Knockout team selection

[Rob Thompson]

Just thinking about how people choose who to put in grade-limited knockout matches. Do you play strong players over the top boards, in an attempt to win on boardcount? Or do you play a fairly balanced team?

The reason i ask is because last night i was captain of a school team in the knockout competition in our local league. 6 boards, total U-750. The lineups were as follows

Us - Them
173 - 148
159 - 140
157 - 136
100 - 115
86 - 100
73 - 106

748 - 745

I chose the "points up high and boardcount if needed" approach, but overall the teams were very balanced. I was wondering what others thought about the way to select players for matches like this.

[Joey Stewart]

There are three good ways to do it.

1. One or two very strong players combined with the rest of the team well under 100 - the bottom guys drag the handicap up so high that the top two are able to singlehandedly win the match regardless of other results. (My school team uses this lineup by default, since there are two of us graded 179 and 190 and the rest have been under 100 for the last few years)

2. The balanced team - usually a good one, and does not suffer as heavily against low rated handicaps but will not provide much punch against a strong opposition and can end up being beaten badly over the board. It is nevertheless the most commonly fielded side.

3. A super squad, made up of the very strongest players available (N.B. Not all clubs are going to be able to pull this off, as such players are rare) and ideally having 200+ at the top and dropping to 160-170 at the bottom.
Relies on winning more or less every board, but is very intimidating to the opposition.

Under rated players are always a massive bonus in these matches too, either because they had one bad season, are an improving junior or simply one of the many people who 'should' be playing to a higher standard but for some reason their rating never goes up by much.

[Alex Holowczak]

The Cannock League has this sort of structure, but the limit is 640 points over 4 boards. My Uni is forced to play with players graded 211, 137, 132 and 110. So we're comfortably under the cap. However, if our opponents had four players graded in the 140s, I reckon they'd be favourites.

[Alan Walton]

Though my team of 3 160's and a 159 should be favourites ahead of your team of 140s

[Neill Cooper]

Apart from large clubs, often you have little choice. My preference would be to play the most under-graded players.

Of course, if you know your opponents team layout your can try to maximise your chances of, say, winning 4 out 6 boards. It reminds me of Non-Transitive Dice: On average die A beats die B; B beats C; C beats A. But as in results predicted from chess gradings, it it only some of the time that they win.

http://en.wikipedia.org/wiki/Nontransitive_dice

[Rob Thompson]

With regard to knowing the opposing team, as we were playing the club i used to play for, i knew (more or less) who we should be facing. However, the secretary of said club is also the teacher who organises the inter-school events at my school, so he knew the sort of team i'd go for.

There was also the added complication that my original team included their board 1 as my board 3, and their board 3 was also from the school, but all bar me of our team play for that club in other divisions...

[Joey Stewart]

Its not unheard of for a player 20-30 points lower rated to draw or even win a game, and the stronger guys HAVE to win their games to win this match - its surprising how much pressure there can be when you know there is no other acceptable result than a win.

[Alex Holowczak]

The grading database is built on the assumption that a difference of x points gives you a (x+50)-(50-x) predicted score as a percentage.

So...

173 - 148 -> 62.5 - 37.5
159 - 140 -> 59.5 - 40.5
157 - 136 -> 60.5 - 39.5
100 - 115 -> 42.5 - 57.5
86 - 100 -> 43 - 57
73 - 106 -> 33.5 - 66.5

The mean result would be 50.25 - 49.75

However, if you look at the match total

748 - 745 -> 51.5 - 48.5 - a difference of 3 points, but 3 points/6 boards = 0.5, which is what you get above.

So according to the statistics, it shouldn't make a difference as to how you compile your team.

It depends on where the threshold of converting a win from a draw lies. Given you have 3 of the 4 highest differences though (boards 1-3 trumped only by board 6), it seems logical that you have more chance of winning the match than your opponents, because the win/draw threshold can be as high as 60 - 40 for you to win the match.

[Alan Walton]

Some people say it is easier for a 180 to beat a 150, than a 150 beating a 120.

My view is people graded 120 would have a higher probability of beating 150s, than a 150 would have beating a 180.

This maybe because the higher proportion of players within the 100-160 bracket, than say those graded above 180

[Eoin Devane]

What was the actual result, Rob? Did your plan work out?

[Rob Thompson]

we actually won 4.5 - 1.5, with wins on 1, 3, 4, 5 and a draw on 2. Such are the vagaries of a junior team

[Phil Makepeace]

The grading database is built on the assumption that a difference of x points gives you a (x+50)-(50-x) predicted score as a percentage.

Isn't it just (x+50), where x cannot exceed 40 or be less than -40?

[Alex Holowczak]

The grading database is built on the assumption that a difference of x points gives you a (x+50)-(50-x) predicted score as a percentage.

Isn't it just (x+50), where x cannot exceed 40 or be less than -40?

Yes, I meant they were the two scores. So a player might win 90 - 10, where x = 40, the difference in the grades, above.

You're right that I forgot the 40-point constraint, though.

[David Williams]

Perhaps your best chance is if there is one board where you are heavily outgraded.

Team A has two 200s and two 50s. Team B has one 200 and three 100s. Should be a win for team B.

But what about Team C, with four 125s? They would be marginal favourites against team B, but will only draw with team A.

[Alex Holowczak]

Perhaps your best chance is if there is one board where you are heavily outgraded.

Team A has two 200s and two 50s. Team B has one 200 and three 100s. Should be a win for team B.

But what about Team C, with four 125s? They would be marginal favourites against team B, but will only draw with team A.

The thing to be careful of is that there might be a rule similar to the Cannock League: Anyone under a certain grade (or for that matter, a default) has the grade for that board immediately set to a minimum threshold. So if the limit is 640, you can't field 210 210 210 d, because you'd be at 730 (the default is 100). Similarly, 181 180 180 39 would overstep the limit, as the total would not be 580, it'd be 641.