My thoughts for the future iteration for BA cricketing series

[wasteyouryouth]

So I've been bowling in a test match in Ashes and thinking a bit about the AI and how we can get their shot selections, scoring rates and also dismissals more realistic. Currently I'm not a fan of the "hidden scoring target", as I feel it takes things out of the control of the player, but we're told if this is removed then the AI will not play attacking strokes go nowhere. What this suggests to me that there is something fundamentally wrong with the way the AI is coded to choose which balls to attack/defend/leave etc.

I've heard a beta tester (I think @WealeyH?) say that one of the reasons the AI is deficient and needs to cheat is because it's too easy for the player to bowl perfect/perfect deliveries every time. While this is true and it would be great if we could figure out a way to make bowling harder, again I think there should be more to it than this. For example, currently if I decide I'm going to bowl a half-volley on leg-stump and hit perfect/perfect on the timings then does that make it a good delivery? Any good batsman will put that way 9 times out of 10, intentional or not, but currently in Ashes it entirely depends on the hidden run-rate target whether they will hit it to the boundary or tap it meekly to the square-leg fielder. So my proposed system:

Every AI batsman has an "Adaptive Heat-map"

Pictured above is VERY ROUGH idea of what I'm suggesting. The player bowls the ball and depending where it ends up on reaching the batsman this will affect what action he decides to take. For example in this image, which I would consider for a test match batsman quite early in his innings, green would be duck/leave, blue would be block and red would be playing a run-scoring stroke (I'm a horrible batsman so don't take this map as gospel ). Then he will play the appropriate shot based on where the ball is (I suspect there is already something similar to this in Ashes but given the issues we have with the AI it doesn't seem anywhere near comprehensive enough).

The main point of this system is that it will change depending on various conditions such as match type, match situation, required-rate, player confidence, player preference (eg, leg/off side favoured players), the field setting etc etc etc. In the final over of a T20 the entire heat-map would be red which would mean we would NEVER have to see a player blocking the ball at the death again.

(One caveat is full-tosses/double bouncers which would obviously need to be accounted for as a full toss bowled by a spinner on off-stump probably shouldn't be blocked )

Currently there is also a problem with the AI where they always know where the ball is going to end up (wacky fun-house seam patch proved this to be the case ) so how are we going to keep the AI from being too OP with this new system where they always make the correct choice for the situation?


AI batsman "Predict" the path/timing of the ball

Earlier I said that the batsman will play based on where the ball ends up but that's not actually the case. What I propose instead is that when the player finishes the bowling inputs (at the point of release in the animation) there are actually two paths generated in the game. One that the ball will take and the other which is the one that the batsman will make his decision and play his shot for. Here's an example using he heat-map:

Let's say it's the first morning of an English test match and you're bowling as Jimmy Anderson. You choose an out-swing delivery and nail your line+length. At the other end is an Indian opener who is not the best at playing swing (). Therefore, based on his attributes and the line+length that the player input, the game generates point A as his prediction for where the ball will arrive at the crease, it will pick a forward defensive shot for him to play. However, simultaneously the game also takes into account the new ball/overhead conditions/attributes of our bowler (Jimmy) and generates the actual path of the ball, which will swing off the predicted line and reach the batsman at point B. As the batsman has now prodded his forward defense down the wrong line, the ball takes the edge and flies beautifully to slip . The best way to think of it is to imagine you're bowling two balls at once, one an out-swinger and one an invisible straight ball. It is the straight ball that the batsman is playing for.

Obviously on a pitch with no movement and bowling to a high-skilled and confident batsman, the points A and B would be basically on top of each other and you'll be tearing your hair out to get a wicket.

One thing I like about this system is it solves a lot of the problems we've had in the previous games. On top of the batsman seemingly omniscient abilities to predict spin/seam/swing it also creates variance in batsman ability: Lower rated batsmen will obviously have a prediction point much further away from where the ball actually arrives and we can have batsmen who are good at playing spin but bad against swing purely based on how much the prediction point differs from the actual ball. I've thought a bit about a system of player traits and here's an image showing a player who I've decided has the trait "can't pick the googly"

This would be to a ball by a leg-spinner that is pitched around off-stump. Our hapless batsman thinks it's going to be a leggy arriving at point A and so plays a push towards the covers but "oh no!" it's actually a googly and the ball spins through the gate to clatter into leg stump.

