Cricket is all about stories that have changed the course of the game. Some of these changes were caused abruptly by certain events like The Bodyline series or the underarm bowling by the Chappell brothers while others like the shape of the bat have meandered through the game. 2018 was going to be an interesting year for me as far as cricket analysis was concerned. I had done my research, developed my methodology and my algorithm. I was nervously getting ready to launch my Cricket Index and expected an uphill challenge to prove that my index would be better than the ICC ranking but then Sandpapergate came up and that was all that was needed to prove that a new way to think about rankings of teams and players was required. Steven Smith, Cameron Bancroft and David Warner were banned from playing cricket in March 2018 for one year but it was not until September of 2018 did Virat Kohli achieved the number one spot by dethroning Steven Smith. During that time David Warner stayed at number five except for September 2018 when he climbed one place to number four and it was not until January 2019 that he started to slip down the ranking. I will concede that Joe Root’s average fell during April to September but a person who is going through a bad patch should not be punished when compared to someone who is not playing. The World Test Championship (WTC) was announced in October 2017 and the points system announced later on but therewere flaws not only in the organisation of the format but the distribution of points too. Surely a better solution had to exist where an equal number of points were shared irrespective of how many Tests were played further the fact that the WTC has only nine Test playing countries and not the entire 12 compounded by the countries do not play each other because of the ongoing conflict between India and Pakistan. The Future Tours Programme (FTP) will continue until 2023 which would make entire series meaningless to a certain extent. Thankfully Australia and England play each other if this was not the case then the Ashes which is the most iconic series in cricket would have been of lesser value.
New Indices and statistical calculations come to the forefront because there is someone who comes up with an idea that a certain system is better than the one that is currently being used or is more popular. Duckworth Lewis Stern came into play after the match between South Africa and England in the 1992 World Cup ended up with South Africa needing 22 runs off 1 ball which ultimately ended up being further revised which meant that England won by 19 runs. The flaw in the “most productive” overs was exposed because South Africa had two overs that were maidens. The Duckworth Lewis system finally got implemented in 1998 with the first international game where it was used played in 1999. That was a case where a change came about because the system was proved to have a significant weaknesses.
I tend to be slightly contrarian so it comes as no surprise that I prefer alternative indices to traditional ones. In my opinion The Jayadevan system is probably a better alternative to Duckworth Lewis Stern but I am practical enough to recognize that there is so far nothing significantly wrong with Duckworth to warrant a change and the debate that will ensue. These days teams can bat until number eight if not lower down this essentially makes the resources left immaterial. There are international as well as domestic teams where the batsman coming lower down the order are probably more dangerous. The loss of resources in the DLS system which is defined as the number of balls left and the number of wickets or the run rates in the Jayadevan system depending on which stage of the innings the game is at have their own shortcomings but a slightly flawed system which is significantly more accurate is better than none at all.
The second reason why new indices crop up is when the main or poular index is not a reflection of reality. A classic example of this would be the difference between the MSCI Emerging Market Index and the FTSE Emerging Market index. The MSCI index has a 15% exposure to South Korea. The problem with this is that the financial markets in South Korea are more closely linked towards developed markets than they are to emerging markets. Similarly, the ishares S&P GSCI index has a 70% exposure to energy. The PowerShares DB Commodity Index has 50%. When I first started looking at Exchange-traded funds I had to make a convincing argument to use the Vanguard Emerging Market ETF and the PowerShares DB Commodity index rather than the alternatives because they were bigger. Just because most of the people used those indices then that does not mean they were right. Today the Vanguard Emerging market ETF and the PowerShares DB Commodity Index is bigger than their former rivals.
The main reason for creating my Cricket Portfolio Index is because I think that the ICC ranking is not reflective of reality. David Kendix who designed the system approached the ICC ranking as an actuary and a statistician which even though is a good system have too many biases to be considered accurate. I view team rankings as a financial portfolio where the 11 players that are fielded for a particular game are the investments. These players in turn are ranked by an algorithm that comprises various factor which I have developed to minimise the biases that are commonly seen in financial investments and ranking systems. I have listed the main index biases below and mention how the ICC deals with them and how my system differs to the ICC specifically in relation to these biases.
1) Classification Bias: We do not live in a cricketing world anymore where each player has clearly defined roles. A great Wicket-Keeper with average or below-average batting will always be lower down in the pecking order than an average keeper who bats extremely well. Batsmen are easiest to classify as they generally do not bowl. Sometimes even they would have to put in a few overs either to give the main bowlers a rest or to add a little bit of variety. Everybody is expected to score runs. Bowlers are no exception. If by chance they happen to score well then those are just bonus runs. The ICC multiplies the batting points by the bowling points and divides that by 1000. If someone does not get any points ie. their batting points or bowling points are zero then they do not appear on the list. In the top ten of the all-rounder list, you probably have five genuine all-rounders. The others are there because they bowled really well and batted a little. Finding out whether a player is an all-rounder is as easy as to check their role in the team. England uses Ben Stokes as their fourth seamer and Moeen Ali as their second spinner. Moeen got dropped because he was not bowling well and neither was his batting backing him up. This essentially meant that the main bowlers were shouldering all the bowling. My system will give All-rounders a higher chance of scoring more points. The penalties in tests will not be as significant as in ODI or T20. My algorithm is designed in such a way that pure batsmen and pure bowlers are also given a fair shot. If you have five batsmen and the wicketkeeper who can get you the runs followed by five bowlers that can get you 10 wickets then there is no need for all-rounders.
