AI field setting in the context of DBC17 Academy settings

I wonder if there is some way within the academy to assign specific field sets to specific bowlers. So when you're creating a player, you're setting instructions for the AI to adhere to and as a default for when you're playing that team/player... Then in the overal match menu before the game you might set the default "team" strategy. Attack, Defend or Moderate (mix of the two) ...then you start the match.

Here's how it could work...

In the academy, maybe it's as simple as "3 Slots" for each created bowler. Default Attack Plan, Default Moderate Plan, Default Defensive Plan. You might like more than three, but lets keep it simple for now. Lets create a "plan" for our created bowler:

Attack "Plans" to select from might be:
  1. Attack The Stumps
  2. Attack the edge of the bat
  3. Attack with shorter deliveries
  4. Attack the legs of the batsman
Moderate "Plans" might be:

  1. Short of a length
  2. Run restriction
  3. Wide outside off stump
  4. 4 - 2 Plan (4 Regular Deliveries, 2 slower deliveries, standard variation plan)
Defensive "Plans" might be:

  1. Full and straight, no variations
  2. Deliveries at the body
  3. Fielders in the deep
  4. Wide Yorkers
You could literally have a dozen, these are just basic "plans" that have say, half a dozen field settings tied to them that the AI can select from or are "recommended" to you when you're in control (cite; the "Flag" option or reduced menu options as suggested earlier). When you select the overal "Team Plan" before the match begins, all your bowlers default to their predetermined individual plans as you've set up in the academy. That way, you can have individual bowlers with plans, along with an overall team goal. Many international teams are more attacking, or defensive but it allows you to set a specific plan for a specific match and have players respond individually to that specific team goal.

What this essentially does is two fold:

  • It allows you to not have to worry about changing fields all the time, when you bring on a bowler, you've already set a "plan" for the team and an individual "plan" for that specific bowler. You just have to bowl to that plan, then you can change the fields manually if that plan isn't working. But there's already a distinct set of plans set for individual players that are unique to that bowler. Furthermore, it's educational! Since your bowlers can't utilize all the plans individually, they have strengths and weaknesses. X bowler might just have a bunch of really aggressive attack plans and very few defensive plans, so bringing him on in a defensive situation has cause-and-effect. You might get a wicket, but you're leaking runs through the field if you don't change it up.
  • It allows the AI to have an objective set automatically for all scenarios. If they're getting their ass kicked, they can default to a defensive strategy automatically and have bowlers respond in kind and fields that match. If you set an "Attacking" team plan before the match starts, the AI can then assess the team, skills and select a strategy to counter that.
Perhaps part of the overal team "Plan" also involves selecting a side after you've seen the pitch report and made an assessment based on those factors too?


This is an interesting idea, though I feel like it'd make things a little too complicated. One thing I like about the way me and blockerdave are envisioning the idea of bowling plans is that it is really just building off what is already in the game. In fact, I am half-anticipating it to be in the game in some form. There is already a system where batsmen have set strengths and weaknesses - it makes sense for the AI to be set up so that it sets the field according to these characteristics. There are also categories of fields already (attacking, defensive and i think moderate?) - the next logical step is that these fields correspond to suitable bowling tactics.

One point I forgot to mention is that this approach also offers a way to make the skills of bowlers more meaningful. For example, the 'fullish, outside off' bowling plan has a *perfect* ball - its line is a couple inches outside off, its length is the intersection of the good and full areas. The higher the skill level of the bowler, the smaller the potential landing spot radius around that perfect ball. So, if they have full skill level for outswingers, then when they bowl an outswinger they are more likely to land it closer to that perfect ball spot.

Also, as I said, the plans would have a feature built into them whereby sometimes the bowler will bowl a 'loose delivery'. For example, instead of bowling 'fullish, outside off', they might bowl a short wide one, or a full one that is heading down leg. Naturally, the more skillful the bowler is, the less often they will bowl 'loose deliveries' (i.e. deliveries that don't follow the plan, and are therefore 'bad balls' for the particular field that is set for the plan). This would be a nice way to mimic the reality of skill differences between bowlers, as this is essentially what it comes down to. Really skillful Test bowlers rarely bowl 'bad balls', less skillful ones bowl them a lot, and this lets the pressure off the batsman as they know that whatever the plan, they will be getting plenty of scoring opportunities.

Another point that I may not have made clear in my original post in this thread (I made a more in depth description of the idea elsewhere on the forum, which included more thoughts on what the different tactics and fields would be like)... is that there would essentially be three tiers of decision making for the AI:

1. It decides whether it should be attacking, defensive or moderate. This would be based on the team score, the individual batsman's score, the captain's temperament, the match situation, etc.

