Flesh and Blood (FaB) has rapidly gained popularity in the trading card game (TCG) community, known for its intricate strategy, rich lore, and dynamic gameplay. As someone who was introduced to this captivating game earlier this year, I’ve had the pleasure of engaging with a fantastic community that shares a passion for exploring its depths. With an ever-expanding roster of heroes and constant shifts in the meta, a critical question for competitive players is: how do different heroes perform against one another?
In this analysis, I’ll uncover the matchups between heroes, highlighting which ones cluster together and identifying the unique characteristics of each cluster. While seasoned players may have an intuitive grasp of these matchups, a data-driven approach can provide invaluable insights into the interactions between archetypes, helping players gain a competitive edge in tournaments.
To dive into this, I ran an analysis using open data from Talishar (an online platform where players can battle with their favorite heroes), made available by Fabrary (deck builder and repo), to figure out which heroes have similar matchup profiles. I applied hierarchical correlation clustering to a matchup win rate matrix to uncover patterns in how heroes perform against each other in the Classic Constructed format. Spoiler alert: the results are pretty fascinating, especially if you’re keen on deckbuilding and preparing for the next big tournament!
Data source
Let’s start with the data. Talishar is a well-known site where FAB players test their decks online. It’s a treasure trove of information for analyzing hero matchups because it gives us real-world win rates across a huge variety of matches. From these win rates, Fabrary provides a matrix — a grid that shows how well every hero does against every other hero in the game. Here is the link to the page with the data and a display of the single matchup win rates.
For this analysis, I focused on a specific time period to capture the most relevant part of the current Rosetta meta, i.e., October 2024. The goal was to dig into the relationships between heroes, which ultimately gave me some surprising insights into the game’s current state.
A Quick Dive into Hierarchical Clustering
Okay, so what’s this “hierarchical correlation” stuff all about? Without getting too technical, it’s basically a way of grouping heroes based on how similar their matchup performances are. I took the win rates from the matrix, correlated the hero-hero matchups and ran a clustering algorithm that groups heroes who have similar strengths and weaknesses.
For example, if two heroes tend to win (or lose) against the same opponents, they’re likely to end up in the same group regardless of their overall win rate. This helps us see which heroes are alike in terms of how they match up against the rest of the field, giving us a clearer picture of how the meta plays out.
The Clusters: Who’s Like Who?
Below are the results of the hierarchical clustering. How to interpret the plot? Heroes that are close to one another tend to have similar matchups. The shorter the lines connecting them, the higher the similarity (see Uzuri and Arakni for a case of very similar heroes).
I will now break down the most significant clusters, providing also the average cluster win rates against each hero, exposing strengths and weaknesses.
1. The Brutes: Levia, Kayo, Rhinar and… Vynnset?
Brute heroes specialize in heavy-hitting attacks and disruption. They excel at creating awkward turns for opponents by using Intimidate or blood debt mechanics to pressure defenses. Rhinar, with his ability to strip away an opponent’s blocking options, and Levia, who can hit hard if her blood debt is managed, share a similar high-pressure, aggressive playstyle.
While they have it easy against Prism, thanks to their high rate of Phantasm-poppers, they don’t seem to shine against heroes who are very good at blocking more effifiently and accumulate value, like the newcomer Florian or the good old Viserai setting up their runechants.
The only ‘anomaly’ in the cluster is Vynnset. Although not a brute but a shadow runeblade, it appears to have many gameplay characteristics typical of a brute.
Here are their average matchup win rate, from most to least favorable:
2. The Lightning Instants: Aurora and Oscilio
The Lightning archetype, focusses on speed and tempo. Aurora goes wide by chaining several attacks fuelled by instants, while Oscilio adds a mix of amped arcane damage. Both heroes excel at maintaining high-pressure situations that overwhelm opponents quickly.
Not keen on blocking and devoid of poppers, unlike for Brutes, Prism seems to make her way pretty easily. Being susceptible to on-hits, also Azalea is able to stop them. On the flipside, they seem to be very good at dealing enough damage before finishing the deck, as can be seen from the Uzuri and Arakni matchups, and they can easily deal quickly against combo decks like Kano or Teklo.
3. The Ninjas: Zen and Fai
Ninjas like Zen and Fai rely on chaining together small, quick attacks to wear down defenses. Zen uses his combo lines to continuously threaten follow-ups, while Fai builds long chains using his Draconic cards. Their relentless, go-wide playstyle makes them difficult to defend against, especially for heroes that can’t keep up with their tempo. Given the recent ban of some key Zen cards (luckily) and Fai not being the most favoured hero, the matchups of these Ninjas are not the greatest. The only big exception being one of the top heroes of the meta, i.e., Enigma. Why is the matchup favored? The wider one goes, the more unlikely the opponent is at preventing their wards to pop.
Among the worst matchups are assassins preventing their long combo lines via discard, Guardans with on-hits and higher throughput Brutes.
4. The Assassins: Uzuri and Arakni
Assassins like Uzuri and Arakni thrive on disruption and precision, using stealth and trickery to outmaneuver their opponents. Both heroes focus on manipulating the flow of the game, often relying on attack reactions and on-hit effects to control the opponent’s decision-making. Uzuri specializes in swapping attacks from hand, catching opponents off guard with unexpected threats, while Arakni uses contract mechanics to chip away at the opponent’s deck and gain incremental advantages. Their disruptive playstyle forces opponents to play cautiously, as any misstep can lead to devastating consequences.
