Eco Advantage and Round Outcomes

 in Categories CSGO, Explainers

Counter Strike is a game dominated by material advantages, and one of the key ways that is expressed is in the balance of the economy. The Performance Above Expectation and Win Percentage Added Per Round statistics both use research into exactly how the economy effects winning chances in the game, but they don’t expose some of the underlying research.

Here I’m presenting some of the data that’s unique performance metrics are based on, and how they influence round outcomes.

The most basic question is how much does an economic advantage improve round win rates? Obviously a team on a full buy is a big favourite to beat a team on full eco, but outside of that obvious scenario there are varying degrees of advantage.

The graph below shows how a team’s win chances are changed by their relative budget deficit or advantage.

There is a solid linear trend, and we can deduce that roughly every $10,000 of advantage creates about a 15% boost to round win chances. There is however an obvious wrinkle between the $10k to $20k range both in the deficit and advantage.

I haven’t tested the theory precisely but this is probably something to do with one team acquiring a specific item that improves their chances. If you split the data into T and CT sides it appears across both, and although the map data is noisier because of the smaller sample size it appears there too (unfortunately there’s nothing else interesting about splitting the data across sides or maps that isn’t attributable to noise).

Remember it’s not $10k to $20k overall expenditure, it’s that much advantage, my feeling is that it may signal a band of eco advantage where the AWP tends comes into play. Unpublished preliminary research shows the AWP can make round win chances more polarised, so these wrinkles would match that conclusion. Either way the movement off-trend is only small.

Regardless of this diversion, this only tells us about win rates. Although it helps us put a number on the precise chances that’s not particularly useful information useful when watching a game.

What we need is more specific round outcome data, so that we can build an expectation of how a team should perform with a big lead. This graph illustrates expected outcomes from a round in terms of the relative balance of surviving players on each team.

The legend at the bottom indicates the balance of players at the end of the round from the point of view of the team that started with the budget balance illustrated on the left axis. The bottom axis is the percentage chance of each player balance being arrived at. The data is a bit noisy so trends aren’t cast in iron, if you look carefully you can see some odd marginal behaviours but overall it’s reasonably reliable.

As an example, a team with a $10k advantage has about a 10% likelihood to end the round with a -3 player balance, but about a 20% chance they end the round with +3 players.

You can see the logic of big buys against an eco illustrated clearly here, there is over a 50% chance a team with a $25k advantage loses only 1 player, and around a 75% chance they only lose 2. I haven’t included bigger margins here as the set gets very noisy, but at even larger margins the trend continues.

There is a subtlety to bear in mind here, just ending the round with a player advantage doesn’t always mean you win the round. There are examples in every tournament of teams losing on time for example.

Practical Applications

When watching a game, taking a look at the total value of team buys as a round starts and checking what the likely outcomes to the round are, you can get a feel for when a team is exceeding expectations or not beyond whether they simply win or lose.

If a team is consistently beating expectations in rounds they aren’t favoured to win this shows up in the overall battle for economy, particularly for T side teams that have that extra potential to disrupt the delicate CT economy.

Following this clearly illustrates when a CT team is potentially vulnerable and a stronger picture of exactly why they might be in trouble after losing one gun round when they have won a series of rounds against T-sde ecos and force buys.

An additional application could obviously be the creation of a statistic based on beating round expectations, a sort of success rate that goes beyond simply winning a round. To do that though this simple overview needs to include more things like bomb plants for T side, and to consider scavenging weapons or other actions that interact with the wider context. Much of this is already included in WPAR, but that is an individual player stat.