Statistics from 256 SolForge drafts

I've been playing SolForge for about a year, if memory serves. A few months ago, I started recording data on the draft tournaments I played, both to try to assess my skill level, and to see if I could learn anything about the metagame. The results after 256 tournaments are visualized below.

I previously published my data after I'd reached 100 tournaments. Each chart below is accompanied by a link back to its earlier counterpart, for easy comparison.

Shortly after I published those earlier results, SolForge's tournament structure changed. The casual draft queue was added, and the payouts were altered. People were concerned about what it meant for their ability to draft regularly.

During my first 100 drafts, I had to take a break at one point to replenish my ticket supply. But since the changes, I've never been unable to draft—for free—when I wanted to, although more than once I've had to multi-forge my excess cards. My win ratio is only a little higher than it was before. I can't say if the new payout schedule is good for other players—I understand that players who regularly go 1-3 are disappointed by it. But it's working fine for me. I feel confident saying that if you have a win percentage in the mid-60s, a mostly full card collection, and are willing to spend your extra cards and silver on nothing but entering tournaments, then you too can be "quasi-infinite".

See the Reddit thread for discussion of this article. If you're just here to download the spreadsheet, skip to the end.

Finishes

Win ratio

100 chart graphic

63.8% is the more precise number this time, up by 2 full percentage points from last time. Can't complain about that.

Finishes

100 chart graphic

The 3-1 bar is gradually pulling away from the 2-2 bar.

Win ratio by day of week

100 chart graphic

Mondays continue to be bad days for me, and there are enough drafts on record that I think this is probably not just a fluke. One possibility is that there are more skilled players around on Monday, but that doesn't make sense to me; I'd expect skilled players to play frequently, not just one day a week. The more likely explanation is that there's some psychological difference in me on Mondays. Am I gloomy about the start of the workweek? More willing to take risks that don't pan out?

Cumulative win ratio at intervals

100 chart graphic

The previous version of this chart wasn't clear, but this one shows definite improvement. Each of these points represents my win ratio since the beginning at a certain point in time, so it's likely to continue to plateau. I might choose a different approach for this chart next time.

When going 3-1, which round was lost?

100 chart graphic

Last time, this chart showed that I was much more likely to lose rounds 1 or 2 than rounds 3 or 4. This time, the chart is a lot closer to flat. We'll keep an eye on it, but I don't expect to see anything surprising.

Factions

% of decks containing each faction

100 chart graphic

Compared to last time, I'm drafting Alloyin less often and Tempys more often. I do think Tempys has gotten a little stronger recently... among other things, it's picked up a couple of powerful Allied heroics in Spite Hydra and Gemhide Ravager.

% of decks with each faction pair

100 chart graphic

The biggest change here is that the Nekrium/Tempys slice of the pie has doubled in size. Accordingly, as we'll see in a moment, my win ratio with that faction pair dropped by over 10 percentage points.

Win ratio with each faction

100 chart graphic

All of these bars have grown slightly taller, but the relationship between them is pretty much the same. Comparing this chart to my most-picked factions suggests that I'm probably under-drafting Nekrium and Alloyin.

Win ratio with each faction pair

100 chart graphic

Nekrium/Uterra, which was already very high, now towers above the other bars. It's not just because of individual high-powered cards like Dysian Broodqueen or Bramblewood Tracker—though those certainly help—but also because those factions, in general, have the fattest creatures. If I could give just three words of advice to novice drafters, they would be these: fat wins games.

Win ratio with each faction, by format

100 chart graphic

This chart excludes Weekend Warrior drafts, since they may use non-standard card pools.

It seems like every faction except Tempys has been growing in strength over time, although probably what we're really seeing is just my win ratio getting better over time.

Win ratio with each faction pair, by format

100 chart graphic

This chart excludes Weekend Warrior drafts, since they may use non-standard card pools.

The tall bar for N/T prior to set 3 is mostly a fluke because I didn't play that faction often. I'm surprised to see the big jump for A/T since Imprisoned Heralds. I guess I've been doing better with it, but I can't really explain why.

Rarity

It's true, I once drafted a deck with nine heroics. Not junk heroics either; highlights included Borean Windweaver, Frostfang Maiden, Hammerfang, Emberwind Evoker, and two Gemhide Ravagers. Yes, I went 4-0.

