Game Theory and Fantasy Sports

Krump wrote:The more I read the more fascinating this gets. I would not be at all surprised to see someone from this thread win at least one of the thursday games

Agreed. So many different takes and opinions on this. After a few rounds we may get a clearer idea of what ways to go in order to give the best chance of winning it all. Would love if someone from on here was able to nab a win somewhere along the line

Beast From The Big East wrote:

It doesn't but surely simple logic would dictate that if someone scores a 55 and someone else scores a 55 then it shouldn't matter who you have as they are equal. You don't get rewarded more points because you happen to have a low ownership team. You should still need to factor in the scoring potential both high end and low end when comparing your POD choice with the more obvious option for the rest of the comp.

The difference to me comparing say a Broncos side with a Roosters is that some teams and in some games there are clear obvious picks who are far and away better scorers. The Roosters for example would be a good team for this type of exercise because they have a multitude of options both starting and on the bench who can all score highly, yet none are consistenly dominant. The general public will probably look at guys like JWH, Cordner and Guerra, yet Moa, Napa, Tauk, Liu are all options who can go big on their day while the big names can struggle.

Shanbon or Milchy as their more qualified than me but my thinking is the Broncos would be better because of the standout option in Parker. Any game that he's involved in he'll be captained by huge numbers of people giving you an instant advantage over most players in that game? Of course that's assuming you find the week he gets outscored and the player who outscores him.

Krump wrote:
Shanbon or Milchy as their more qualified than me but my thinking is the Broncos would be better because of the standout option in Parker. Any game that he's involved in he'll be captained by huge numbers of people giving you an instant advantage over most players in that game? Of course that's assuming you find the week he gets outscored and the player who outscores him.

This is the key problem for me. I see the reasoning with the idea of going with someone other than Parker and agree that whenever he is playing there will be many that will captain him. Broncos would be a good option if you get lucky with a dud Parker game and correctly pick who outscores him. But to my mind we need to differentiate between stand out option and the best option. Week to week Parker is the standout option in that forward pack because he scores much higher than everyone else and he barely ever scores poorly and even then it is still a very good score and not many can beat it.

Compare that to a team like the Roosters. They don't have a standout option. Rather instead they have players you could call the best option. Guerra/JWH if you think it will be a tough, grind it out game, Cordner if you think Roosters will get some solid attacking chances. But the level of that player is not that much greater than the others so you have a greater likelihood of your POD outscoring them.

It doesn't matter what teams are playing the same logic applies in these one off match days. When you break it down the bare numbers i.e. Tauk out scores Cordner 60% of the time (just made up example here) the play comes down to your estimates of ownership.

I have spreadsheets set up to give me floor/ceiling/consistency ratings that I'll always use for my projections and hopefully this means for me I can work out which games Tauk is more likely to outscore Cordner. So the basic season numbers might say 60% but this week I might project that to be only 40% then I need to try work out what % ownership each player has and do my little calculation

Krump wrote:The more I read the moreShanbon makes sense. If you want a good ranking every year play it safe. To win you need to take risks and there isn't really a difference between rd1 and Rd 19 except that taking a chance in rd1 doesn't automatically cost you a trade

This is absolutely spot on. We aren't trying to finish in the top 100, we are trying to finish number 1. Best chance at that is getting solid PODs with a strong base. The likes of Fifita, Hunt, Gallen

Shanbon wrote:On the Parker v Gillett if ownership changes to 50% Parker 15% Gillette then Parker becomes the better play. Boils down to Gillette needs to be owned by less the 1/4 of the amount of Parker owners.

Milchy can you throw out the number of games Norman out scores Hunt and the number of games he outscores Milf please

Norman outscored Hunt only twice (bit unlucky, he scored 57+ (Hunt's average) three times, but didn't beat Hunt on any of them as Hunt scored 70+ in each of those weeks. Norman also didn't play round 20 when Hunt pumped out a lowly 15)

Hunt is generally quite solid though, only provided 2 crap scores. A 17 in the rain against the dragons, and a 15 when they got pumped by Manly late in the season.