Another thing this system solves is the effect pitch degradation has on the batsmen. As the ball starts to move unpredictably just have some random balls where A + B do not match up regardless of what is bowled. Would be great to see some horror balls where it keeps low or spins way more than usual.

Ok so one thing I haven't addressed is the timing. Obviously batsmen can be fooled by changes of pace (or just beaten straight up by fast deliveries) and not only movement. It's harder to depict visually but I think it's a similar idea. Batsman predicts the ball will arrive at point A in 0.5 seconds, but it actually comes at 0.60 and he misses it because he's too early. In stuff like T20 where you're constantly changing your pace to try and fool the batsman this is could get pretty complicated but it would ultimately be down to the batsman's timing attribute how successful they are against different paced deliveries.

Summary of what I feel at the main pros of this system compared to the current one:

- AI no longer has just a number they are aiming for and instead plays based on their "adaptive heat-map" which will change depending on many variables, leading to more natural shot selection and scoring in all formats
- Allows much more noticeable variation between AI batsmen based on attributes, ability and mentality
- Lets the player feel like they can fool the batsman with movement/changes of pace and get realistic dismissals
- Conditions have visible effect on how the AI batsmen play


Alright, that was just me throwing out my ideas again. I'm no coder so I have no clue how feasible this system would be to implement into a game and there's obviously a million variables you can account for in a game of cricket. Just a bare bones proposal and I expect I've overlooked some stuff so I'd be interested to hear the thoughts of anyone who bothered to read all this.

Cheers!

Warning: I've no idea if any of this makes any sense. I'm just running with your idea and mixing it with thoughts I've had in the past.

I definitely think the quality of the delivery (Perfect/Perfect) should not be as big a determining factor for whether or not the AI batsman plays a delivery. Ideally, Perfect/Perfect delivery would mean the length and movement is spot on. Drop below and there are reductions in the effectiveness of the ball in terms of accuracy and intended movement. I don't think speed should be reduced unless the timing is way off.

When it comes to AI batting whatever system I think of it in terms of whittling down the shots available to the batsman until you are left with a single shot they would play to each possible delivery. So you go from all shots being available then factors such as batsman skill, batsman mentality, match situation, conditions, opposition get you down to the shot that is played. So it's not always the case that the shot selected will be the batsman's best shot but the shot that is most suited to the situation - so conservative batsman that need to attack would be forced out of their comfort zones in T20 and the same for attacking batsman needing to defend in test matches.

I like the idea of thinking of it as heat maps. The Shot Zones image is just a quick example I threw together. Each circle would correspond to a set of shots (Normal, Attacking, Defensive, Leave - maybe grounded aggressive and special shots too) that would be selected prior to the delivery so that when the ball pitches in the zone that shot is played. In theory it'd be a normal cover drive, a lofted cover drive, a defensive push directed towards cover and leaving the ball. Green they are good in that zone, amber they are Ok and red means it's not a strength. That would be determined through however skills would be set up. Maybe it'd be for individual shots or something more simple like ability at front foot, back foot, leg sides, off side, straight etc.

Spoiler: Shot Zones

Let's say a Zone has five possible varieties of a shot - they begin with an equal likelihood (20%) of being played. Then you introduce the other factors which adjust the percentage. Mentality (attacking) would push an attacking cover drive to 40%, leave and defensive to 10%. Match situation - if it's say the last over of a T20 - pushes the attacking cover drive up to 60%, leave and defensive down to 0%. That's a basic start, other things would need to be thrown to refine the calculation.

For me the key is refining the skills that are involved with how the shots are selected. In the image below I've thrown in a few example skills that might be useful to explain my thoughts about the process. So, it isn't that the AI will know where they ball will go but will have a pre-determined response for each scenario (which zone the ball pitches in).

