Berserk as a Conditional Advantage

On Arena tournaments, when both players don't berserk, they are under equal conditions. The data on the chart above proves that winrate against the same-rated opponent (within a +-50 rating point range) is approximately 50% for every group of players. Higher-rated players show a near-60% winrate because this sample is the smallest.

In this article we will look at the probability of winning your next Arena tournament game across 3 other scenarios: you berserk, your opponent does, and you both berserk. All games were analysed from White's perspective in order to clearly identify the scenarios described above and avoid counting the same game twice.

Winrate When White Berserk

Firstly, we will look at the probability of winning when you will use berserk the next time. As expected from the previous analysis, winrate from lower-rated opponents starts from 58.3% for players rated below 1200, and reaches 88.8% for the strongest group of blitz players – almost only 1 lost game out of 10.

What is more interesting, the lowest rated group has the highest winrate (50%) berserking against the opponents with a similar rating. On average it is approximately 40% for the rest of the players pool.

Berserking against the higher-rated players is not a good idea if you want to win: around 20% for players rated below 1600, 27% – for players with 1600-2400 ELO, and even the strongest players win only a third of the games.

Winrate When Black Berserk

In most cases you have higher chances to win when your opponent berserks. Only players rated above 2400 show a decreased winrate when being berserked compared to games when they themselves berserk – a drop of 8.4%, to 80.4%.

When matched against an opponent with a similar rating, players below 1200 have a lower winrate – 45.7% – than when they berserk themselves – 50%. The main reason may be psychological pressure, something more experienced players can better cope with. For other rating groups the picture is the following: 60% winrate for players rated 1200-2000, and around 64% for players with 2000+ ELO.

All rating groups have higher winrates playing against stronger opponents who have half of the clock. Moreover, in this scenario players rated 1600 and above show a winrate roughly 50% higher in relative terms compared to games where they use berserk themselves: 39.1% for the 1600-2000 group, 49% for 2000-2400, and 48.8% for players above 2400.

Winrate When Both Berserk

When both players berserk, the results sit between the two scenarios above. Winrates against lower-rated opponents remain high across all groups: 59.9% for players below 1200, rising to 83.9% for players 2400 and above. Against equally-rated opponents, the figures drop noticeably – 38.4% for the lowest-rated group, rising to 62.5% for players above 2400. Against stronger opponents, both-berserk winrates are the lowest of all three scenarios: 17% for <1200, and 41.2% for >2400.

Stronger players still win more often, but the edge that berserk was supposed to create has disappeared entirely. The data confirms what was suggested previously: berserk is a conditional weapon, and its condition is asymmetry.

Conclusion

The results show that 2000+ players have the most to gain, as their winrates are consistently high regardless of the scenario, and they can berserk more confidently, maximising tournament score. For the majority of players, the picture is less encouraging. Berserking reduces available time in games that were already difficult. Sub-2000 players who berserk against stronger or equal opponents are, statistically, making their situation worse rather than better.

All in all, berserk is not a universal tool. It is most effective when used selectively, against weaker opponents, and – what is more important – when your opponent does not berserk back. Understanding it allows you to convert berserk activation into tournament points rather than time loss.