[Affiliate Program]
[For Sale]

Sign up to PayPal and earn $5! Home CD-ROM
International Test Cricket


Feedback received

Send feedback about any of the articles.

Paul Clarke said on 5 June 2000 in response to his own article A Potted History :-)

Its funny, now I've committed this to paper, I've remembered a whole raft more (well a few anyway) A very good game for the Amiga (it had a picture of either Peter Such or Graeme Hick ducking a bouncer on the front) Fully changeable fields, editable teams, line and legnth etc I think it was just called'cricket'. Then of course Ian Bothams (written by an Italian company I think..the mind boggles)

Reading other articles, the subject of the sucker ball is mentioned. (along with the percentage shot) To be fair this is a problem that plagued footy sims until about two years ago. The problem is developing a realistically fallable AI. Your corresponants notes about iterative processing bears up statistically though this only works for a multiple node process (thought process!?)

Once this problem is corrected, we may see other games jumping on and using it, it seems though that with some software, realistic stats don't even come into it.

Get Cricket 2000 to pay an innings for you on auto mode and youll see for yourself. Lofted straight drives and sweeps for 4 and 6, and a grand 20 over total of about 50-60.

I'm glad the "bored of batting" syndrome is also well recognised...perhaps the answer is, to choose one, two or three players maximum to 'be' during a match (star batman / star bowler and star allrounder maybe or whatever) It limits it somewhat, but cricket lacks the fluidity of football,(where, hopefully the players gell into one 11 man unit) so this 'boedom' problem will allways raise its head.

Anyway, thanks again.

Paul Robson said on 4 September 1999

As the author of a shareware game, I'd like to add a few comments about your "Arcade Games" section.

Bowling AI You are entirely correct about this. On my game I did this reasonably effectively by (as you suggest) incorporating the player ratings, so feet move slower, big shots are less likely to succeed (but blocking is not so affected) etc.

This does have a gameplay problem. I think when the gameplayers play the game, it is quite difficult to adjust between the two batsmen, especially in such a long game as cricket. If you have a dasher and a blocker at either end, you have to be careful as to what shots you play; this can impede the flow of the game. Playing my game, I found it irritating to bat with the tail (perhaps that's realistic ?)

Batting AI is another matter altogether. Most of these implementations are just lazy coding; a very simple cause & effect scenario - you on drive a bouncer and are caught all the time sort of model. This has two bad effects.

Firstly, you can find a way of always getting a batsman out. (On a demo of the original Gooch game I managed to get 12 wickets in the two over demo); most games are not this bad (some are) but as one can see from your hints 'n' tips page, these loopholes are there.

Secondly, you can always score runs. After a while it becomes easy to detect percentage shots and you can more or less score at will. As with any "cheat" if you know it is there it is extremely difficult not to use it, at least some of the time.

The solution I tried worked quite well. Firstly it is important to realise that a real "physics model" is actually almost impossible. There are too many interacting variables. So one is left with the cause and effect model.

One of the elements of the model is limited the shots. You are limited to 8-16 shots. Everything else is variable; the position of the player, the ball position and direction.

For each of these shots a realistic and semi-random model has to be worked out and play tested to death. There is no way around this requirement. But one must still avoid absolute fixed results - an attempt to sweep a fast yorker or leg glance an offside wide will work.... just not very often)

The trick which I don't think anyone else used, was to run the data (stroke, positions etc) through the model several times, (the number depending on the batsmans skill) and take the "best" result. Thus if a Lara did try and sweep a yorker, he would have a greater chance of success - it would be like him being allowed to have several tries at it, and identical balls and strokes would have different results. The Laras would have 9 tries, the Tufnells 1 or 2.

It avoided the predictability problem pretty well. After about two years I noticed that if you played on a square turned and bowled left arm wristspin over the wicket to a right hand bat, pitched it on the edge of the crease and turned it back to hit the stumps, you could bowl a side out for < 30. But I didn't find another.

The hard (and I actually think it can't be done) bit is the realism of bowling at a computer player. When bowling one tries to develop plan or trap for a particular batsman over a sequence of balls, and that can't be simulated in a single delivery. So, as with every cricket game I've played, you end up bowling a mixture of deliveries and waiting for a wicket to fall.

It isn't possible to lay traps in any simulation (of any game I know of)

Arijit Sen said on 31 Ausgust 1999

The article of Troy (What I think would make a great cricket game) was great. The all features suggested by him was excellent. I think heavyweight companies like EA Sports or Codemasters should come forward and make this kind of games. They could add this extra features as well

bulletThird Umpire Facility
bulletCrowds disappointed against not given decisions
bulletCrowds furious against and there team play (No bottles, of course)
bulletSome crazy animals entering the field during play
bulletMan of the match videos with real player voices
bulletRealistic player images ( Sachin looks like Arcade version of Sachin)
bulletPlayer names individually pronounced
bulletMore details on batsman's shot, bowlers bowling, fielding, captains tactics, how the game is going.
bulletAction Replay Facility at anytime and automatic
bulletVideos after winning the match, losing the match, scoring centuries, scoring fifties, taking 5+ wickets.
bulletDatabase manager. It will record the database of the players since the game was installed
bulletAnd more I could not remember just now.

