Discussion Tip a Top 30: Using last year's points to predict the future

[Rowsus]

Definitely one of the better analyses published on this site.
Without trying to pinch KLo30's thunder, I would like to add a bit of my own analysis, based entirely on the numbers KLo30 has posted above.
While the figues back up, that we should expect around 15 players from this 2014's top 30 to drop out of the 2015 top 30, it is usually not "new" players that replace them. There are are 149 players listed in the 5 years above (it excludes McDonald, as he only payed one game in his top 30 season). There are 72 individual players that make up the 149 top 30 positions.
4 players appear in all 5 years
5 players appear in 4 out of 5 years
13 players appear in 3 out of 5 years
20 players appear in 2 out of 5 years
30 players appear in 1 out of 5 years
So only 30 of the 149 top 30 players appear once, that's around 20% of the 149 spots are "one timers".
The other scary thing, as far as people trying to narrow down their selections for 2015 is, only 8 of the 72 players on these lists have retired! Of the 42 players that appear on these lists more than once, only 2 have retired!!!
Let's re-arrange KLo30's tables to show the 5 Groups.

Taking a quick glance at these first 3 tables, the first things that hit me are:
In the 4 out of 5 table, Boyd, and possibly Swan are faltering probably due to age/wear and tear.
In 3 out of 5 table, 2 are on a 3 year streak. 5 had one year out of the top 30, then bounced back. Barlow had 2 years out (both injury affected) then bounced back. Griffen is at a cross roads. Goddard is looking to become the first player to make it back into the top 30 after dropping out twice.
In the 5 out of 5 table we have 3 Super Premiums, and Watson, who is probably more of a 106-110 player, than a genuine 110+ player.

Of the 11 players that have missed making the top 30 at least 2 of the last 3 seasons, only Gibbs is not near the end of his career!
All the players above Montagna have a strong possibilty to make it again this season.
Just for completion's sake, here are the players that made the lists only once.

There are all in all 19 players that have dropped out of the top 30, then made it back in in a later season. No one has managed to do this twice. 12 of these 19 were out for one year, 6 were out for 2 years, and one was out for 3 years. 5 or more of these 19 could claim it was injury that kept them out.

I will continue this analysis when I get up in the morning (it is 00.22am here now), The second part of the analysis will look at the age, and record of the players that managed to regain their top 30 status.

 

[Philzsay]

Brilliant thread all round!

To state the fairly obvious, I note that there is only two current SC backmen out of the 72: Andrew Walker and Brian Lake.

 

[Jurn Stern]

Brilliant thread all round!

To state the fairly obvious, I note that there is only two current SC backmen out of the 72: Andrew Walker and Brian Lake.

Don't forget trusty old Hodge!

 

[Philzsay]

Don't forget trusty old Hodge!

Oops!

And to think I browsed through the names twice to make sure I had them all!

 

[KLo30]

Defenders find it difficult to average 100+ in any one season, let alone make and stay in the top 30 from one season to the next. For the period above, here is the frequency table of 100+ season averages by position.

In 2014 5 classified defenders had a season average of 100+, 1 in 2013, 4 in 2012, 6 in 2011 and 4 in 2010.

To no-one’s surprise Brendon Goddard featured in all bar one (2014) year, when he was classified only as a midfield player. Below is a list of all defenders scoring 100+ season averages from 2010-2014.

Only four players, Goddard, Brett Deledio, and the two Heaths (Scotland and Shaw) appear on more than one occasion.

Obviously, Supercoach has a habit of reclassifying players from season to season. However this doesn’t appear to increase a defenders chance of ‘backing up’ in their new position with only Goddard, Deledio and Hodge going 100+ the following season.

Only Malceski stays as a defender this year, albeit at a new club, with D.Swallow, McVeigh and Hanley all moving to midfield only status. It’s not out of the realms that none of these players average 100+ in 2015.

 

[Drew]

Wow this thread just keeps getting better and better! Incredible work guys!

 

[Darkie]

Very interesting analysis KLo. I think this ties in nicely to the way I've been thinking about my team for 2015 - but it's nice to have some additional numbers that might support that thinking!

I've seen some teams (admittedly, more commonly on other sites) that seem to have been based around getting as many of the class of 2014 in as possible. If the game was simply about selecting the players likely to score the most points, this could be quite a good approach. The issue, of course, is that there is a budget constraint - so it's arguably much more about identifying value than it is about identifying the top scorers alone. The class of 2014 are by definition the highest priced players, so value may be much harder to find on that list than high scoring!

