AI Team vs AI Team Spectate- unrealistic

[goonalan]

So I set up two T20 teams-

1) Sri Lanka, all players set to 'aggressive', and all Attributes distributed (all points spent), all batsmen Skills set to 20+, all bowlers likewise 20+. Basically I created a super-team (at least stat/skill point wise).

2) Papua New Guinea, all players set to 'conservative', and all Attributes set to minimum (no points spent), all batsmen skills set 0 or 1, all bowlers likewise. Basically I created the worst team I could without reducing all skill points to 0.

Then I spectated as the two teams went head to head (no weather interruptions and bright sunshine) on all difficulty levels- Rookie, Amateur, Pro, Veteran and Legend- same ground for all games.

Sri Lanka won every time of course, and for 3 of the 5 matches bowled PNG out, however- and here's my problem. Sri Lanka didn't win by a lot.

Note in all five games SL batted first (3 times they were put in by PNG, twice they won the toss and chose to bat)

Rookie- SL 182/1 vs PNG 123 All out (in 17.2 overs)
Amateur- SL 178/2 vs PNG 145 All out (in 19.2 overs)
Pro- SL 218/1 vs PNG 196 All out (in 19 overs)
Veteran- SL 198/3 vs PNG 193/8
Legend- SL 196/2 vs PNG 176/9

And so my question is why?

Why doesn't a team made up of superstars (all with massively pimped stats- almost all maxed out) destroy, or at least convincingly win every time against a team which has been made as weak (almost) as it can get- what's with the AI?

Also why is it that the 'conservative' team (with terrible skills and attributes) managed to score more boundaries than the 'aggressive' team (with massively pimped stats) in 2 of the 5 matches, and in the other three games still manage to club sixes and fours all over the park?

Also odd is the fact that over 5 games (10 innings) over 90% (both teams obviously) of the players hit a boundary in their first 3 balls faced. In point of fact the highest and fastest scorers for PNG- in all five matches were 6, 7 & 8 in the batting order- even when one of these guys has absolutely 0 (zero) batting skill points, and they faced many fewer balls than many of the lower batsmen.

Is there a way to make the AI reflect reality a little more convincingly- or at least to reflect the respective players skills and attributes?

Any advice gratefully received.

 

[Fantazm]

weirdly the rookie skill level probably gave the most "realistic" result.

 

[cricket_online]

Unfortunately you can't create Chris "The Phantom" Martin in DBC 14. Chances are he would open his account off the 3rd ball he faces by hammering a 150+ kph full delivery on the stumps over 3rd man for a 6.

 

[T.J.Hooker]

The game isn't designed for realistic ai vs ai spectating. P vs P and p vs ai are the main game modes, so that's where BA have devoted their resources.

 

[goonalan]

The game isn't designed for realistic ai vs ai spectating. P vs P and p vs ai are the main game modes, so that's where BA have devoted their resources.

Thanks for the above, and I understand that the game is not designed for ai vs ai, and appreciate that, the problem came about because in game, as in Player vs ai, there seemed to be very little difference between playing an ai team with maxed out skills et al, and playing a team with a bunch of zeroes for scores.

I also picked Sri Lanka and T20 because the SL team has played 100s of T20s and have massive averages- just in case average scores et al were an influence on things.

So I got to thinking why am I spending an age going through ai teams messing about with all of the players stats and attributes in an effort to make them more 'realistic' when the reality is in game it doesn't really make a whole lot of difference.

The question as stated however remains, and I'm not trying to argue with anyone- I love the game, but has anyone found a use for the stats/abilities et al. I want to play be able to play and beat Netherlands and feel like I done good, and then feel like the king of the world when I beat SL- I don't want to know ahead of time that it don't matter whether it was Netherlands or SL it was only the difficulty level that counted.

Is that odd?

So if anyone has any suggestions as to how to make Sri Lanka (or whichever) team more like world champions, and therefore tougher to beat... be my guest.

The other thing is of course some of the stuff just seemed plain wrong, when I made the Sri Lankan team I maxed out Malinga on his bowling skills- straight 22's all the way- in five games he took the joint least lowest wickets and conceded the most runs (not including PNG bowlers). That just made me feel like a fool for going through and changing all the players stats.

Thanks again.

 

[AshleyWilliam]

AI v AI make up a lot of career mode interactions actually. To discount that ai v ai doesn't matter is a little naïve I think.

I think what Goonalan is tying to point out is that AI skills and attributes make absolutely bugger all difference. In Rugby League Live 2 if you played as Melbourne Storm v Cronulla you can expect to win easily cos the sharks are weak - their overall stats of their individual players match their actual onfield abilities. Same goes for AFL Live - play Geelong and you'll have a tough match. When simming results of Ai v AI and watching AI v AI games you'll generally see expected results with occasional upsets.

DBC 14 on the other hand is frankly completely and utterly ****ed up in this regard, to be blunt, compared to other games made by the same developer. Sorry if this offends.

I remember seeing a scorecard posted by the administrator of this forum where he struggles to bowl out Russia playing as a top international team (think it was Australia) in a test match - classic example.

 

[grkrama]

You certainly feel the difference while batting against a russia and an aus or sl, between facing a anderson/irfan /dinda.