This development are suggested for arcade type games only. ( Cricket World Cup 99 is an example)

Hugh Bradley said on 1 July 1999

General Cricket Games Discussion

I thought I would prepare a brief response to the general comments on cricket games. I would like to start by stating that in the main I am agree with your requirements for good cricket games, both simulations and arcade alike.

I would suggest that your division of games should note that simulations are often developed for a further purpose. I my view, a simulation is a fascinating exercise, but isn't a game in the sense of competition. (I do understand the attraction of simulators: Back in about 1986 I wrote a test cricket simulator on my BBC Micro, which generated ball by results for a given match. As I see it, this provides a substitute to the experience of following a match, what I would call the "soap opera appeal". Furthermore, the desire to answer those "what if" questions can be partially satisfied.")

For a game, whether person vs. person, or person vs. AI, some element of control is required, at which point the simulation aspect breaks down. Simple captaincy elements are most commonly added, i.e.  select bowlers, batting order, set aggression levels, even the field placements. This addresses the tactics of an individual match.

Alternatively, management aspects can be added. In this case, the game addresses the strategy for a cricket team/club. The most obvious requirement here is the selection of appropriate players to generate a balanced team capable of coping with the wide range of opponents.

Both captaincy and management games place extra requirements on the detail of a simulation. It is vital to build in more complexity to allow the human to influence matters. An example is field positions - to bring these into a game, you need a model that determines a trajectory for the ball. A good captain wants to be able to ask his bowlers to pick a line and length, which again requires a detailed model to handle any effects. For a management game, the way a player develops with time, age and experience, and reacts to his results must be handled. A good manager picks strong characters as well as talented cricketers.

Tim Astley's PBeM Ultra Cricket system would seem to be a good model of a management game. The emphasis here is on the strategy of building a squad, improving players with time and getting the right balance between player types (for one day and test matches). Secondly, the player must develop tactics for the match itself, including bowling sequence, instructions on dealing with declarations/run chases etc. The emphasis is on plans, since the manager cannot interfere when the actual game is played.

Empire's ICC combines the captaincy and management elements together. In my opinion, the management is the more successful, but has been effectively hobbled. The management job of selecting players, handling finances and contracts works and is supported by ageing of players, financial rewards and the vagaries of the international programme. However, the player has little evidence to make decisions, so is effectively blind. In addition, he must captain every match to bring the most out of his players as the AI is penalised when running the human's team. On the captaincy side, the lack of adequate feedback makes it difficult to develop intelligent tactics. The player doesn't know whether a batsman is picking the gaps in a field or the bowler is straying from the desired line. With no indication of an AI bowler's aggression settings, the player doesn't know how to react as a batsman. In real life, the batsman can judge the bowler's effort and rein in attacking instincts if necessary.

Further improvements are needed. A better means to work on enhancing a players abilities is needed for management games. Second eleven matches, tours, playing abroad off-season, for example. Enough recorded details to know if a player has succeeding in important/difficult situations before, and a confidence in the game model to know that means he has a better chance of doing so again.

On the subject of Arcade cricket, I have to concur with the analysis of problems with the genre. In common with most computer games AI, a player can find the repeatable moves to succeed with more regularity than in real life. Naturally, this leads to low AI batting totals, and high human totals. Of course, this is the problem with cricket - you only need ten good deliveries to bowl a side out. The biggest problem as I see it, is that for any level of realism, you need 20+ over games. To recreate first class/international cricket, you need 40+ overs. Any task repeated 240 odd times is bound to get boring. Playing real cricket is fun because its a sport, which involves exercise and adrenaline.

Having said this, I must admit to enjoying the latest Brian Lara Cricket. The problem is that I don't have the concentration to be a test batsman. Certainly not to be 11 test batsmen in a row. So I play silly shots after a while and get out, damming my team to low totals. So I'm forced to bowl as well as the generated innings are not similarly afflicted. What I would recommend, however, is to alter the game so that, for human vs. AI games at least, the best balance is for the player to control only one cricketer. Then you can pick a batsman and bowler. No player ability settings are required, except to control the game difficulty. If you are good enough to play as Gary Sobers, so be it. Then you only have to concentrate when batting for one, and possibly bowling. I would not bother to implement fielding, or if I did, only graphically if the ball actually approaches my chosen player. BLC has the engine, but not alas the functionality, to implement this now, and that way, one could play through a season, gaining a level of interest akin to that from simulations.

With such a change, and gradually improving control methods for batting and bowling, arcade cricket has a lot still to offer. Computer power is only now becoming good enough to represent a cricket match in a way that offers a player the chance to spot and react to the flight of a ball. As it improves, arcade simulations of batting will certainly improve. If the bowlers hand animation and ball in flight are detailed enough, we can do away with the on pitch targets and have a realistic, batsman's eye view.

Revised: Saturday, 16 March 2013.
Copyright 1998-2013 by All rights reserved. No part of these pages may not be copied without the permission of