2. It decides which bowling plan *within* the category of attack/defense/moderate it should employ. This then determines which subset of fields in the category it chooses. There should be maybe 3 different but similar fields for each plan that are randomly chosen - this mimics the captain's 'gut feeling', e.g. whether they should have a 3rd slip or a cover.

3. It decides how long it should stick with the chosen plan. Again, the initial plan it chooses should be based on the batsman's set strengths and weaknesses. So, if a batsman walks in and his strength is offside and his weakness is legside, then the AI would *always* choose the 'bowl at the stumps' plan. However, if the batsman doesn't get out, then after some period of time (which is determined by the captain's temperament) the AI would change to a randomly determined new plan, within the category of attack/defense/moderate already determined by the factors in the first tier thinking.
 
Last edited:
@Cricketonline to get back to your post about field placing, I wasn't wanting to place fielders anywhere I like but more anywhere within a certain distance from the set postion. It would be incremental 1m circle in close up to 10m circle on boundary. Would that be so hard to include?
 
They haven't removed the aggressive/brute/conservative/opportunistic attributes, have they? Surely they would be sufficient to play the sort of role we are talking about with regard to strategies, right?

those are still there but there's no development. I was hoping for (at a minimum):
  • tactical approach (to define captaincy style) - the existing bowler mentality could be co-opted for this as i mention in the op
  • temperament (to define ability to deal - batting or bowling - with crunch situations such as delivering/facing a hat-trick ball; the death stages of limited overs; a run-chase; a batting collapse)
  • tactical skill (to define how well a bowler can bowl to a plan or according to his strengths/batsman's weakness; how often a captain changes his field/follows the ball; how a batsman can pace an innings/cope in different situations)
these things not only add realism but that key word "variety". DBC14 doesn't offer a lot of variety in opponent, and i hoped DBC17 would address that. maybe they have another approach - i hope so.
 
@Cricketonline to get back to your post about field placing, I wasn't wanting to place fielders anywhere I like but more anywhere within a certain distance from the set postion. It would be incremental 1m circle in close up to 10m circle on boundary. Would that be so hard to include?

Just from development/testing perspective it would definitely ramp up the effort as the Dev and QA will have to test the AI for those additional field spots. If they don't the risk is the AI may behave inconsistently at one or more of these field positions. As an example, you bring in a fielder in by 1m from the standard Cover position. The AI has only been trained to recognize 'Cover', 'Cover Point', 'Deep Cover' etc. field positions. In that case AI may place the ball on the off side, interpret the fielder being at Cover and try to sneak a quick run and get run out. Similar scenarios may result in cheap AI wkts due to run outs or balls being hit in the air by AI as they wouldn't have an idea the fielder is present in the non-standard field position.

I'm not saying this can't be done but this definitely will increase the amount of effort Big Ant will need to put in. This feature would be great to have as it will add longevity and new tactics in the game but not sure it can happen in this iteration where the focus has to be to nail standard field settings and AI. Once that's done Big Ant can explore additional challenges such as the one you describe. But that's my take on it based on what I've seen in various cricket video games over the years and Big Ant folks would be in a much better position to speak on it.

Edit: I'm waiting for machine learning to make its way in to DBC as that would lead to an 'always fresh' AI and it will easily help address the sort of feature you asked for. The AI will never go stale and and add a different dimension and longevity to the game. Not sure how close we are to have machine learning in the game but to me that's the way to go to have a decent AI in a video game.
 
Last edited:
Just from development/testing perspective it would definitely ramp up the effort as the Dev and QA will have to test the AI for those additional field spots. If they don't the risk is the AI may behave inconsistently at one or more of these field positions. As an example, you bring in a fielder in by 1m from the standard Cover position. The AI has only been trained to recognize 'Cover', 'Cover Point', 'Deep Cover' etc. field positions. In that case AI may place the ball on the off side, interpret the fielder being at Cover and try to sneak a quick run and get run out. Similar scenarios may result in cheap AI wkts due to run outs or balls being hit in the air by AI as they wouldn't have an idea the fielder is present in the non-standard field position.

I'm not saying this can't be done but this definitely will increase the amount of effort Big Ant will need to put in. This feature would be great to have as it will add longevity and new tactics in the game but not sure it can happen in this iteration where the focus has to be to nail standard field settings and AI. Once that's done Big Ant can explore additional challenges such as the one you describe. But that's my take on it based on what I've seen in various cricket video games over the years and Big Ant folks would be in a much better position to speak on it.

