over 3 days (31 Oct to 2 Nov 2026) whose 95% range includes losing. About 20% of its picks could not be matched to a fixture we record, so the figure is indicative only.

Sources and archived copies

Show the sources

1.ProSoccer: predictions: read 2 Nov 2026; the table, tip column, average odds and the technology claim
2.ProSoccer: about: read 2 Nov 2026; the neural-network claim
3.ProSoccer: home page: read 2 Nov 2026; the "Best of successful predictions for yesterday" list
4.ProSoccer: football archive: read 2 Nov 2026; day pages listed only from 9 Feb 2016 to 8 Jul 2018
5.Ledger file: every number on this page: generated 2 Nov 2026

Updates, corrections and right of reply

Published 2 Nov 2026. We did not contact ProSoccer. Re-check due on or after 1 Dec 2026 (full 30-day window, 31 Oct to 29 Nov). Spot an error? Email [email protected] or tell us on Telegram, and we will correct it and note the change here.

The Price Check #9 · More from The Price Check →

Keep reading

NerdyTips review PredictZ review Forebet review