For all I know this may be similar to how it actually works now and it just needs refining - which is where those new skills could come on useful.

I think you've got to take into consideration batsman skill, batsman mentality, batsman preferences, conditions, match situation, opposition tactics (the bowler and field settings) and the batting team tactics.

It's trying to find a logical path for each batsman to play a certain shot.

Spoiler: All shots to chosen shot

 

[Gamer Pradosh]

I am really really happy with all these contributions by all of you guys..It's not just a thought put in but the effort taken to portray those thoughts with such images and explanation is just brilliant..Thanks for making the thread meaningful so far guys..

Also @wasteyouryouth that's a really good thought as well, I like the use of shots becoming available to the player but the shot he plays depends on skills and various other factors, which is what is required..Use of different shot to same zone would make bowling lot more interesting and strength and weakness would actually make a lot of sense and have a factor in the game.. Situational shots will make AI play depending on format and situation of the match too..

Another great post..

 

[Flat_Track_Bully]

I want more variation in animation. Different animations to the same shot depending on timing. More fluid motion. At the moment its far too cumbersome. Same goes for the bowlers. Just bringing this area up in the game will impact immersion massively.

 

[wasteyouryouth]

Use of different shot to same zone would make bowling lot more interesting and strength and weakness would actually make a lot of sense and have a factor in the game.. Situational shots will make AI play depending on format and situation of the match too..

I definitely think some kind of grey area for certain zones would be good. Either different shots or different footwork. A batsman early in their innings, or with a low judgement or footwork skill, might appear indecisive; playing back foot shots when front foot shots would be better. Maybe the process would be: 'Front Foot or Back Foot?' -> 'On Drive/Straight Drive/Leave/Defend?' -> 'Lofted/Aggressive/Normal/Precise?'

A big thing is trying to flesh out suitable skills that really influence how AI approaches batting. When we control a player these skills would also contribute to making batting easier or more difficult for us.

 

[sonimrajk]

There should be different animations for precision shots and normal shots.

 

[Nathaus]

I'm going to play Devil's advocate a bit, and say that the thing I hope for most in the next BA cricket game (assuming there is one) is for:
1) the development/creative team to have a vision of the cricket game they want to make and stop listening to all the micro-feedback about timing windows and what not and have some conviction in what they've created. Get rid of the physics sliders and actually build proper physics in to the game.
2) Get rid of the majority of sliders and actually hard code the physics in to the game engine. Unless people want to play cricket on the moon, the physics will be constant wherever you play (if it's variable in real life, make it variable in the game i.e. gravity can change by 0.01m/s, air pressure can change at altitude, etc).
3) if CA want 6 hitting to be a piece of pie because it's more 'entertaining' then they don't understand the appeal of cricket to the enthusiasts. If that's the case, have arcade physics and realistic physics, or have batting assists show you the line of the ball, control timing, etc so every shot you hit can go for 6 if that's what you call 'entertainment' just like in racing sim games with driving assists.

I appreciate the irony of me saying ignore micro-feedback and then proceeding to include my own micro-feedback

 

[Nikhil Ajmera]

I really Like all the major changes that you guys have suggested.
I would like some minor changes as well. these are not even changes but are some additions that would improve the game and not make it monotonous.
1. New celebration system - Something like PES or FIFA where you could press certain keys for certain kinds of celebrations after a wicket, after scoring a fifty/hundred/, After winning a match (Like the players running to the pitch and other celebrations).
2. Bowler raising the bowl after getting a Fifer in an innings.
3. A Short form of presentation where Man of the Match and Man of the Series is Awarded.
4. Tour mode should have a success rate. Winning certain no. of matches leads to certain % of tour success. So that playing a tour means something. Winning a certain tour unlocks certain celebration or something else for your career mode.
5. Better Commentary. Please this is a must.
6. Better Fielding annimations.
7. I actaully like the gameplay a lot. So I'd go for better AI than a change in Gameplay.