Only when you are bowling to AI its a bit hit or miss, i can surely say i have never felt a difference between bowlers in other cricket games, here there is individuality to bowler to a good extent, atleast in higher difficulties.

 

[AshleyWilliam]

You certainly feel the difference while batting against a russia and an aus or sl, between facing a anderson/irfan /dinda.

Only when you are bowling to AI its a bit hit or miss, i can surely say i have never felt a difference between bowlers in other cricket games, here there is individuality to bowler to a good extent, atleast in higher difficulties.

Bowlers seem the same to me when batting regardless of skill/attributes. Tall bowlers are very hard to play however. Irfan is much harder to play than Steyn just cos of his height. But is he a better bowler in real life? No.

At least in former games they had different bowling actions

 

[Langeveldt]

When the game came out I simmed about 20 games, one of which was an Ashes series, on Pro difficulty.

The leading runscorer and highest strike rate came from the dashing English left hander James Anderson.

When I posted these findings I received the usual petulant response from Big Ant asking something like "why would I watch AI games". Well its an option in one of your games, and it forms an integral part of career mode.

I've rather given up on the situation improving, especially finding that many of the sliders make FA difference to anything at all. They can give the game a fancy set of attributes and menus, but it changes nothing. It's actually really insulting coming from a developer who openly mocks other efforts in the genre (yes I was unfortunate enough to buy those too).

I really hope the situation when a human is bowling is a little more encouraging, and I hope these sliders actually mean something. Maybe we can test these outcomes too?

 

[grkrama]

Bowlers seem the same to me when batting regardless of skill/attributes. Tall bowlers are very hard to play however. Irfan is much harder to play than Steyn just cos of his height. But is he a better bowler in real life? No.

At least in former games they had different bowling actions

which difficulty are you playing in, im playing on legend so that could be the case, its visible mostly from veteran and upwards.

I certainly find irfan easier to play than steyn in game. ya he does bounce a bit being a giant and leftarmer in game but its much easier to score of him than steyn. Also other thing would be to check the bowler skil as we are getting get best, and so skilling is certainly varying from team to team being done by different creators and so uniformity is a bit questionable.


also i cant remember the last cricket game that had various bowling actions. Also even though bowling actions are the same in dbc there is individuality to each bowler if you study them a bit interms of deliveries and trajectories which are unique to each bowler due to the skilling, pretty fun against spinners too, you can certainly tell a difference between an ajmal vettori and a hafeez or boyce.Anderson deliveries certainly dipp n swing more than stokes etc.

So from a batting perspective i feel the skills and attributes of bowlers/team i play against hold pretty well.

 

[lastig]

You certainly feel the difference while batting against a russia and an aus or sl, between facing a anderson/irfan /dinda.

Only when you are bowling to AI its a bit hit or miss, i can surely say i have never felt a difference between bowlers in other cricket games, here there is individuality to bowler to a good extent, atleast in higher difficulties.

I agree in this regard. The reason it feels different is that fe: Iran is tall and gets a lot steeper bounce, whilst a shorter bowler will deliver a skiddier delivery. On low slow wickets you can play all deliveries of the front foot. On a quick bouncy wicket you will get into trouble doing that.

 

[AshleyWilliam]

which difficulty are you playing in, im playing on legend so that could be the case, its visible mostly from veteran and upwards.

I certainly find irfan easier to play than steyn in game. ya he does bounce a bit being a giant and leftarmer in game but its much easier to score of him than steyn. Also other thing would be to check the bowler skil as we are getting get best, and so skilling is certainly varying from team to team being done by different creators and so uniformity is a bit questionable.


also i cant remember the last cricket game that had various bowling actions. Also even though bowling actions are the same in dbc there is individuality to each bowler if you study them a bit interms of deliveries and trajectories which are unique to each bowler due to the skilling, pretty fun against spinners too, you can certainly tell a difference between an ajmal vettori and a hafeez or boyce.Anderson deliveries certainly dipp n swing more than stokes etc.

So from a batting perspective i feel the skills and attributes of bowlers/team i play against hold pretty well.

Ricky Ponting Cricket, Ashes Cricket 09, International Cricket 10

 

[Langeveldt]

I've been thinking about this recently and I'm not sure T20 is a format which gives us accurate results anyway. Lets have a look at some other indicators as to what might be going on with the AI in the original study. I've been getting the impression recently that sliders work, but not enough. So you'll be facing a Zimbabwe as opposed to a Russia, or bowling to a Tim Southee as opposed to a Chris Martin. It is impossible to create a player who is useless at something. Now we all know that a Tim Southee can win you a game of T20 with the bat, but he can't win a test match with the bat, so I think we need to sim longer forms of the game, something I will do again.

I also think the in game AI is the same as the AI vs. AI in terms of input and output (I think generally the human bowlers are better than in AI vs. AI however). I don't see why they would code something entirely different for a game mode that hardly anyone uses.

Run rates

Sri Lanka RR - 9.72
PNG RR - 8.33

Similar? Probably resulting in the tight games in this format.

Now lets look at what I think is a more important indicator of AI strength, average per wicket.

Sri Lanka - 108
PNG - 17.7

This is a far more heartening stat, which means something must be working.