An approach I've used to identify potential value is to look back at 2013's top scorers in each position (the SCS Team Planner works well for this, as players are ordered by 2014 starting prices, which are based on 2013 averages) and thought about the pros and cons of these players as we head into 2015. This also helps provide a little more perspective on certain players.

As a result, I've ended up with a preliminary team that includes only three of the class of 2014- Ablett, Pendlebury and Selwood. In my view, these are the three safest bets as ultra premiums as well - they're repeatedly in KLo's tables with top 5 averages, and are all pretty durable.

In contrast, I have as many as seven of the class of 2013, because I've tried to pick a number of fallen premium types that I think represent value. The additional four are Danger, Swan and Goddard (both in the forward line, where top 30 players are harder to find) and Steven (a 50-50 for me, and chosen partly for structural reasons).

As it turns out, of my 12-15 possible keepers (15 being very generous/lucky) only two - Ablett and Selwood - averaged more in 2014 than 2013. I suspect this would be pretty unusual - but I think it results from a fairly aggressive focus on value. I've tried to do this without selecting players that are marginal in their position (you don't really want four M8s, even if they are cheap!).

This approach was also informed by the analysis in this thread - which is well worth a read or re-read:

http://www.supercoachscores.com/thr...with-an-Analysis-of-last-3-SuperCoach-Winner)

On my reading, past winners look to have been (1) very value-focused and (2) potentially a lot less pure GnR than I might have guessed. They look to have chosen 8 or 9 top 30 players (see Rowsus' table at post 29 of the link above, which I read as being comparable to KLo's work) in their starting sides though ... making my three look pretty skinny.

I think there are a couple of reasons for this. The class of 2014 is mid dominated, and you can't really pick more than perhaps 4 or 5 mid keepers in your starting side. Of the non pure mids that made the top 30, I see more value in the ruck (NicNat and Leuy vs Jacobs and Mummy, eg) and fwd lines (Swan vs Gray, Harvey, eg) than what the class of 2014 offers ... so only three it is.

Are others in a similar position as we head into 2015? Or is three very light? I don't see many of the class of 2014 as that attractive on a value basis - which I think is very much part of the point of KLo's work! - but I'm interested in others' thoughts.

I also thought Rowsus' analysis of prior winners was well worth highlighting

 

[Ben's Beasts]

Are others in a similar position as we head into 2015? Or is three very light? I don't see many of the class of 2014 as that attractive on a value basis - which I think is very much part of the point of KLo's work! - but I'm interested in others' thoughts.

I have just the 4 from the class of 2014 in my current side so I don't think that 3 is too light at all.

There are quite a few great value picks on all lines available to us this year.

 

[chels]

Exactly the sort of analytical thinking I have been looking for (and seldom finding). I do love playing with data. My congratulations to the Klo and Rowsus and the other contributors - I feel honoured to be amongst you.

 

[IDIG]

Welcome to the site chels. Another Essendon supporter and analytical thinker to the site

 

[freowho]

Awesome thread. I've got 3 at the moment. Ablett, Rockliff and Murphy. I've just been browsing a few injured players and it looks to me like the 1st year back is good for a bit of conditioning and it is the second or third year back where they can return to previous form. I think Murphy is as good a chance of 110+ as Dangerfield.

 

[Courtesans]

Gez, what a thread this is. Congrats fellas.

 

[OzRulesFootball]

Awesome thread. I've got 3 at the moment. Ablett, Rockliff and Murphy. I've just been browsing a few injured players and it looks to me like the 1st year back is good for a bit of conditioning and it is the second or third year back where they can return to previous form. I think Murphy is as good a chance of 110+ as Dangerfield.

ssshhh

 

[Rowsus]

So to continue with my analysis from post #21 above.

Let's look at the players that have at some stage dropped out of the top 30, then made it back in. Since 2010 there have been 19. The table is listed in order of their ages at July 1st, the year they made it back into the top 30. Youngest player first.

Apart from injury as an excuse, it appears you need one of two things, before you back a player to rebound back into the top 30. They need to be younger than around 25 Years 6 Months, or they need at least 4 previous seasons in the top 30. We can see by the table (and common-sense), that the older a player gets, the harder it is for him to maintain his grip on the top 30, or to rebound from a bad season back into it.
So how does that look for the players that dropped out of the top 30, going from seasons 2013 to 2014?

