Page 1 of 1

Tigers v Heffalumps

Posted: Sat Aug 22, 2015 5:47 pm
by IanCalvert
Simon Webb's anti-Heffalump strategy is well known: the weaker Tiger should play for a win against the stronger Heffalump by heading for complicated or unclear positions such that neither player has much idea what to do and hope that the Heffalump makes a serious mistake first.

Less well known is Webb's recommended team strategy for a team of Tigers playing a team of Heffalumps, much stronger players. He recommends that Tigers with White should play the opening safely and then "offer an early draw" which should at least "gain a psychological advantage"... if not an actual half point. [ I wonder if the Tiger team expected score, in the statistical sense, might even increase a very little from this psychological advantage.]

However as a 1b3 player, I find his team strategy unconvincing. At least with larger teams, when a Normal distribution for Tiger team's total score might well be assumed, I think Tigers should aim for maximum variance, to increase the probability of tail events like Tigers winning. The maximum variance will be achieved by all the Tigers adopting the traditional anti-Heffalump , go for the win, strategy.

What do others think are appropriate strategies for a team of Tigers against a team of Heffalumps?

Re: Tigers v Heffalumps

Posted: Sat Aug 22, 2015 6:04 pm
by David Robertson
I'm likely to have an above-average interest in the answers here :)

Re: Tigers v Heffalumps

Posted: Sat Aug 22, 2015 6:34 pm
by MartinCarpenter
Well the fundamental idea of 'shortening' the match is very valid. You're obviously much more likely to get a 'lucky' upset over 4 boards than 8.

Fully optimal probably won't often be as simple as killing all your whites, you'd want to identify a few boards you think you can probably neutralise and some where you've got a decent chance of a win.

Look at this card for a rather good example of my team (very efficient tigers by and large) falling foul of this sort of thing last season:
http://www.chessnuts.org.uk/ny5/match.p ... =382599173

The Sheffield E top 3 in that match are all legitimately very dangerous, so their best chance was always going to be killing the lower boards and hoping to come through there. Which they actually did quite nicely to some extent. We probably cooperated a bit too much vs the ideal really. We did I thought had it under control for 4-4 at worst, but it was thin margins and span out of control at the end.

There is an alternative scenario - here you're just a little weaker all the way down the match. In that case I think you want to keep everything alive as long as possible and hope to get them nervous/get a present. Generally as the stronger team you're lucking to tick off quite a few boards as 'under control' for a win or a draw before the match starts.

An example of this sort of thing:
http://www.chessnuts.org.uk/ny5/match.p ... =371916806

If you can only do that for a few of the boards you're slightly on edge before things kick off. Oh, a really crushing early win is priceless. Happens from time to time and I can vouch for it being incredibly disruptive to the stronger team. To the extent that the stronger team almost do better with steady but slightly weaker players than stronger ones who can crash and burn. It might be worth explicitly seeing if you think there's a chance of producing that on one or two boards and gambling.

Look at what happened here:
http://www.chessnuts.org.uk/ny5/match.p ... =322633729

Well, actually, about 4 of those boards were legitimate enough contests (our 5/6 would have been on 7/8) but really an awful lot of this was down to me going down in quite horrible flames very early on. That got everyone nervous and.....

Re: Tigers v Heffalumps

Posted: Sat Aug 22, 2015 6:37 pm
by IM Jack Rudd
Pushing up variance is good, but pushing up expectation is better. The point of the early draw offer is to push up expectation.

Re: Tigers v Heffalumps

Posted: Sat Aug 22, 2015 8:44 pm
by Tim Harding
In the World Senior 50+ team tournament back in March - over four boards - the Slovakia team ("weaker Tigers") with an average rating difference of about -95 employed this strategy against England A (Short-Nunn-Speelman-Arkell). They agreed quick draws as White, holding out comfortably against John Nunn (who was ill, though whether they knew that is unclear) and punishing errors on board 4.

Since the scoring was by match points and there was only one other team [GER] of comparable rating to them in the tournament, Slovakia repeated the strategy against Germany. Again the Germans accepted the two draws and later a third one. As I recall, the match was drawn too because Tischbierek missed lots of wins in the last game against Timoscenko [who even missed a win himself close to the end].
Of course the strategy would have misfired had Keith exploited his opening advantage or Tischbierek won that game.

The upshot was that Slovakia comfortably won the gold medals and England came third.

So this looks to have been a deliberate match strategy by Slovakia: narrowing the front plus maybe as said above affecting expectations in the games continuing.

There was a lot of discussion about this on the Senior Chess board here at the time.

Re: Tigers v Heffalumps

Posted: Sat Aug 22, 2015 10:16 pm
by MartinCarpenter
4 boards is random enough to start with of course :)

One thing worth noting is that safely killing a game with white isn't easy. A definite skill, and also useful to pick your opponents if you can.

Quite easy to slip into a slightly passive position which lets the tigers grind at you with black to go with their white games - the last place you want to end up is being ground at over a lot of boards at once. Nil chance of survival/upset then and you'll very likely lose by a healthy margin.

