Predicting the Predictable - II

I believe that ODIs follow a predictable pattern and, as the title of this post suggests, the aim of this and the previous post is to prove the same. And since this is a continuation of the previous post, I would strongly recommend anyone who has not read that post to do so before reading any further.

In the previous post, I had highlighted some of the most important attributes and also, in my opinion, flaws of the one day game, namely:

• There exists some kind of a par 1st innings score that gives both teams an equal chance of winning
  
• If the game moves away from this par score, the chances of winning and losing change dramatically. And so do the margins of victory and defeat.

I am sure a lot of you will question the existence of these attributes based on one graph. To begin with, the game has changed over the years. Add to it the fact that the nature of the game played in various parts of the world is different. Hence, to say that the above attributes represent ODI cricket in general appears to be unfair and flawed.

Fair enough. Frankly, even I had my doubts before I started this analysis couple of weeks ago. Since then, I have analysed the One-day game to see if the above mentioned attributes do exist and are common to ODIs in general. I did this by grouping ODIs together based on different parameters.

First, here is a comparison of games played in this decade (2000 thru 2005) with all games played since 1971.

Note, as mentioned in my previous post, these graphs do not include games decided by the D/L (or similar) method, less than 45 overs (as they distort the numbers) and those that did not finish in a +ve result (i.e. "No Results" due to rain). Also note that in this (and all graphs), the line going from bottom left to top right represents the % times the team batting first has won after scoring the runs represented in the "x" axis, and vice versa. This gives us a par score (50% chance of winning or losing) where the lines intersect.

(Please click on this and other graphs to enlarge)

As you can see, apart from the difference in the par score, both those graphs are similar. And as expected, the par score in the games is a bit higher for games played in this decade.

Now, given the fact that ODIs in the sub-continent are normally higher scoring than the rest of the world, I split the games between those played in the sub continent and elsewhere and did a similar comparison...



The sub-continental graph does appear "wonky" but on close inspection, it exhibits exactly what I am trying to prove, i.e. outside of the par score, the result becomes predictable, and in this case, its just more pronounced. And again, as expected the par score is higher for the sub-continental games. Guy who might still be holding their sheet of paper tick or cross your answers :-)

I was also aware that this split based on regional lines may not be correct, as not all grounds in the sub-continent are high scoring and vice versa. In order to eliminate this, I split the games based on the average run-rates for individual grounds. I used 4.50 RPO as the basis for the split (simple because this gave me about 1,000 ODI s on either side). And here is what I came up with...



And, just to be absolutely sure that all the above graphs did indeed exhibit more or less similar characteristics, I merged all of them into one graph...



Funny enough, any which way you slice and dice this damn game, it does not seem to change. I am sure if I had picked up a random group of ODIs, I would have seen the same result. i.e., existence of some kind of par score, and either side of it, the game starts to get one sided.

Now, you may ask, so what, is this really a problem? In my opinion, yes, because given the 1st innings total, and with enough knowledge of the ground, era, etc, it becomes easy to predict the outcome of the game. And I am talking of games involving two evenly matched teams, (not Australia Vs Namibia kind).

And anyone need further convincing need not look any further then the last few games of the VB series being played in Australia. In fact, next time you sit down to watch a One-Day game; try to guess the result after the first innings is over. And see how many times you are correct.

Now, logically speaking, if there is a par score, it implies that most 1st innings scores should be finishing around that score, no? Well, this is another aspect of this whole analysis that struck meas odd (and something i have not been able to understand). See the % of games that fall into these different "run-bands"...



Talk about similarity here!!

Now, it would be appear that the ideal situation would be for more games to have the 1st innings scores around par score as it would be difficult to determine the result of those games. Frankly, this would make the whole game boring and the excitement would only kick-in at the death of the second innings (say 5-10 overs left). A bit like the 250-270 games we are so used to seeing in the sub-continent.

So is there another solution by which no matter what the situation of the game at the mid-point, it becomes difficult to predict the result? Or in other words, can the element of par be eliminated?

In my opinion, it can. But not by making flat pitches or introducing meaningless rules like Powerplays or Super-subs. All they do it increase the runs being scored in the game but, as I just proved, does not change the nature of the game...



Anyway, to understand how this par can be eliminated, I will need to compare ODIs with other "Set-target, Chase-target" kind of games like Test cricket and Baseball (actually, none other come to mind)!

And as a starting point, I will analyse how easy (or difficult) it is to determine the par score in Test cricket and why.

But that in my next post.