With matchups on average more favourable than that of ninjas, they are weak against illusionists. On the flipside, the fear way less the non-blocking brutes.
Among the assassins it is interesting to note that Nuu did not end up in the same matchup cluster – althoug still close. I assume this distance is driven by the difference in equipment and the greater amount of reactions pumping up its attacks. This may ultimately drive slightly different outcomes with increased hand disruption (in primis with the Mask of Recurring Nightmares) and surprise damage. As to what makes her matchups more similar to those of Boltyn, I will leave this interpretation to the reader.
5. The Combo decks: Kano and Teklovossen
Kano and Teklovossen both belong to the combo archetype, but they approach it in very different ways. Kano focuses on burst damage with arcane spells, often finishing games in a single explosive turn by casting spells at instant speed, catching opponents off guard. Teklovossen, on the other hand, plays a more defensive, long-term game. This mechanologist hero builds up resources and sets up powerful armor pieces over time, preparing for a late-game strategy where they unleash the devastating Mechropotent. Teklovossen’s strength lies in its ability to survive while assembling its powerful endgame combo, creating a formidable challenge for aggressive opponents.
While very different, both these decks need time to set up their combos, and there’s nobody who likes more the fact of not being attacked than illusionists setting up a devastating board state. Ninjas also tend to be faster, while assassins don’t seem to struggle in disrupting them and emptying their deck from key cards. On the flipside, they tend to have the upper hand against one of the currently favoured heroes, Florian. This is likely because of the latter lacking disruption and relying on sending earth cards to the graveyard in part via blocking.
6. The New Earth Heroes: Florian and Verdance
Florian and Verdance are masters of manipulating Earth energies, focusing on a unique strategy that revolves around sending Earth cards to the graveyard and later banishing them to unlock powerful effects. Both heroes share a mechanic where once more than eight Earth cards are banished, their hero abilities activate, providing significant value in the form of additional auras and arcane damage upon life gain. Although one is a wizard and the other a runeblade, what makes them similar is the strategy revolving around taking the time to get their abilities online to gain extra value in later game stages. Until then, their hero ability is nonexistent.
As with combo decks, illusionists like Prism and Enigma shine against them as they prioritize setting up a board full of auras and allies. Go-wide ninjas are also able to poke them before they can get their value back from their ability. Nevertheless, their matchups are mostly favorable, in particular against brutes, guardians and warriors.
As a final note, Teklovossen appears to be favored in comparison to other matchups, as he is supposedly given more time to prepare for the late game.
A note on higher-level clusters
Small clusters can form bigger clusters, telling us something about an overlap in the underlying strategy. An apparent case is that of the Earth cluster (Verdance and Florian) and Combo cluster (Kano and Teklovossen) , which both prioritize setting up their strategies over time. For example, both pairs benefit from having the game extend into later stages, where their full potential can be unleashed, making them more vulnerable to aggressive early-game strategies.
I wil leave the interpretation of other higher-level clusters to the readers, adding a note of caution as sometimes noise can give false signals the higher up one moves.
Limitations and What’s Next
Like any data-driven analysis, this one comes with its own limitations. The data is based on Talishar matches, which may not fully represent all competitive FAB play, especially at high-level events. Player skill and deck tuning can also create variances in matchup outcomes. Plus, the game’s meta is constantly shifting with new releases and updates, so this is more of a snapshot in time rather than a permanent truth.
Nevertheless, some core hero archetype mechanics remain more or less constant. Such analysis should serve as an empirical indication of what each archetype is strong/weak against when picking up a hero.
When it comes to player skill, while certainly its role is very important, knowing which ‘default’ one starts from based on the current standards is a very useful information. That is to say, one can win with an unfavored hero, but the level of skill and luck needed will be higher than when playing a top tier deck.
What’s next? It would be exciting to track how these clusters change over time as the meta evolves. A deeper dive into time-based shifts could reveal how heroes rise and fall in relevance and help players stay one step ahead of the competition.
Final Thoughts
Running this analysis has deepened my appreciation for the intricate relationships between heroes in Flesh and Blood. By clustering heroes based on their matchup similarities, we can observe patterns that reveal how the meta operates — and how you can prepare more effectively.
Understanding these clusters can be a game-changer for players during deck-building. For instance, if you know that a certain class struggles against a specific archetype, you might include targeted tools in your deck to counter them effectively. Adapting your strategy based on matchup insights can provide a crucial edge in tournament settings.
Whether you’re an aggressive player aiming to exploit weak defenses or a control specialist planning to outlast your opponents, there’s much to learn from this data. How will you use these insights to enhance your gameplay? That extra layer of understanding could be the difference between victory and defeat at your next tournament.
Acknowledgments
I would like to extend my thanks to the Trieste Flesh and Blood community at the Arcana LGS for their continuous support and encouragement. Special thanks to Sebastiano, Davide, Piero, Goran and Fuad, whose insights and thoughtful feedback on the initial versions of the manuscript helped shape and refine this analysis.
