Some results from Attribute Testing AI vs AI match

[Beefybeef66]

As mentioned elsewhere, I have ran a quick demo Test to try to gauge how much impact the attributes have on scoring.

Please excuse the silly names. Worst are unsurprisingly the dummy team (Gray-Blue as opposed to Royal Blue).

I didn't record pitch type but think it was hard, with some cracks. Weather was sunny.

After losing early wickets, the hosts started to build a score.

 

[Beefybeef66]

On the second day, the batsman carried on where they left off, setting the dummy team a target of 581.

Wagon Wheel attached, showing shot variety.

[DOUBLEPOST=1482666578][/DOUBLEPOST]In response, the dummy side could only manage 169 runs. They were forced to follow on.

[DOUBLEPOST=1482666654][/DOUBLEPOST]The third (and last) day. The dummy side fell short with a second innings of 103.

Also, interesting that the pitch deteriorated over time (Quick Pitch Wear was on). The pitch became slightly more helpful for spinners but majority of problems for batsman came from pace and bounce.

Final scorecards are attached below - my conclusion would be that the attributes make a moderate amount of difference.

 

[Beefybeef66]

I have attached partnership and batting mentality information.

Unfortunately, though batsmen do appear to behave differently, I am not sure exactly how mentality impacts behaviour - further work here is required. (Will also edit the mentalities to give more of a mix)

It might be that reducing the stats in turn has an impact on mentality so probably worth trying this with more evenly matched teams. I.e. conservative batsmen are try to score more off bad balls. Aggressive seems to have an effect though.

I intend to investigate further - with different pitch types and mentalities. Hope this is useful.

 

[Langeveldt]

Really good stuff, thank you. Confirming what I already think. Attributes matter, mentality matters, but not enough.

I can also confirm that players with max attributes are more aggressive.

I think AI vs AI isnt the ideal way to test in future, (I know I do this), but they serve up a lot of rubbish balls.

 

[Beefybeef66]

Thanks. As Ross has said, you can't really learn anything from AI v AI. But I find it interesting to see how it plays out none the less.

I did run a return match (Green, Soft, Uneven). Similarish results but lower scores (132, 188 v 381/4). Noticed that the AI declares quite early, maybe is based on how far ahead they are rather the potential to bat the other team out of the game. The pitch was tricky to start with lots of seam but was dead even by the second day.

Also with the fast bowlers there were lots of byes.

 

[Offie McOff face]

The declaration seems to be programmed as about a 250 lead. Certainly always seemed to be about that in DBC14, and I suspect it's not changed for this version.

 

[Saeglopur]

If I remember correctly, first innings declarations were always at the 550 mark too, which was always annoying when I was unbeaten on 200-300+, haha! I remember my career team once declaring before Tea on the first day as well - I think it was a Test too, so completely unnecessary, haha!

Really intrigued to see anything else you guys find out about mentality impact. I'll be disappointed if it's negligible. I also really want the AI to be smart about playing to a player's strengths and weaknesses. If a batsman is weak on the drive, he should try to avoid playing that kind of shot because he doesn't want to edge it behind, or a bowler will play more to their best deliveries, etc. The judgement attribute could come into it too, so players with good judgement would avoid their weaknesses more often, things like that. It's probably asking a lot but has anyone seen anything like this or does the AI just play every shot regardless?

 

[Beefybeef66]

Interesting about the declarations, wonder if this is the same in Human v AI. If so, it would be good to see more variation in declarations and some high AI scores.