 

[jaffar100]

I really Like all the major changes that you guys have suggested.
I would like some minor changes as well. these are not even changes but are some additions that would improve the game and not make it monotonous.
1. New celebration system - Something like PES or FIFA where you could press certain keys for certain kinds of celebrations after a wicket, after scoring a fifty/hundred/, After winning a match (Like the players running to the pitch and other celebrations).
2. Bowler raising the bowl after getting a Fifer in an innings.
3. A Short form of presentation where Man of the Match and Man of the Series is Awarded.
4. Tour mode should have a success rate. Winning certain no. of matches leads to certain % of tour success. So that playing a tour means something. Winning a certain tour unlocks certain celebration or something else for your career mode.
5. Better Commentary. Please this is a must.
6. Better Fielding annimations.
7. I actaully like the gameplay a lot. So I'd go for better AI than a change in Gameplay.

One point to add...

Include cup presentation after bilateral series as well like it was in DBC14...

 

[wasteyouryouth]

I'm going to play Devil's advocate a bit, and say that the thing I hope for most in the next BA cricket game (assuming there is one) is for:
1) the development/creative team to have a vision of the cricket game they want to make and stop listening to all the micro-feedback about timing windows and what not and have some conviction in what they've created. Get rid of the physics sliders and actually build proper physics in to the game.
2) Get rid of the majority of sliders and actually hard code the physics in to the game engine. Unless people want to play cricket on the moon, the physics will be constant wherever you play (if it's variable in real life, make it variable in the game i.e. gravity can change by 0.01m/s, air pressure can change at altitude, etc).
3) if CA want 6 hitting to be a piece of pie because it's more 'entertaining' then they don't understand the appeal of cricket to the enthusiasts. If that's the case, have arcade physics and realistic physics, or have batting assists show you the line of the ball, control timing, etc so every shot you hit can go for 6 if that's what you call 'entertainment' just like in racing sim games with driving assists.

I appreciate the irony of me saying ignore micro-feedback and then proceeding to include my own micro-feedback

I completely agree.

 

[Spoobir]

I'm going to play Devil's advocate a bit, and say that the thing I hope for most in the next BA cricket game (assuming there is one) is for:
1) the development/creative team to have a vision of the cricket game they want to make and stop listening to all the micro-feedback about timing windows and what not and have some conviction in what they've created. Get rid of the physics sliders and actually build proper physics in to the game.
2) Get rid of the majority of sliders and actually hard code the physics in to the game engine. Unless people want to play cricket on the moon, the physics will be constant wherever you play (if it's variable in real life, make it variable in the game i.e. gravity can change by 0.01m/s, air pressure can change at altitude, etc).
3) if CA want 6 hitting to be a piece of pie because it's more 'entertaining' then they don't understand the appeal of cricket to the enthusiasts. If that's the case, have arcade physics and realistic physics, or have batting assists show you the line of the ball, control timing, etc so every shot you hit can go for 6 if that's what you call 'entertainment' just like in racing sim games with driving assists.

I appreciate the irony of me saying ignore micro-feedback and then proceeding to include my own micro-feedback

I agree with this apart from one thing and that's ball speed. There's no way I can go back to how slow the default ball speed is but what I play with is probably too fast for a lot of people.

 

[wasteyouryouth]

I agree with this apart from one thing and that's ball speed. There's no way I can go back to how slow the default ball speed is but what I play with is probably too fast for a lot of people.

I don't know if it is now but bowling speed should be linked to difficulty with realistic speeds being the hardest difficulty. If that's too easy for people then other areas should be addressed.

 

[Spoobir]

I don't know if it is now but bowling speed should be linked to difficulty with realistic speeds being the hardest difficulty. If that's too easy for people then other areas should be addressed.

That was how it was on DBC14 and I was too rubbish to bat on Legend difficulty which had the highest speed.

 

[blockerdave]

So I've been bowling in a test match in Ashes and thinking a bit about the AI and how we can get their shot selections, scoring rates and also dismissals more realistic. Currently I'm not a fan of the "hidden scoring target", as I feel it takes things out of the control of the player, but we're told if this is removed then the AI will not play attacking strokes go nowhere. What this suggests to me that there is something fundamentally wrong with the way the AI is coded to choose which balls to attack/defend/leave etc.

