ICU Ratings

If you have any information that would improve this article please contact [email protected].

Dawn of the Elo System in Ireland

The statistical system invented by Arpad Elo (who was a professor of physics and a chess master) was adopted by the USCF in 1960 and later by FIDE in 1970. Prior to 1967 the LCU had been operating a non-Elo rating system and a non-Elo Ranking List had been published by the ICU in 1966. In 1967 Oisín Ó Siochrú and statistician Jim Prendiville persuaded the LCU to both adopt the new Elo system and cede control of it to the ICU.

A number of tournaments were chosen to establish a new list based on Elo's system: Irish Championships 1966, 1967; Ulster Championships 1966, 1967; Leinster Championships 1967; Williamson Shield 1966, 1967; Ulster Team League 1966, 1967; City of Dublin 1966, 1967; Oireachtas 1967. The first live run of the system was the Myles Farrington Memorial 1967.

This was the beginning of a period (1967-1975) where the Elo system was in use but the rating lists were fragmented: an ICU list for top players only and provincial lists for the rest. Other names associated with this period are John Gibson (grading committee chair in 1972) and Paul Cassidy (grading committee chair in 1975). Some of the surviving documents are:

• 1968 note showing some of the earliest Elo calculations
• 1968 rating list 22 players headed by Heidenfeld, Littleton, Henry, Keogh, Cassidy
• 1969 rating list 32 players headed by Moles, Heidenfeld, Keogh, Henry, Littleton
• 1970 rating list 40 players headed by Moles, Patterson, Littleton, Heidenfeld, Keogh
• 1972 rating list 28 players headed by Moles, Patterson, MacGrillen, Henry, Heidenfeld
• 1972 note discussing the ICU and LCU lists

UCU Ratings, by David McAlister

My own involvement in ratings started circa 1973. From recollection the UCU was a bit concerned about the establishment of a 1900 rating requirement for the Irish Championship as there were relatively few Ulster players on the national list. It was at a meeting on 1 November 1972 that the idea of a UCU Rating List was first mooted. Myself and the secretary, Arthur Pinkerton, were tasked with producing it using results from the period 1969-1973. It was Arthur who did the leg work in providing old results, whereas I did the calculations and the first list was available late 1973. A UCU Rating Committee of myself, Arthur, Ron Henderson and David Blair was appointed the following autumn.

In the mid 1970s, Brendan Hayes did a great deal of work to produce a truly national rating system. Up to about 1976 the ICU, LCU and UCU all had rating lists and he coordinated this into a single list. At that time I was the UCU Grading Officer, so I think I must have been a member of the ICU Grading Committee.

My day-to-day involvement ended circa 1979 and Arthur Pinkerton took over (for the next 20 years). Robert McCaughey took over when Arthur retired to England (but I think that was at a time when the UCU was in one of its disputes with the ICU and had to do its own ratings). I was peripherally involved again when Vincent Bissett was in charge but all you had to do then was send SwissPerfect files to him.

Since the split in 2005, Drew Ferguson does the UCU Ratings.

I wrote up a history of the early years (1966-1970) of ICU ratings.

Establishment of a National List

By the mid-70s, due in part to the amount of "donkeywork" involved in rating calculations (especially after the expansion of chess interest due to the 1972 Fischer-Spassky match), the LCU list had become out of date and the rating of non-Leinster players had been abandoned. However, Brendan Hayes (pictured right), an actuary, who was appointed as chair of the grading committee in 1975, had a vision of a unified list with liaison officers in each province to help distribute the work. By the time Brendan finally retired from involvement with ratings in 1980, this vision had become reality.

Brendan temporarily retired in 1977 and Colm Fagan, another actuary and possible colleague of Brendan's, was grading committee chair for 1977-78.

Surviving documents from this period (1975-1980) include:

• 1975 letter from Brendan Hayes discussing the then current state of affairs and plans for the future
• 1976 letter from Brendan Hayes to USCF
• 1976 letter from Arpad Elo to Brendan Hayes
• 1978 rating list in spreadsheet form (recently compiled from source documents by Fred Harte)
• 1980 article by Brendan Hayes describing the ICU rating system

Sadly, in September 1992, Brendan Hayes disappeared in mysterious circumstances in a case that made national headlines.