Norman beat Milford 9 times (plus 1 draw) out of 20 weeks they both played.
Milford is quite erratic, a lot of sub 30 scores (although only 1 in the latter half of the season)

Sounds like you are well and truly on top of it for the season Shanbon. Hopefully we see your name up there a few times as a Thursday Night comp winner

One thing we haven't really looked at here is scoring trends against certain opponents, and effect of victory/loss on scoring.

For a season long comp, weekly opponents are only a minor factor in whetehr to get a player or not, as theoretically you'll have them for the whole year anyway.
I've heard people thinking of Slater for the start of the comp as Melbourne have an easy draw - so that sort of thinking certainly has merit, but its not the deciding factor.


For daily fantasy though, the ground condition and opponent has a large impact on scoring.
If one team is going to win by 50, you damn well want their kicker on your team, and you probably don't need their big tackler.

I don't think it makes sense looking for a POD half over Thruston in a game you expect the cowboys to win comfortably, for example.
Saying that, the ownership of every other half in such a game could be so low that it might be worth taking a punt on Thurston having an off night.

Milchcow wrote:
One thing we haven't really looked at here is scoring trends against certain opponents, and effect of victory/loss on scoring.

For a season long comp, weekly opponents are only a minor factor in whetehr to get a player or not, as theoretically you'll have them for the whole year anyway.
I've heard people thinking of Slater for the start of the comp as Melbourne have an easy draw - so that sort of thinking certainly has merit, but its not the deciding factor.


For daily fantasy though, the ground condition and opponent has a large impact on scoring.
If one team is going to win by 50, you damn well want their kicker on your team, and you probably don't need their big tackler.

I don't think it makes sense looking for a POD half over Thruston in a game you expect the cowboys to win comfortably, for example.
Saying that, the ownership of every other half in such a game could be so low that it might be worth taking a punt on Thurston having an off night.

Exactly. Going for the PODs is what will allow you to have a shot at winning a week but in my opinion it shouldn't be at the expense of a clear points beast. A comfortable cowboys game means JT will likely score highly, or at the very least have a high floor as he should get a fair amount of simple halves points if the cowboys do well.

A hard tough game between another top 4 contender meanwhile is the perfect game to bet against JT as he could only struggle through to a 35 yet many participants will pick JT and not even think about the matchup so there is a better chance of picking a half that outscores him.

Milchcow wrote:
One thing we haven't really looked at here is scoring trends against certain opponents, and effect of victory/loss on scoring.

For a season long comp, weekly opponents are only a minor factor in whetehr to get a player or not, as theoretically you'll have them for the whole year anyway.
I've heard people thinking of Slater for the start of the comp as Melbourne have an easy draw - so that sort of thinking certainly has merit, but its not the deciding factor.


For daily fantasy though, the ground condition and opponent has a large impact on scoring.
If one team is going to win by 50, you damn well want their kicker on your team, and you probably don't need their big tackler.

I don't think it makes sense looking for a POD half over Thruston in a game you expect the cowboys to win comfortably, for example.
Saying that, the ownership of every other half in such a game could be so low that it might be worth taking a punt on Thurston having an off night.

In a situation like that you would be better off taking Morgan over Thurston, although he only outscored him 3 times in the 17 games they played together.

multiple.scoregasms wrote:

In a situation like that you would be better off taking Morgan over Thurston, although he only outscored him 3 times in the 17 games they played together.

But how did Morgan go compared to Thurston in the games Cowboys scored 30+ points in?

I can't check that without looking at games individually, but I suspect it would be rare.

