Incursion Victory Conditions - Err, wot?

So, I just ran a round in Incursion, and lost. These things, they happen. However, the conditions themselves of the loss are a bit out of kilter to me.

Screenies:
Results:

Statistics:

Match ID:
20160612-63e40a34-6eb1-4201-8336-2b0c6687609c

My actual gripe:
Previously, I had thought that Incursion victory was determined in the event of a tie by average team score, apparently this is not the case, it’s being averaged by player scores.

Average by totaled scores would put Blue at 124/5 or 24.8
Average by totaled scores would put Red at 108/5 or 21.6
Blue would win.

However, if one does the average by the lowest + highest divided by 2 to determine the 'average -
Average by high+low / 2 would put Blue at 36+15 / 2 or 25.5
Average by high+low / 2 would put Red at 42+12 / 2 or 27
Red would win (and did.)

Could this be changed to reflect the fact that it’s actually a teamplay mode, and take either highest team score, or actual average of teamscore / players instead of the current method?

Are you sure there wasn’t any damage done to a sentry?

I’ve had a couple of losses where I thought I was winning, until the game ended and I found out they knocked a percentage or two off at the last minute, to take the win.

Not that I doubt what you’re saying, I’m just wondering if you may have missed some obvious cause, inadvertently…?

Yea, I’m certain, they had taken out our first sentry with 1m 30s to go on the map (it was a nice ISIC backdoor) and no other damage was shown to the last sentry. Also, IF it were to be scored based on non visible sentry damage (note score of 50 - 50) it would be nice if it would show it somwhere (e.g. 49.9 to 50) though that wasn’t the case in this instance.

Yeah, the invisible damage would’ve been my second thought… Isac is stupid good at that. A well played Kleese is arguably better.

I feel like the game says average team damage, but I think it actually tally’s total team damage. I’ve never lost a truly tie game, due to team score, so I’m thinking invisible damage is the actual culprit.

I read somewhere that it’s based on average player score, not damage, though I can’t seem to find it now. Mathematically, it’s accurate for this match, I just disagree with the method being used to determine the average.

I’d be interested to see that topic, if you find it again.

I don’t like the notion of averages being the determining factor. I’d rather it be straight team score because the ugly truth of the matter is, you can get stuck on some teams with some really (keeping it PC :sweat_smile:) skill deprived players… And averages reduce the chance of a really skilled player “carrying” the team, and getting the deserved victory for scoring best overall and making his/her total team score higher than that of the opposition with the better “average”. Granted, it doesn’t happen often on a team of casual gamers, but it can, and does.

Well, a bit of searching while waiting for matches has left me with this:

That thread and the accompanying screenshots seem to indicate that it’s first by Sentry life, second by an average score of some kind, then third by 1st blood. This makes sense to me as it fits the victories and losses I’ve had. It still leaves me wanting a shift in the mechanic behind average score determination or just changing it to reflect total team score.

1 Like

This.

Total team score. That’s what I like.

I had this same thing happen to me and a few people on reddit said it was because they got a fraction of damage, but not a whole point and that is how they won.

Somehow, that still seems wrong to me. Thinking back on it, I know at least once during the open beta I got a win based on average team score, though for the life of me I can’t find any official confirmation. (The dearth of available information for Battleborn makes me a touch sad.) It would be really nice to get some confirmation of how things work for this from say… the Devs. I think that a great big data dump out to a wiki or something along those lines would be greatly useful.

In a 5v5 match total team score and average team score are technically the same thing. Where average team score comes into play is when the game is 4v5, as it doesnt count the person that didnt play. So while your team score may be lower, your average score is actually higher since there is only 4 players. I have actually won a game that was 4v5 because our average score was higher, even though we had a lower total score.

Actually, depending on how one does their math, they aren’t. Look at my first post - right below I did a few bits of math. The Sentry scores where the same, my team had a higher score, higher total average player score, but a lower min/max average and we lost.

