Competitions Pieman Rd 11 v Brisbane 7:40pm Sat 8 August (Gabba)

[dogwatch]

DISCLAIMER: The following discussion only refers to the Pieman Mk II concept and is unrelated to the regular Pieman 2020 comp.

And here are the hypothetical Kelvin results. What I did was randomise each team's selection strategy with randomisation based on the following weighting:
1+1+1+1+1 = 37.5%
2+1+1+1 = 25%
2+2+1 = 12.5%
3+1+1 = 12.5%
3+2 = 12.5%

Then, according to each team's randomised strategy I selected random players from the 22 selected to play Brisbane. I weighted these toward picking players with the lowest handicaps (i.e. those more likely to be kicking goals) so players like Keath, Daniel, Wood and Cordy were unlikely to appear.

Here are the results. So it's congratulations Sharkey66 on winning a hypothetical Kelvin! *

* Achievements don't come much hollower than that.



As you can see, you can effectively choose a banker any time you want simply by doubling up ... or even tripling up. It does of course cost you a pick of another player each time you do it. The other obvious benefit of this approach is that you have 5 certain "live" selections every week, unlike the regular comp where sometimes you are down to one or two players in your team actually playing in any given week.

A couple of interesting features to note were

• Two of the top 3 places went to teams who chose 3x plus 2x but the other place went to a team that chose 1+1+1+1+1. I don't know if that's an indicator of strategy or just the way it played out in one game. I did a what-if and took one goal off Wallis just to see how it would change things and the top FOUR places went to players who had 3+2. So that startegy succeeded this week for some but equally it could flop big time (see dogwatch's score!)
• The top 20 or so scores were all fairly close. No Vandermeeresque blowouts. That's what you want because it keeps everyone interested and in with a chance. And even if you get a zero score (like I did ) there's an opportunity for a complete re-set the following week.

With the above I didn't do any bonuses for getting 4 different goalscorers. I am thinking along the lines of say a 20 point bonus for 4 different goalscorers (and possibly a 40 point bonus for 5 goalscorers but that is probably overkill). If I had applied the 20 point bonus Sharkey66 would still have won but Bont2Bruce would have come in second on 105 pts. It's just an option I am thinking about at this stage to keep a teasing balance when you try to work out which is the best approach (doubling/tripling up or going for 5 different players).

FYI. Comments and ideas welcome.

 

[grassman75]

So am I reading it correctly I can pick Wallis 5 times in the one pie team and get his score 5 times but I take the risk if he doesn't score I get nada

It's shame you cant pick a defender to have goals scored against him....Gardy would have been $1:01 fav.

 

[dogwatch]

So am I reading it correctly I can pick Wallis 5 times in the one pie team and get his score 5 times but I take the risk if he doesn't score I get nada

...

No. Maximum of 3 times.

It's not just a nannystate clause to protect you from yourself. (You're old enough to figure these things out for yourself.)
It's also to prevent excessive swings/blowouts that might spoil the comp. However it's something I'm open to persuasion on.

 

[BEaston]

So am I reading it correctly I can pick Wallis 5 times in the one pie team and get his score 5 times but I take the risk if he doesn't score I get nada

It's shame you cant pick a defender to have goals scored against him....Gardy would have been $1:01 fav.

the most you can pick a guy is 3 times. But yes, that's the concept.

I think with our dearth of goal scorers the 2+3 mode has an inherent strength

 

[BEaston]

dogwatch could you run those teams for every game this season to see the results?
(sneaky feeling I come out on top - likely the only way I'll win a pieman, in a pretend comp)

 

[grassman75]

I like it, just thought I was misreading it.

I see your 3 Wallises and raise you 2 Vandermeers

 

[dogwatch]

dogwatch could you run those teams for every game this season to see the results?
(sneaky feeling I come out on top - likely the only way I'll win a pieman, in a pretend comp)

You may not be serious but ...
I can keep yours the same if you want but I really want people to choose their own if we're going to iron out the wrinkles. Anyway you can pick those same players each week yourself if you like.

At this stage my gut feeling is that it's a better comp than the existing Pieman but if everybody hates it I'm prepared to recant.

the most you can pick a guy is 3 times. But yes, that's the concept.

I think with our dearth of goal scorers the 2+3 mode has an inherent strength

Maybe. That's the sort of thing I'm aiming to find out.

It's also why I'm contemplating a 4-goalkicker bonus.

 

[BEaston]

You may not be serious but ...
I can keep yours the same if you want but I really want people to choose their own if we're going to iron out the wrinkles. Anyway you can pick those same players each week yourself if you like.

At this stage my gut feeling is that it's a better comp than the existing Pieman but if everybody hates it I'm prepared to recant.

Maybe. That's the sort of thing I'm aiming to find out.

It's also why I'm contemplating a 4-goalkicker bonus.

No I meant just for the proof of concept trial, just to have more data than just one game. I understand the selected players and handicaps will change each week in the Real Mk II but I think running retrospectively for the whole season might show up any patterns/flaws as well as showing how big any potential gaps between top and bottom could be. Not sure how long this would take, or how much free time you have, but it could be worth doing.
(when I said winning a pretend comp, I meant it.... PoC trial of a trial haha)

I definitely want to choose my own players for the actual Mk II trial

 

[TedDougChris]

I'm still in with a good chance of 40th !!!!