Beast From The Big East wrote:Sounds like you are well and truly on top of it for the season Shanbon. Hopefully we see your name up there a few times as a Thursday Night comp winner

Doubtful all my research and fancy spreadsheet numbers are based around moneyball DFS scoring, hopefully I make even more through that. When I get home I'll see if I can dig up some interesting stats

In my breakdowns I'm able to look at how a player scored in games with under or over 37 total points I was thinking of adding in a stat for his teams under overs.

I can also see avg in certain weather, on certain grounds, against different opponents ect.

Shanbon wrote:In my breakdowns I'm able to look at how a player scored in games with under or over 37 total points I was thinking of adding in a stat for his teams under overs.

I can also see avg in certain weather, on certain grounds, against different opponents ect.

  Giving me a hard on don't forget sharing is caring .
There must be a way to rip ownership numbers, that's really the only thing your missing.

Ripping ownership numbers would be cheating in a paid moneyball comp and I'd assume as its a gambling website they would have top notch security.

The legal way to do it would be manually looking at previous weeks top 100 team break down and seeing if you can find a pattern to help your estimates

A few little stats for from Moneyball last season.

In GPPs i played in the winning score avg was 380 from 9 players (42.2 per player)

All players avg was 26.2

Paul Gallen played 18 games at lock Corey Parker 24, Gallen was a top 5 scoring lock each week 78% of the time while Parker was 67% of the time.

all 5/8's averaged 26.05, Milford scored less then 17 points 1 out of every 5 games but scored over 39 points 2 out of every 5 games

The game theory article is quite interesting. I play AFL SC with a few blokes who are avid poker players who talk about things mentioned in that article. Also how we make decisions rashly/emotionally (ie making impulse trades) that can affect your fantasy ranking. Some good points also have been brought up in the posts thru this thread. Of course having a feel for the NRL teams is exceedingly important.

Firstly in regards to the sub-optimal tactics in the article, a few people have mentioned in this thread that there is a difference between %ownership for everyone and %ownership of the top 10k or so. The latter is the more important data (if you're outside of 10k by say rd 5 you might be too far back anyway). I know some people in AFL SC scrape the data off the website for top 100/1k/5k/10k ownership - and this helps me pick guys who are hurting me for not having them, as well as picking POD's to make ground.

On that last point saying that someone (ie Parker/Smith) doesn't give you a leg-up on the competition I don't really agree with, because having them also means that they're not hurting you. If you can find someone who 100k less who can replicate >60ppg then why not have this person + Parker/Smith? If you can maximise value in your team then fair enough, but these guys have over a sustained period produced great scoring. Let's not forget these guys are the best capt options in DT.

Also, and this was brought up, scoring can be influenced purely on the opponent. A few years ago the dragons had a tough defence, so attacking players suffered against them. Halves generally score better against weaker opponents (same goes for backs). There'll be teams who give up fantasy points, and those who are stingy. So if you see someone starting the season with a good schedule then that might edge out a 50-50 call (excluding keepers - this would require comparison of a season long schedule).

Having said this I think the byes/origin plays a bigger role than any of the above. So many points on offer through good bye management. This is in my view one of the downsides of NRL fantasy.

some good points Superbucks.

Do you play daily fantasy?

Shanbon wrote:some good points Superbucks.

Do you play daily fantasy?

No I don't, I've had a look and it seems pretty cool, just not sure how much time I would have if I played that lol

Those wanting an easy-to-use projections generator tool connected to a DFS lineup optimiser, I have just made one of those and launched it on FanLeague. I believe it's the first in Australia (apart from FanFooty which I launched for AFL a day earlier!).

http://www.fanleague.com.au/premium/createweekly.php

You can use historical data and then set your own "nudges" based on how you think each player will over- or underperform, then just one click and it calculates the best team that will fit under the cap. I will add DFS providers like Moneyball and Top8 once I get a list of their Round 1 player prices and caps.

I will be rolling out a bunch of features on the site over the next month that should help with a lot of things mentioned in this thread. But not all! Sounds like Milchy has even more complexity in his spreadies, with data I don't have.