Grading Matters, by Fred Harte

I took over as rating officer from the late Brendan Hayes in 1981 and stayed in the post until 1989. In the mid-80's I wrote a series of 4 articles for Fiacla Fichille under the heading Grading Matters to describe the workings of the ICU rating system as it was at the time. For the purposes of this history, the series of articles have been combined into one document, taking the opportunity to correct a few mistakes in transcription and repairing a couple of omissions. Other than that I have left the text unchanged, although there were some passages that I think could have been better expressed.

The ICU bulletin Fiacla Fichille was launched in December 1980 under the editorship of Tom O'Sullivan and continued quarterly until its demise in 1987. From 1984 through 1986 it was the vehicle for the annual rating list. Its successor, The Irish Chess Journal, was launched in December 1987 with Michael Crowe as editor. It eventually published rating lists, but that was sometime after my tenure. In 1986, 87 and 88 I had the lists published as booklets and a copy of the front page of one of them is shown to the right.

During my time in office, while Arthur Pinkerton and Dick Rellis were responsible for calculating ratings for Ulster and Munster respectively, I effectively had overall sole responsibility for the conduct of the rating system. No doubt it was not like this in the early days when all decisions presumably had to meet with the approval of a committee. With the arrival of Brendan Hayes in the mid-70's, my guess is that the influence of a committee may have been diluted as Brendan would have had the drive and mathematical expertise to take matters in hand himself.

I was also responsible for the operation of the Irish National Master award scheme. It was a fiddly, time consuming procedure with pen and paper calculations. For ratings I used a Texas Instruments Programmable 57 calculator for which I paid £20 in 1981.

Surviving documents from this period (1981-1989) include:

• 1981 instruction list for Fred Harte's TI57 programmable calculator
• 1984-86 series of articles by Fred Harte describing the rating system
• 1985 letter from USCF to Fred Harte
• 1985 article on USCF system (referred to in above letter)
• 1988 rating list 2595 players headed by Dunne, Delaney, Curtin, Barry, Orr
• 1989 fax from Fred Harte to Michael Crowe (handover notes)

From DOS to Windows, by Gerry Graham

I first got involved in chess organising in the mid-1980s, at which time the ICU rating officer was Fred Harte and he did it all with a mixture of calculator, pen and paper. He issued a rating report once a year, there were no interim updates as far as I remember and I'm sure if I looked hard enough in my attic I could copy of the 1985 rating list.

In 1989 Eamon Keogh commissioned (for the ICU, but at his own expense) a DOS-based rating program from Time-Out Software which was a huge leap forward. I believe it was Dick Rellis of Waterford that took over from Fred Harte and was the first rating officer to operate the DOS-based rating program. It was around this time that rating lists began to be published 3 times a year (September, January and May). When Dick stood down, I think John Crowley did the job for a few years. However, the DOS program had the drawback of not being able to introduce players at their initial rating performance. Instead all new players joining were given a base rating of 1000, which caused problems.

The idea for a new program came out of a chat at the bar during the Kilkenny congress of 1998. John Crowley convinced me that we'd never hold on to a rating officer very long because of the serious workload involved in operating the DOS-based system. I knew that one of my club-mates, Ray O'Brien, had the necessary computer skills to help out. Ray very kindly agreed to write a new program which would run on Windows. Pat Reynolds was the main technical adviser; his knowledge of several rating systems, as well as the flaws contained in the ICU system, was critical. My own input was relatively minor, I got the lads together, encouraged them along and helped with beta testing to stamp out bugs.

Input to the new system was in the form of report files produced by the pairing program SwissPerfect, which all tournament organisers were required to use. In addition, organisers were provided with a regularly updated file of ICU player data for import into SwissPerfect which meant that all they needed to input was each player's ICU code and all the other data would be entered automatically. When the tournament ended, the organiser would send the SwissPerfect report on floppy disk to the rating officer who could then import the data into the ratings program.

The new system was announced in November 1999.

The MSAccess System, by Ray O'Brien

Prior to 1999 there had been a DOS system in use for several years (run by John Crowley I think) however it would not run on Windows. Gerry Graham was on the ICU Committee at the time and was searching for a replacement system. He approached the company who developed the system used by the British Chess Federation with a view to providing a system for the ICU, however that didn't work out, presumably because the BCF system is so different from ours.

Discussing this with Gerry, I suggested that it should be possible to build a system using MicroSoft Access and he asked me to look into it. After spending a couple of weeks studying MSAccess I concluded it was feasible. Together with Visual Basic for building forms, it worked well, with excellent support for linking screens to data views.