2) Survivorship and Attrition Bias: The ICC ranking removes the results of matches scored between the 37 and 48th month every May. Further, the matches played up to 24 months gets counted but those before that get a 50% weight. This index has tried so hard to remove survivorship bias that it has gone and created a different set of problems. Giving a weight of 50% to matches older than 24 months implies that batsmen either become 50% worse off after two years or that 50% of your team no longer exists with a side being completely changed after four years. Can you imagine what the results would be if such a thing actually happened in teams? If players retire or are dropped then their results will still be reflected for a period of time. Chopping and changing of players in teams have little impact on the ICC rankings. My solution to the survivorship bias was to take the performance of the playing XI in the team. A team can be ranked number one before the start of the test series. If they make changes in their teams and bring in uncapped players then they will lose points and may end up starting that series with a lower score than their opponent. As mentioned earlier I view each team as a portfolio and each player as a constituent of that portfolio. If a player is not playing then he has no place in that portfolio.
3) Selection Bias: By ignoring previous data of players and teams; there is a heavy selection bias that seeps through all the pores of the ICC ranking. Past performance is not a guarantee for future results but those results should not be ignored. A player should not be allowed to either jump up rankings or have his performances negated just because they do not fall within a specified time. I do not have any intention of removing player performances just because they are dropped or a period of time has elapsed.
4) Series and Team Bias: The ICC gives bonus points for a series win. I do not believe this should be the case. I assign scores to players and their scores are reflected in the team scores. There is no point in assigning a team further points. Playing longer series is more rewarding than playing shorter ones. This should not be the case. I completely agree that there are weaker teams and stronger teams. I do believe that if a team is deemed worthy to be playing test cricket then they should be able to play on the levels demanded by Test Cricket. I do not give any weight to stronger or weaker teams. The ICC considers weaker teams to be where the difference in points is greater than 40. Considering this India which has 125 points as of 15th May 2018 has only Bangladesh, West Indies, Zimbabwe and now Ireland and Afghanistan as weaker teams. Australia who is ranked third has only Zimbabwe, Ireland and Afghanistan as weaker teams. Australia has cancelled the series against Bangladesh for financial reasons. Based on the ICC ranking system; they have cancelled a test series against an equally strong side. Classifying a team as stronger or weaker is meaningless especially in T20’s.
In my system players are rewarded and penalized for certain milestones. Additional batting points in Tests are given for reaching a minimum of 50 and then in 25 run increments with an equivalent system for ODI & T20. Points are given for 4’s and 6 as well as maidens in One day and T20’s but they don’t figure in Test matches as the amount of runs that a player accumulates is more important than how they get it which brings in the question of strike rate into question. Needless to say strike rate is crucial is the limited overs format but how or whether to factor it into Test matches. The third Test between Australia and India is an example where the high strike rate of Rishabh Pant that nearly won the game for India was probably as important as Hanuman Vihari’s 23 off 161 balls that helped India draw the Test instead of losing it. Fielders get more points for catches than Wicket Keepers but Wicket Keepers benefit from stumpings because it requires more technique and skill. I do not differentiate how the bowlers get their wickets because at the end of the day you need to take wickets to win a game so what matters is how much they get and what their strike rate is. Economy rate comes into play for ODI and T20 games. The algorithms for Tests, ODI’s and T20’s are different but it has been designed in a way that points scored by a player in one format can be more or less compared to one in a different format. A person getting 1,500 points in Tests is as good as a player that scores 1,500 points in a T20. However the weight I give at a Team level differs and I have made this as simple as possible. One Test is the same as five ODI and one One ODI is the same as two T20’s.
My system is by no means perfect and I have no doubt that I will tweak it more in time and people may find discrepancies in it and I welcome constructive criticism. One of the biggest flaws in my system is the high likelihood that people who play more games are likely to get more points. My solution to this was to measure the performance of a player in a year by using the points he scored and the performance of a player over time by annualizing his performance. In effect I would have two playing XI’s. Retired and banned players are immediately taken out of the individual rankings.
The final component in the ranking system for teams is the “cash” element. This is simply the surplus or deficit that occurs when a player is taken from the playing XI and replaced with another player. The end result is a close reflection of how strong the team that takes or last took the field actually is based of the performance of the individuals in that team. This is the main reason why teams of one era can be compared with those of another.
I look forward to discussing this with anyone so feel free to post comments here on Twitter or DM me there.