As far as scores in the conditions for 4v5, I don’t particularly object to the idea that if a person has been dc’d for more then 50% of the round not being counted, but it seems like someone could exploit the team score not counting the 5th player if they are not connected at the end of the round. Having not managed to win a 4v5 round at all, I doin’t know how that would work.

Ex - 5 man premade runs a round of incursion. Sentry scores are tied at the last 30s. If the lowest score person disconnected close to the end of the round and they were not counted, it will/could artificially inflate the team score based on the remaining players. Some numbers to show this idea a bit more clearly:

Team A score 100 points, 5 players, 22,25,28,20, 5
Min Max average - 28+5 = 33/2 = 16.5
Average of all Players = 100/5 = 20
Min / Max average of players if the lowest score disconects - 28+20 = 48/2 = 24
Average of all players = 95/4 = 23.75

Team B score 100 points, 5 players, 15,30,20,17,18
Min Max average - 30+15 = 45/2 = 22.5
Average of all players = 100/5 =20
Min / Max of average if lowest score disconnects - 30+18 = 48/2 = 24
Average of all players = 85/4 = 21.25

In the case of Team A vs Team B, Team A could actually flip the victory in their favor by having the player with 5 points disconnect before the end of the round. Again, some of this is speculation and potentially sky is falling, and would be really great to have an official dev posting explaining the tie break conditions.

1 Like

I’d hope that’s not the case. If you see the score as 50-50 your team’s going to play very differently in the last few minutes of play than if their score is 49-50. If that IS the case the very first itty bitty bit of damage should reduce the percentage to 49 on the interface so people know what the actual situation is.

The game says average, and i would assume that would be the normal way (total score/ players). Taking the average of the best and worst players would be really odd…Your basically making 3/5 players not matter at all. Why should a team that scores 50,49,49,49, 2 lose over a team that went 50, 5,5,5,3?

My thought is there is something not calculating correctly in the end of game average.

I have no idea at point they dont consider a player for the average score. I have had people leave in the last 1 minute and not effect the outcome.

I’ve seen a post where it happened and someone complained about it,this scenario.
Someone should have won by team score but one of the opposing team left and they lost,cuz it’s average player score.
Can’t find the post (actually didn’t look,too lazy) but I swear it exists,I have seen it.
I guess it helps. Guess it’s also why they use that,you can actually still win being down one player,I think you can’t with what you ask,or it’s significantly harder,no?

I agree with what you’re saying there completely, that’s precisely why I made the thread. As you can see in the screenshots I provided, my team lost due to the min / max average computation, as that is the only way I can see that they would’ve won.

Sentries - Tied.
Team score - My team was higher.
Average score counting all players - My team was higher.
Min / Max averages - My team was lower.
Result - We lost.

In the event of players leaving, I’m not entirely sure what would be a best solution. I definitely don’t want to see teams that hold off 4v5 penalized due to a lack of a player being factored into the average, or factored in at a 0. If the average of all players is used and the 5th player is technically a 0 due to DC early or late, then the team that held out 4v5 looses. (Mathematically speaking, pulling a 0 out of an average is ugly at best.) However, if scores are pulled based on connected players at time for the match, it still leaves a loophole that could be exploited to push average scores higher. On that note - if a player DCs from a match and is unable to reconnect due to the match being over, what do they recieve for the match? If the team that they were on wins, do they get a win for the match? Do they get a disconnect with associated lack of reward? These things I really don’t know. I play 'em to the bitter end.

Perhaps a solution would be to look at disconnection times - if they’re in the round less then 5 minutes, discount their score and take a complete average at 4 players based on the players that played the round. If they’re connected for 5 - 15 minutes, calculate them in, but run the average at 4.5 players using the whole score. If they’re in for more then 15 minutes, average all scores at 5 players perhaps?

Again, I see that it could be a loophole benefiting 4 of 5 players at least, potentially all 5 of them depending on the DC / Reward system, but I seriously doubt that anyone has done this deliberately. (I really doubt that anyone would, it doesn’t seem to be a high enough payout, but I still think the victory conditions need a bit of work)