 

[Sharkey66]

Woohoo won a game I didn't even know I was playing.
Don't know if it would be too much work, but would like to see the original piaman go back similar to the early days as more of a set and forget comp, ie pick your teams at the start of the year, no or limited trades and banker with this new game played concurrently.

 

[dogwatch]

Woohoo won a game I didn't even know I was playing.
Don't know if it would be too much work, but would like to see the original piaman go back similar to the early days as more of a set and forget comp, ie pick your teams at the start of the year, no or limited trades and banker with this new game played concurrently.

This is a really valuable perspective for me as I was wondering whether many (or any) people feel this way.

My original plan about 10 years ago was for a set-and-forget comp so people didn't have to remember to do something every week. Or they sacrificed their chance of winning if they forgot to do so.

However the impact of LTIs led to the creation of trade periods. There was also a push from some players to be able to exercise some tactical influence on their chances as the season progressed, hence the banker concept.

I can definitely see both points of view but the comp has drifted more toward the regular interaction model rather than set-and-forget. I don't know whether that's because it's more popular that way or because the other (set&forget) players have just stopped playing.

It's probably not too much work to run both in parallel, but they may have too much in common to justify having both on the same board. I'd be happy to hear other viewpoints on this.

All good food for thought anyway. Thanks.

 

[dogwatch]

No I meant just for the proof of concept trial, just to have more data than just one game. I understand the selected players and handicaps will change each week in the Real Mk II but I think running retrospectively for the whole season might show up any patterns/flaws as well as showing how big any potential gaps between top and bottom could be. Not sure how long this would take, or how much free time you have, but it could be worth doing.
(when I said winning a pretend comp, I meant it.... PoC trial of a trial haha)

I definitely want to choose my own players for the actual Mk II trial

The difficulty in doing it retrospectively is knowing what handicaps I'd have issued that week. I can't unknow what I know now.
Or maybe the handicaps won't matter that much? I could just use the same handicaps as Rd 11 for the purposes of the exercise and do it over a few representative rounds (eg a 5 goal game, a 7 goal game and a 14 goal game).

I'll see how much spare time of got to muck around with it.

 

[BEaston]

The difficulty in doing it retrospectively is knowing what handicaps I'd have issued that week. I can't unknow what I know now.
Or maybe the handicaps won't matter that much? I could just use the same handicaps as Rd 11 for the purposes of the exercise and do it over a few representative rounds (eg a 5 goal game, a 7 goal game and a 14 goal game).

I'll see how much spare time of got to muck around with it.

yeah I was thinking just leave the teams and handicaps as they were for the single round you did. As they are random it would be somewhat reflective of a normal round anyway (maybe?). I dunno just thought it might be useful, but might be pointless haha.

 

[dogwatch]

yeah I was thinking just leave the teams and handicaps as they were for the single round you did. As they are random it would be somewhat reflective of a normal round anyway (maybe?). I dunno just thought it might be useful, but might be pointless haha.

If I leave both Pieman teams and handicaps unchanged it should be easy. Only problem would be the actual change in WB playing personnel in other rounds (eg lots of games with no Naughton, Dale and Lloyd).

What I can do is make the entire thing a fantasy and plug in some randomised goals in games where we (hypothetically) kicked 5, 9 and 14 goals.

Hold my beer ...

 

[dogwatch]

OK here's a five goal round - not based on a game we actually played, just randomised goalscorers (weighted slightly toward the lowest handicapped players).

Goals were attributed to Richards, Lloyd, McLean, Libba and Bruce:

 

[dogwatch]

And here's a hypothetical 9-goal game (same method as above).

Goals were attributed to Vandermeer (2), Dale (2), Wallis, Macrae, Lloyd, Naughton and Dunkley.

Dogwatch still yet to score after 3 hypothetical games ...

 

[dogwatch]

Finally here's a hypothetical 14-goal game.

Goals were attributed to Dale (5!!) BSmith (2) Wallis, Bont, Gardner (yay), Lipinski, Bruce, McLean, Vandermeer



FYI the actual frequency of each of the selection approaches (combos) was as follows. It also shows the weighting given to each combo in the randomisation:

Combo	Weighting	Actual	Count
1+1+1+1+1	37.50%​	35.00%​	14
2+1+1+1	25.00%​	15.00%​	6
2+2+1	12.50%​	12.50%​	5
3+1+1	12.50%​	7.50%​	3
3+2	12.50%​	20.00%​	8

 

[BEaston]

I think it's a really good concept.

the one issue I see with it, which I think you already mentioned, is that it requires everyone to set their team every week. The current pieman has an element of set and forget which might suit some people. A pretty small issue though, certainly not a roadblock.

 

[dogwatch]

I think it's a really good concept.

the one issue I see with it, which I think you already mentioned, is that it requires everyone to set their team every week. The current pieman has an element of set and forget which might suit some people. A pretty small issue though, certainly not a roadblock.

Clearly you only play if you are willing to participate every week ... a bit like you have to do in footy tipping comps.

Maybe if someone forgets on any given week they get the same team they chose last week. Might have to think about subs if one or more of that team is out though (omitted, injured, etc).

So in theory you could play it as a set and forget but it probably wouldn’t go too well.