I've heard a beta tester (I think @WealeyH?) say that one of the reasons the AI is deficient and needs to cheat is because it's too easy for the player to bowl perfect/perfect deliveries every time. While this is true and it would be great if we could figure out a way to make bowling harder, again I think there should be more to it than this. For example, currently if I decide I'm going to bowl a half-volley on leg-stump and hit perfect/perfect on the timings then does that make it a good delivery? Any good batsman will put that way 9 times out of 10, intentional or not, but currently in Ashes it entirely depends on the hidden run-rate target whether they will hit it to the boundary or tap it meekly to the square-leg fielder. So my proposed system:

Every AI batsman has an "Adaptive Heat-map"

Pictured above is VERY ROUGH idea of what I'm suggesting. The player bowls the ball and depending where it ends up on reaching the batsman this will affect what action he decides to take. For example in this image, which I would consider for a test match batsman quite early in his innings, green would be duck/leave, blue would be block and red would be playing a run-scoring stroke (I'm a horrible batsman so don't take this map as gospel ). Then he will play the appropriate shot based on where the ball is (I suspect there is already something similar to this in Ashes but given the issues we have with the AI it doesn't seem anywhere near comprehensive enough).

The main point of this system is that it will change depending on various conditions such as match type, match situation, required-rate, player confidence, player preference (eg, leg/off side favoured players), the field setting etc etc etc. In the final over of a T20 the entire heat-map would be red which would mean we would NEVER have to see a player blocking the ball at the death again.

(One caveat is full-tosses/double bouncers which would obviously need to be accounted for as a full toss bowled by a spinner on off-stump probably shouldn't be blocked )

Currently there is also a problem with the AI where they always know where the ball is going to end up (wacky fun-house seam patch proved this to be the case ) so how are we going to keep the AI from being too OP with this new system where they always make the correct choice for the situation?


AI batsman "Predict" the path/timing of the ball

Earlier I said that the batsman will play based on where the ball ends up but that's not actually the case. What I propose instead is that when the player finishes the bowling inputs (at the point of release in the animation) there are actually two paths generated in the game. One that the ball will take and the other which is the one that the batsman will make his decision and play his shot for. Here's an example using he heat-map:

Let's say it's the first morning of an English test match and you're bowling as Jimmy Anderson. You choose an out-swing delivery and nail your line+length. At the other end is an Indian opener who is not the best at playing swing (). Therefore, based on his attributes and the line+length that the player input, the game generates point A as his prediction for where the ball will arrive at the crease, it will pick a forward defensive shot for him to play. However, simultaneously the game also takes into account the new ball/overhead conditions/attributes of our bowler (Jimmy) and generates the actual path of the ball, which will swing off the predicted line and reach the batsman at point B. As the batsman has now prodded his forward defense down the wrong line, the ball takes the edge and flies beautifully to slip . The best way to think of it is to imagine you're bowling two balls at once, one an out-swinger and one an invisible straight ball. It is the straight ball that the batsman is playing for.

Obviously on a pitch with no movement and bowling to a high-skilled and confident batsman, the points A and B would be basically on top of each other and you'll be tearing your hair out to get a wicket.

One thing I like about this system is it solves a lot of the problems we've had in the previous games. On top of the batsman seemingly omniscient abilities to predict spin/seam/swing it also creates variance in batsman ability: Lower rated batsmen will obviously have a prediction point much further away from where the ball actually arrives and we can have batsmen who are good at playing spin but bad against swing purely based on how much the prediction point differs from the actual ball. I've thought a bit about a system of player traits and here's an image showing a player who I've decided has the trait "can't pick the googly"

This would be to a ball by a leg-spinner that is pitched around off-stump. Our hapless batsman thinks it's going to be a leggy arriving at point A and so plays a push towards the covers but "oh no!" it's actually a googly and the ball spins through the gate to clatter into leg stump.

