Levski Sofia 1–0 CSKA 1948
Final score · Half-time 0–0 · All of yesterday's results
Our backtest says that, with the sharp market making Levski Sofia the most likely outcome at 63% (fair odds 1.60), Unibet offered 1.62, a gap of +1.5% above the fair price (Value flag). This was reconstructed on 26 September 2026, not recorded at the time. Recent form — Levski Sofia: WWLWD; CSKA 1948: WLWLW.
After the whistle
Match stats
| Match stats | Levski Sofia | CSKA 1948 |
|---|---|---|
| Possession % | 51 | 49 |
| Shots | 12 | 7 |
| Shots on target | 3 | 2 |
| Corners | 3 | 7 |
| Yellow cards | 2 | 2 |
| Red cards | 0 | 0 |
What the market said before kick-off
- Levski Sofia win 62% happened
- Draw 25% did not happen
- CSKA 1948 win 14% did not happen
Pinnacle closing price with the margin removed, next to what happened. It is the market's view, not a tip: an outcome priced at 30% happens about 3 times in 10.
Next matches
Line-ups, events and match stats: API-Football.
Starting line-ups
Levski Sofia 4-2-3-1
- M. Soula MF
- J. Perea FW
- A. Oko-Flex FW
- C. Makoun DF
- Maicon FW
- Hevertton Santos DF
- M. Grujić MF
- K. Dimitrov DF
- S. Vutsov GK
- G. Trdin MF
- A. Mitkov MF
Regular starters not in the XI
- Álex Centelles — not in the starting XI (started 5 of the last 5)
- Serginho — not in the starting XI (started 5 of the last 5)
CSKA 1948 4-2-3-1
- P. Vitanov MF
- G. Rusev FW
- B. Tsonev MF
- H. Cuellar DF
- M. Zemzemi MF
- O. Gašević FW
- Elias Franco FW
- A. Hoffmann DF
- D. Medina DF
- A. Bozhev GK
- Frédéric Maciel FW
Regular starters not in the XI
- Jules Meyer — not in the starting XI (started 4 of the last 5)
- Atanas Iliev — not in the starting XI (started 4 of the last 5)
Not in the XI does not mean injured: it can be rotation, a suspension or a knock. We only compare with the team's last 5 line-ups.
Pinnacle, margin removed (home / draw / away): 62 / 25 / 14% when the line-ups came out, 62 / 25 / 14% at kick-off. The price barely moved after the line-ups.
In the big leagues, line-up news is usually already in the price by the time it is official.
The most likely result is not the same as value: value is where a bookmaker pays more than the fair price, and the two can point to different outcomes.
Value opportunity before kick-off Value flag
Market discrepancy: before kick-off, Levski Sofia (1) was priced at 1.62 by a soft bookmaker, above the fair price of 1.60 — a +1.5% gap. An observation about the price, not an instruction to bet.
A single result where a bookmaker pays above the fair price — separate from the main prediction below.
Most likely result
| 1 | X | 2 | |
|---|---|---|---|
| Sharp, recorded | 63% | 23% | 15% |
Other estimates
| Our model | 61% | 23% | 16% |
|---|
Sharp, recorded = Pinnacle when we logged the match (it can be days old). Our model = our own estimate, which on average predicts worse than the market.
Confidence reflects only the sharp probability of the selection: high from 50%, medium from 40%, low below. It says how likely the result is, not how big the price gap is. Model agreement shows how close our own model is to the sharp market on this selection — when it is Low, the market and our model disagree, so treat the flag with more caution.
Value lifecycle Settled
“Detected” is the moment the flag was first shown on the site and written to our ledger (recorded live since 26 September 2026). For earlier matches the flag was reconstructed afterwards from the prices we had stored, and the time shown is when the rule would have fired: a backtest, not a live record. When the gap closes, we say so instead of pretending the value is still there.
Match context
The last 5 head-to-head meetings we tracked split 4–1–0 (Levski Sofia wins – draws – CSKA 1948 wins).
Levski Sofia scored 2.2 and conceded 1.2 across their last 5 matches in our results database; CSKA 1948 scored 2 and conceded 1 in 5.
Since opening around 1.54, the sharp market price on Levski Sofia has drifted to 1.63 — the market has moved against this outcome.
These figures come from our own tracked results and odds snapshots for this fixture — not from a generic stats feed.
Market-relevant trends
We show the trend and whether it is already in the price. Most sites show the streak and let you assume you should bet it.
Why this price?
Recent form and head-to-head from our settled results — context behind the fair price. The price itself comes from the sharp market, not from these alone.
Public, timestamped record
Backtest: reconstructed on 26 September 2026 by applying today’s rule to the prices stored before kick-off. It was not shown in this form at the time; its seal was created at the reconstruction, which for matches already played was after the match.
This hash fingerprints the prediction as recorded: teams, kick-off time, our model probabilities, the Pinnacle prices recorded for it (at logging or, if there were none, the first quoted after logging and before kick-off, which can be written in after the match) and the log time. A later edit to any of them would change it (a rescheduled kick-off changes it too). It does not cover closing prices, the result or the value-bet record above; value flags recorded live have their own seal in the public record. Verify in the public record →
Every day a snapshot of the full public record is published to a public Git repository and timestamped with OpenTimestamps. Timestamped snapshots are never overwritten and the Git history shows any difference, so a later change to a past row is visible by comparing two snapshots. The corrections file lists the corrections we announced. Verify on the public log →
Since 20 June 2026 the public record is snapshotted daily (07:40 Bucharest time: 04:40 UTC in summer time, 05:40 UTC in winter) to an external log. A prediction is externally timestamped before kick-off only if it was in a snapshot taken before the match; predictions logged on match day, or whose Pinnacle price was added after logging, first appear there after kick-off. Value flags have been recorded live only since 26 September 2026; earlier value flags were reconstructed on that date.
Methodology
The fair price comes from the sharp market itself: we strip the margin from Pinnacle-style odds — the most accurate public predictor in existence — and use our calibrated Elo + form model as a cross-check and as the fallback where no sharp price exists. Value is flagged only where a softer bookmaker prices a selection above that fair price, never on long shots (implied probability under 30%), and never beyond a plausible band. Most of the time there is no value — and we say so.
Measured by TipsAudit
- Most accurate football prediction site? Ranked against the closing line
- Is Statarea accurate? Audited against the closing line
- Is ProSoccer.gr accurate? Audited against the closing line
- Is FootballPredictions.net accurate? Audited against the closing line
- Independent tipster audit: verified track records
- How to check if a tipster is legit
- Can you verify a Telegram tipster? What we found reading their tips
- Are free football prediction sites accurate? We measure them in public
- Which bookmaker has the lowest margin? Live index
- Free value bet finder for football
- What is closing line value (CLV)?
- Free football predictions dataset with closing odds (CSV)
- Our full track record: every prediction, losses included