Edit: I'm waiting for machine learning to make its way in to DBC as that would lead to an 'always fresh' AI and it will easily help address the sort of feature you asked for. The AI will never go stale and and add a different dimension and longevity to the game. Not sure how close we are to have machine learning in the game but to me that's the way to go to have a decent AI in a video game.

perhaps a compromise might be to add just a couple of positions with variation - e.g. mid-off/on can be wider or straighter, gully can be squarer or finer; slips can be nearer or deeper.

that gives you a bit of extra tactical flexibility but doesn't make it the wild west!
 
perhaps a compromise might be to add just a couple of positions with variation - e.g. mid-off/on can be wider or straighter, gully can be squarer or finer; slips can be nearer or deeper.

that gives you a bit of extra tactical flexibility but doesn't make it the wild west!

I'm all for as much variety we get as long as the AI is competitive. At the end of the day the game will live or die on the basis of the quality of AI. If the AI isn't good, competitive or challenging enough no amount of field sets would save it.
 
I'm all for as much variety we get as long as the AI is competitive. At the end of the day the game will live or die on the basis of the quality of AI. If the AI isn't good, competitive or challenging enough no amount of field sets would save it.

Which is why I was pretty underwhelmed when I saw the lack of mental skills in the academy.

Variation in opponent is a part of a good challenging AI. It should be different having your opponents 20/3 and in walks Ian Bell or Steve Waugh.
 
Which is why I was pretty underwhelmed when I saw the lack of mental skills in the academy.

Variation in opponent is a part of a good challenging AI. It should be different having your opponents 20/3 and in walks Ian Bell or Steve Waugh.

Having a varying opponent will change the career mode as well. Currently every team and every match seems same. It doesn't matter whether you are playing Australia or India or New South Wales or Essex, it's all the same. You get same field sets, same approach to bowling or batting so on so forth. If we have different types of team behavior (which is a different topic from field sets), it will add much needed variety to the game and the career mode.
 
Having a varying opponent will change the career mode as well. Currently every team and every match seems same. It doesn't matter whether you are playing Australia or India or New South Wales or Essex, it's all the same. You get same field sets, same approach to bowling or batting so on so forth. If we have different types of team behavior (which is a different topic from field sets), it will add much needed variety to the game and the career mode.

I don't believe it's a different topic to field sets they're related, albeit behaviour encompasses more.

Changing the team captain, a bowling change, a wicket, a big partnership... all should present you with a different challenge. I really hope it's not too late for @BigAntStudios to look at mental skills/attributes, how they effect the outcome etc.

I have A LOT of notes I've made about this and I'd be more than happy to share them here or in private
 
I don't believe it's a different topic to field sets they're related, albeit behaviour encompasses more.

Changing the team captain, a bowling change, a wicket, a big partnership... all should present you with a different challenge. I really hope it's not too late for @BigAntStudios to look at mental skills/attributes, how they effect the outcome etc.

I have A LOT of notes I've made about this and I'd be more than happy to share them here or in private

I would say share them with Matt and see if that approach is viable based on what's being implemented in the game.
 
Matt loves private messages. Fact.
 
A few comments:

We're being careful with the skills we are including in the game, as the primary goal is to make the skills, mentality and abilities of a player meaningful, and there be clarity about the kind of impacts that differences will have on a player. It's an iterative process - lessons learnt from the first game, combined with being realistic about the steps to be taken in enhancing that. Would certainly not want to be too ambitious and end up with something that takes steps backwards on gameplay.

Obviously not the whole lot - but there's a reasonable slice of the first post that contains things that will be possible. Fields will be able to be tagged with things like the type of bowler, the match type and the match situation that they are intended to be used for. That will then feed into how the AI picks which fields to use. There's a few ideas here that could be easy enough modifications over what we've got already - but I loathe to commit to anything until it's in the game.

And finally - don't PM me. The discussion of the idea is far more valuable than the actual idea - things are refined and added to when people talk about them. That's why posts on the forum are good, and I will read them and follow the discussion, while I'll give your PM a quick glance, probably not reply and forget you ever sent it when it's hidden by newer ones.
 
Don't worry I was never gonna pm you...

It's reassuring you were going down this track anyway, and if there's anything in the OP or any of the discussion here that makes it into the game and takes us forward I will be delighted.

For me the goal is

1) feel I'm facing different plans/fields due to the bowler's strengths / batsman's weakness
2) feel that different opponents react differently - going defensive quicker/slower etc.

Anything that moves us towards that is great
 
Thanks for all the thought and effort you guys have put into these sort of posts.

when we get a better game at the end of the year it'll be partially on discussions like this that have taken place over the past few years and that's a great thought to have
 

Users who are viewing this thread

Top