NRL Streaks — What's Hot, and Whether It Means Anything
Current runs by club, and the measured answer to the question the list raises. Updated with each week's results.
Which NRL teams are hot right now?
| Club | Run | Length | Last game |
|---|---|---|---|
| Redcliffe Dolphins | winning run | 8 | 2026-09-12 |
| New Zealand Warriors | covering run | 5 | 2026-08-30 |
| Newcastle Knights | run of overs | 4 | 2026-08-30 |
| Penrith Panthers | winning run | 4 | 2026-09-13 |
| Gold Coast Titans | run of overs | 3 | 2026-08-29 |
| Redcliffe Dolphins | covering run | 3 | 2026-08-29 |
Which NRL teams are cold?
| Club | Run | Length | Last game |
|---|---|---|---|
| Gold Coast Titans | Winning runs (against them) | 6 | 2026-09-04 |
| Cronulla-Sutherland Sharks | against the line (against them) | 4 | 2026-08-30 |
| Penrith Panthers | over the total (against them) | 4 | 2026-08-28 |
| Sydney Roosters | Winning runs (against them) | 4 | 2026-09-13 |
| Sydney Roosters | over the total (against them) | 4 | 2026-08-29 |
| Wests Tigers | over the total (against them) | 4 | 2026-08-29 |
| Newcastle Knights | against the line (against them) | 3 | 2026-08-30 |
| Redcliffe Dolphins | over the total (against them) | 3 | 2026-08-29 |
Does a NRL streak predict the next game?
Of 36 run lengths tested in the NRL, 2 clear the noise band — and about 1.8 would clear it on chance alone, so a count this high turns up about 54.3% of the time with nothing going on. No run length here tells you anything about the next game.
The test is the one the list invites: take every game in the record, work out the run the club was on going into it using only earlier games, then see what happened. The comparison is not 50% and not the league average — it is that club's own rate in that same season, with both the graded game and the run itself removed, because a club on a six-game losing run has a bad season record largely because of those six games and comparing the run against a figure that contains it compares the run with itself.
Even that is not enough. Running the identical calculation over data whose runs have been destroyed — the same clubs, the same season rates, outcomes shuffled within each season, so the true answer is known to be zero — the method still reports an edge of two to five points. That is bias, not signal, so it is measured and subtracted. The Edge column below is after that correction, and the Noise column is the spread across those shuffles. An edge inside the noise is not a finding.
Winning runs
| Run | Next games | Continued | Expected | Edge | Noise (1 SD) | Beats noise? |
|---|---|---|---|---|---|---|
| 1 in a row | 1,000 | 51.7% | 48.4% | +1.1 | ±1.6 | no |
| 2 in a row | 516 | 54.7% | 51% | +0.9 | ±1.7 | no |
| 3 in a row | 282 | 52.8% | 54.6% | -3.4 | ±2.5 | no |
| 4 in a row | 148 | 60.8% | 56.7% | +0.2 | ±4.1 | no |
| 5 in a row | 90 | 61.1% | 58.5% | +1.1 | ±5 | no |
| 6+ in a row | 171 | 68.4% | 62.6% | +3.7 | ±3.5 | no |
Runs against the line
| Run | Next games | Continued | Expected | Edge | Noise (1 SD) | Beats noise? |
|---|---|---|---|---|---|---|
| 1 in a row | 1,109 | 49.1% | 48.3% | -1.6 | ±1.4 | no |
| 2 in a row | 543 | 49% | 47.7% | -0.5 | ±2.1 | no |
| 3 in a row | 265 | 50.6% | 46.8% | +3.4 | ±3.4 | no |
| 4 in a row | 134 | 52.2% | 46% | +4.6 | ±5.2 | no |
| 5 in a row | 69 | 56.5% | 44.9% | +10.9 | ±5.3 | yes |
| 6+ in a row | 71 | 45.1% | 43.8% | -0.9 | ±6.7 | no |
Runs over the total
| Run | Next games | Continued | Expected | Edge | Noise (1 SD) | Beats noise? |
|---|---|---|---|---|---|---|
| 1 in a row | 1,083 | 50.5% | 46.7% | +1.9 | ±1.3 | no |
| 2 in a row | 541 | 47.3% | 46.5% | -1.9 | ±2.3 | no |
| 3 in a row | 255 | 50.2% | 46.5% | +2.5 | ±3.4 | no |
| 4 in a row | 127 | 51.2% | 45.4% | +4.9 | ±4.2 | no |
| 5 in a row | 65 | 55.4% | 45.4% | +6.8 | ±6 | no |
| 6+ in a row | 68 | 47.1% | 47.9% | -3.4 | ±6.4 | no |
Why a hot streak is not an edge
A run is public. Everyone can see it, including the people setting the line, and the line moves to absorb it — a club on four straight wins is priced as a club on four straight wins. What is left to bet on is whatever the market got wrong about the run, which is a much smaller and much noisier quantity than the run itself. That is why straight winning persists in these tables while covering the line does not: winning is about the team, covering is about the team relative to its price.
The break-even strike rate at standard −110 pricing is 52.4%, not 50%. A run length that continues 51% of the time is a losing bet made at a profit-looking number.
More NRL
Built 2026-10-12. Across all four codes, 21 of 144 run lengths clear the noise band against about 7.2 expected by chance.