Another thing this system solves is the effect pitch degradation has on the batsmen. As the ball starts to move unpredictably just have some random balls where A + B do not match up regardless of what is bowled. Would be great to see some horror balls where it keeps low or spins way more than usual.

Ok so one thing I haven't addressed is the timing. Obviously batsmen can be fooled by changes of pace (or just beaten straight up by fast deliveries) and not only movement. It's harder to depict visually but I think it's a similar idea. Batsman predicts the ball will arrive at point A in 0.5 seconds, but it actually comes at 0.60 and he misses it because he's too early. In stuff like T20 where you're constantly changing your pace to try and fool the batsman this is could get pretty complicated but it would ultimately be down to the batsman's timing attribute how successful they are against different paced deliveries.

Summary of what I feel at the main pros of this system compared to the current one:

- AI no longer has just a number they are aiming for and instead plays based on their "adaptive heat-map" which will change depending on many variables, leading to more natural shot selection and scoring in all formats
- Allows much more noticeable variation between AI batsmen based on attributes, ability and mentality
- Lets the player feel like they can fool the batsman with movement/changes of pace and get realistic dismissals
- Conditions have visible effect on how the AI batsmen play


Alright, that was just me throwing out my ideas again. I'm no coder so I have no clue how feasible this system would be to implement into a game and there's obviously a million variables you can account for in a game of cricket. Just a bare bones proposal and I expect I've overlooked some stuff so I'd be interested to hear the thoughts of anyone who bothered to read all this.

Cheers!

some great stuff here. the main complication i can think of here is the compute power needed to generate so much of this on the fly, but the principle is spot on.

 

[Stil]

Some really great, detailed points in here. Love the illustrations from @wasteyouryouth.

On a broad level, I think a partnership between Big Ant and a larger studio with more money and resources would do a lot to improve the next game in a substantial way. The likes of EA and Sony might not think it profitable to invest in a whole dev team for a cricket game, but if they could be convinced to lend some money and development resources to Big Ant in exchange for a profit share, it might make business sense for them.

I mention Sony because they make a Major League Baseball game which is, in my opinion, perhaps the most authentic and accurately tuned video game representation of a real sport. The fact that the developers of this game focus on ball throwing and bat-on-ball physics, as well as AI “swing vs. don’t swing” logic, would make them a great potential development partner for a cricket game.

TL/DR: I think Big Ant finding a big development partner for their next cricket iteration might really help give us the nearly-perfect game we’re looking for.

 

[flake]

Brilliant stuff from some of you guys on this thread. Hope they take some of this on board !

This has mainly to do with non physics issues as you guys are covering that side pretty well.

I agree with what was said regarding the foreign tours FEELING like you are there in terms of that little map animation (pretty cool), sounds of the crowd, pitches looking obviously different, drier looking outfields in certain places etc

I really feel a few solid cutscenes like a man of the match, series, presentation would be a solid addition, even if you get the same presenter appearing on the mic doing the same voice over. I feel this wouldn`t be too hard to implement if you keep it pretty basic. Adds to that vibe.
Possibly get a few guys to do voice overs for the players receiving awards, one or two from each country for the different accents. This could be assigned when editing a player. Similar to what they did for DBC17 where the community helped with that. Makes us feel the players actually have personalities and adds to that realism aspect.
Possibly a cutscene of a team getting off a bus prior to the game? As seen on tv during World Cups.
Maybe after a batsman is getting screwed over ball after ball, his partner could be seen to have a quick word between balls. Same with a struggling bowler, where the captain goes upto him in between balls. All these little things make the product feel that bit more realistic.

A brief note on bowling and batting animations, it would be good to keep adding more as iterations continue to come around, as you can never have enough animation styles, especially for bowling.
Also, as mentioned plenty of times, the fielding animations could do with a bit of an update, especially the throwing.

Here`s hoping Big Ant deliver another cracking game around World Cup time next year ! I wont be watching much of that since the ICC screwed Zim out of a place but nonetheless ! Odds on getting the ICC license might be slim but we can hope !