Re: Tigers v Heffalumps

Posted: Sat Aug 22, 2015 11:45 pm
by Roger de Coverly
IanCalvert wrote: Less well known is Webb's recommended team strategy for a team of Tigers playing a team of Heffalumps, much stronger players. He recommends that Tigers with White should play the opening safely and then "offer an early draw" which should at least "gain a psychological advantage"... if not an actual half point.
If you are the match captain of the relative Heffalumps, you ban early draws, unless players are seriously worse and need a "respect" bailout. Remind them how many grading points they are losing.

In terms of bashing the Heffalumps, it is of course true that titled Heffalumps know all theory, or at least that's what they would have you believe. In practice, as they get older, they get lazier, so if you hit them with new ideas preferably from top 100 players, they may not be as familiar with it as their reputation might suggest.

Re: Tigers v Heffalumps

Posted: Thu Aug 27, 2015 9:50 am
by MartinCarpenter
Having idly thought about this a bit more, I think the main thing is to make sure you've got a 'realistic' narrative of how you might draw/win the match.

If your expected scoreline over 8 boards is 3-5 or 3.5 - 4.5 then that narrative is basically just playing well/getting a little luck, so fine to do that.

Once it slips out to 2.5 - 5.5 or 2-6 I suspect you need to do more than that. So you need to work which boards you can hope to reasonably randomise the result on - time trouble addicts, theoretical knock outs, simply random play depending on the players. And which you can have decent chances of neutralising. Partially to improve your overall chances a bit, and partially just to make the team believe they can get a result.

You still won't have great chances of course, but I'm sure you'd get a few results over time. What I can say is that just 'hoping' genuinely doesn't seem to work at all: York A say have typically had an expectation of 6-2 vs the weaker teams in the Yorkshire league and we've run up something daft like +76, =2, -3 over the last 9 seasons.

We'd have expected a big plus score over time of course, but still! Once it gets out to 6.5 - 1.5 and above, then I guess you just try and enjoy it on an individual level :)

Re: Tigers v Heffalumps

Posted: Thu Aug 27, 2015 5:25 pm
by Roger de Coverly
MartinCarpenter wrote: York A say have typically had an expectation of 6-2 vs the weaker teams in the Yorkshire league and we've run up something daft like +76, =2, -3 over the last 9 seasons.
It rather suggests York A are quite ruthless. No agreeing draws early because they don't like the position, or agreeing draws once the match is settled.

Re: Tigers v Heffalumps

Posted: Thu Aug 27, 2015 6:16 pm
by MartinCarpenter
Well, we're not perfect but we do have some very efficient players and much the same core as White Rose to keep people in line.... Still, from what I remember, when Sheffield/Bradford A have fielded teams with ~6-2 expectation they've done very well too.

Partially perhaps because most of the teams in the league haven't often really seem to try anything specific/quite believe. I can remember an awful lot of matches where = to +/= on a bunch of boards has eventually turned into a 'surprise' 6-2 win.

Of course a 6-2 expectation simply is quite hard to overcome - a 1.5 point cushion is an awful lot of accident insurance. Still I guess you could get up to maybe 10-15 per cent odds of doing something?

Re: Tigers v Heffalumps

Posted: Thu Aug 27, 2015 9:48 pm
by MartinCarpenter
Some best case figures for weaker teams (all boards being win/loss with no draws.).
6 - 2 ~ 8% draws and 3% wins.
5.5 - 2.5 ~ 16% draws, 7% wins.
5 - 3 ~ 22% draws, 14% wins.

Ideal case for the stronger team is of course all wins/draws for a huge percentage overall. For a 6-2 match (25 grading pts/board), quite a few people do actually get surprisingly close to achieving that in practice. York's score over time probably isn't especially dramatic.

To see how 'shortening' a match might sometimes work - a 4 board match with a pure 75 - 25 split is about 20% to draw, 5% to win. That's actually slightly ahead of the expected score for a 5.5 - 2.5 match.

All hugely ballpark of course, because you very rarely get a match with even strength differentials on every board :) Or players who can smoothly switch between random and solid. Also the boards in a match aren't remotely independent events!

Re: Tigers v Heffalumps

Posted: Sat Aug 29, 2015 7:01 pm
by IanCalvert
Martin Carpenter wisely wrote, "I think the main thing is to make sure you've got a 'realistic' narrative of how you might draw/win the match.

If your expected scoreline over 8 boards is 3-5 or 3.5 - 4.5 then that narrative is basically just playing well..." .... but maybe all the other ideas including improving expectation and getting "good" , colour dependent, game strategies are even more important in the close match case.

Maybe in some matches , discussing , before a move is made, the games where individual Tigers (against specific opponents with specific colours) might be aiming to :

(i) Maximise the chance of winning by trying to avoid a draw (e.g. against Berlin players) or (ii) Avoid losing or (iii) trying to maximise the chances of a draw (if an individual Tiger is weaker than his Heffalump opponent ..or ill)

might well also make a big difference to the probability of the team winning a close match , as mathematical analysis suggests is possible.