I wrote all the code and based it on formulae used in the existing DOS system. Gerry extracted the data from the DOS system to an Excel file and did considerable work cleaning it up before it was imported to Access. Pat Reynolds provided some technical guidance on the statistical issues, advice on the possible effects of changes and some testing of the system. I used the same K-factors as the previous system which was based on the Elo system and two articles describing the Elo system in Chess Magazine from the 1970s (I think) to provide background on how the system should work. The first version of the rating system was demonstrated at Kilkenny in 1999.

The bonus system was taken from the earlier system using rules outlined by Pat. It was, as I understand it, an attempt to avoid deflation in the ICU system, which seems to have been a recurring issue. Under earlier systems ratings had been artificially increased by a few percent to compensate for a downward drift. The general belief was that this was caused by large numbers of young rapidly improving underrated players taking rating points off the more established players. I'm not sure if this holds statistically but we do seem to have lower rating compared to other countries. Of course players involved in international tournaments and foreign players in weekend tournaments in Ireland also help to restore the balance.

Earlier systems started new players at a fixed level and I felt this could also be a problem. Therefore I added the concept of using a performance rating over a player's initial games as their starting rating. One other idea I considered but didn't implement was to have a minimum floor rating so that juniors wouldn't drop below say 600 or 800 points.

Between 1999 and 2005 the system was upgraded several times adding abilities such as membership tracking, data exports and report generation. The last version (1.7) was in 2005 (see screenshot right).

As I recall Herbert Scarry was rating officer when the system was introduced and he was followed by Vincent Bissett. Before John Crowley and the DOS system I think Fred Harte performed early ratings calculations by hand. Even allowing for fewer players and tournaments it must have been a major undertaking. Although I wrote the MSAccess program, I was never the rating officer myself.

From Windows to the Web, by Mark Orr

In 1997 I started a web site called The Irish Chess Archive which eventually grew into the first official ICU site, www.icu.ie, in 2004 and from then until 2012 rating lists were published 3 times a year on that site. The calculations were still done using Ray O'Brien's MSAccess desktop app and rating lists were copied to the web server for display.

One rating per player every four months is just the tip of the iceberg as far as the data potentially available to players was concerned. I felt strongly that a great deal more rating information could and should be made available to the ICU's members with the help of the web so, with this in mind, I volunteered to be rating officer in 2006 in order to learn how the rating system worked. For the next 4 years I got to grips with the MSAccess system and also wrote a lot of new software (in Perl) to aid the process of accurately getting data into and out of the application without fundamentally changing how the system worked. However, even with this additional software, the rating officer still had a lot of work to do and I gained a deepened respect for my predecessors.

In 2010, I started a complete rewrite of the ICU rating system as a web application (using Ruby and Rails) with the following main goals:

• improve the rating information available to players
• simplify the processes of rating tournaments and publishing rating lists
• provide tools for tournament controllers to upload, view and edit rating reports

Finally in 2012, this site (ratings.icu.ie) was born. Members could login and get detailed information about their rating performances and the rating officer's job was much easier compared to what it had been. All calculations were performed on the web server and the venerable MSAccess desktop application was retired.

List of Previous Grading Committee Members

Prior to 1981, ratings were run by a Grading Committee, with a chairman and members, rather than a single Rating Officer as we have today. Information here on dates and committee members is a bit uncertain, but as best as we can tell:

• 1967-?? Oisín Ó Siochrú, Jim Prendiville
• 19??-?? John Gibson
• 19??-75 Paul Cassidy
• 1975-77 Brendan Hayes, J. G. Kennedy, M. A. O'Sullivan
• 1977-78 Colm Fagan
• 1978-80 Brendan Hayes

List of Previous Rating Officers

Since 1981 we have had Rating Officers who are responsible for day-to-day operation of the system. In the 80s and possibly the 90's, assistance was provided by provincial liaison officers.

• 1981-89 Fred Harte, Arthur Pinkerton, Dick Rellis
• 1989-91 Dick Rellis
• 1991-97 John Crowley
• 1998-99 Pat Reynolds
• 1999-00 Herbert Scarry
• 2000-05 Vincent Bissett
• 2005-07 John Shearan
• 2007-14 Mark Orr
• 2014-?? Andrew Kildea

Note: Mervyn Honner was elected as rating officer for 1998-99 but discovered he couldn't spare the time and Pat Reynolds took over.