The charts in this section exclude High-Rarity Draft tournaments, since the number of legendaries received in that event was fixed and not relevant to the normal drop rate. I didn't bother to exclude Guaranteed Legendary Draft, because I only played it once.

% of decks with at least one legendary

100 chart graphic

Last time, 22% of my draft decks included a legendary, but now it seems I had just been unusually fortunate.

This chart only accounts for legendaries that I actually drafted, not ones that I was offered but passed, so you could pull legendaries a little more often than this if you wanted to. In an event that will surely never occur again, I was once offered Binben, Lightning Herald twice in the same draft, and passed them both. It's a build-around card that didn't really fit with the deck I was putting together.

Legendary drop rate by drafts played/day

chart graphic

If this chart is hard to understand: on days when I played six drafts in the same day, I acquired legendary cards at a rate of one third of a card per draft.

It seemed to me like I started pulling legendaries less frequently once I started drafting more frequently. This chart is a clumsy way to try to investigate that. But it doesn't support the hypothesis at all. I'll keep tracking this just for kicks, but it seems likely that the effect was all in my head.

Win ratio with vs. without legendaries

100 chart graphic

After 100 drafts, I was actually doing worse with a legendary than without. Now I'm doing better, but it's still not a huge difference—not the kind that would make you worried that legendaries are overpowered.

Win ratio by number of heroics

100 chart graphic

This chart is a lot more boring if you ignore the extremes, where the sample sizes are low. Even so, it might suggest that it's a little better to have more heroics (which is a contrast to the earlier version). I can certainly think of several heroics that are snap picks if you see them, and Bramblewood Tracker in particular is probably the best card in draft right now.

Card Type

When I posted my first set of results, people were surprised at my high average number of spells per draft (7.21). I took their advice to heart and was able to shave almost a full card off that average. I might be able to go lower, but probably not much... and there's such a thing as having too few spells, too.

Win ratio by number of spells

100 chart graphic

It looks like the optimal number of spells in a draft deck is four or five, though I wouldn't say we have a large enough data set or a sufficiently pronounced effect to make a rule out of it.

Average spells per draft by finish

100 chart graphic

It's perplexing how these numbers go up and down, but I think the general message is still that having fewer spells is better.

The Elite tournament

Thanks to skipping the December Elite constructed tournament, and playing frequently over the holidays, I'd amassed enough Elite tickets to enter January's Elite draft tournament four times. Unfortunately, my results were poor:

I started recording my drafts because I wanted to know if I was good at SolForge. Now, with a 64% win ratio, but such a disappointing performance in the Elite tournament, I'm less certain about it than ever. I've been thinking about this in the days since the tournament, and I've had two insights.

The first is that two of my Elite decks were with the two faction pairs that I've historically done the worst with (A/N and T/U), whereas my best performance was with my best pairing (N/U). Normally I let my factions be decided by whatever good cards I see in the first few picks. That's well and good when I'm doing a normal draft, but in the Elite tournament, with more at stake, I should have forced a faction combination that I know I can do well with. I collected this data, and then I didn't use it when I needed it.

My second realization is that I rarely feel like I deserve either my wins or my losses. If I see good cards in the packs, and draw more levelled cards in the games, then I win, and if not, then I lose. The gaps in power—especially between cards of different levels—are usually too great to overcome otherwise.

(Why isn't my win ratio 50%, then? I don't know. Sometimes you do get the opportunity to overcome bad luck with smart plays. And often I can tell that I'm playing against an inexperienced opponent. Is 65% just a win ratio that means "I'm not new to this game?")

The spreadsheet and script

The ODS spreadsheet on which this data is based is available here, as is a blank copy for your own use. Some columns have been added and some formulas have changed since the previously released version. If you were using that earlier version, I think the safest way to transfer the data would be to copy and paste a column at a time. Not all of the new (or old) columns may be of interest to you; it's fine to hide any that you're not using.

The Ruby script that I wrote to generate the charts in this article is also available. For notes on its dependencies and how to use it, see my comments in the previous article. Note that it now expects the CSV file to be named "sf_drafts.csv" (for little other reason than my own natural programmer's capriciousness). Bug reports are welcomed, and, if you're lucky, might actually get worked on at some point.