Looking at this table, we don't have too many "younger players", and the 4 at the top are hovering around the cut off mark (they are all born withing 14 days of each other!). On a pure "blind" test (ie remove the names, and just look at their stats) Dangerfield, Cotchin then Swan look to be the most likely to bounce back into the top 30. Griffen, Boak along with Goddard and Bartel look like some chance to be just in, or just outside the top 30 in 2015. Griffen is problematical, being at a new Club, and Goddard/Bartel because of their 8 and 9 year records respectively.
You wouldn't want to be backing in Minson, Walker or Steven to do it, on what we have seen so far. K Jack has been known to be up and down, but I wouldn't back him to bounce back too high, if he does manage to make it back to the top 30 at all. The others seem to be too old to take a risk on and/or play for poor teams, which adds to the risk involved.

Once again, great thread KLo30. I hope you don't feel I'm trying to steal your thunder.

 

[Maso]

wow ! just wow

 

[Hondo]

Sheez this is some thread !

I've skimmed it once and been blown away. I'll need multiple goes at it to absorb it all.

So what does this say about picking 5 premium mids over $600k? Blinkered into playing last year's SC season a year too late? The stats say that strategy is fraught with danger I think?

 

[Manikato1]

Whats your team looking like then Manikato? Or is it a secret? Would be interested to see what it looks like with only one top 30 scorer in it from 2014

Hi Siwel

I don't post teams as I realised a long time ago I am no where near good enough a coach to make suggestions to others and I would rather live or die with my one decisions so I just stay completely out of it. But given most of the players that were listed to drop were mids or rucks and you asked I can tell you the current mid/ruck line up looks like this.

Gablett, Pendles, Prestia, Wines, A Swallow, then rookies depending on NAB Cup such as Heeney, Gore, Boekhorst etc.

Rucks Nicnat, Berger

Backline still work in progress but have $192k cash and I prefer to keep around 100k.

 

[PC]

Rowsus that is an A+ grade post!!

Brilliant thread as well - thumbs up all round.

 

[KLo30]

Once again, great thread Erich. I hope you don't feel I'm trying to steal your thunder.

Not at all Rowsus, I value your contributions to this thread and the depth they provide.

Cheers KLo

 

[Darkie]

Thanks Ben, Freo - I appreciate the feedback, especially from the two of you. For what it's worth, I think Murphy is a bit under-rated as an SC scorer. He's been top 30 for 4 of the last 5 years, and very likely in 2009 as well (113 average). I prefer Danger in part because I think he has more potential upside above 110. He can also be quite handy as a VC option, because his ceiling is higher (albeit that Carlton play more early games this year).

Hondo - yes, that's largely the way I interpret it. Certainly if you were choosing five $600k mids, it suggests you should be careful and selective in doing so - because, on average, the players on these lists are over-priced by around 9 points per game (or around $47k). Once you've chosen your captaincy options, there's a pretty good argument that you should start looking for (1) select high priced players that might be under-priced (due to being subbed or carrying injuries, eg), or (2) somewhat cheaper options (below $600k) - and/or simply use more of your mid spots for rookies.

As a comparison, I did some back of the envelope numbers on how fallen premium type players performed - an approach similar to the one I mentioned in post 27 above (going shopping for 2015 players in the class of 2013, not 2014) - but obviously more formulaic. For these purposes, I focused on those that had had a larger than average fall in their scores, based purely on KLo's tables.

Looking at the class of 2012, eg, these players ended up with averages 13 points worse in 2013 than 2012 (see the bottom of the second column from the right). Beams' average declined 21 points in 2013, so he makes the shortlist, having fallen further than average (-21 vs -13).

I then checked how those players performed in the following year - in Beams' case, 2014. [This info was sourced separately - KLo's tables are obviously set up to look at something different.] Last year Beams increased his average from 102 to 116, so that was obviously a win, and that would have been recorded as a +14. [He also happened to receive a discount heading into 2014, but that's a separate issue.]

On average, players fitting these criteria outperformed their prior year averages by 6 points per game. That means they were around $31k under-priced (ignoring discounts).

Obviously if you add a bit of real world knowledge/common sense, you should be able to do better than that. The two additional overlays I tried to apply in selecting my own side were players with a proven record of good scoring, and players that weren't too late in their careers. As it happens, these are essentially the same two factors Rowsus identified above ... although there is a lot more analysis to back it up now!

To put all of this in perspective, choosing a random class of 2014 player for your 2015 side (not that anyone chooses randomly!) means you're likely paying around $47k too much, on average. Choosing one of the fallen premiums from 2013, instead, means you're buying a player that is likely under-priced by around $31k on average - a $78k difference. If a typical team has around six of these players, that's a difference of $468k in total - nearly 5% of your budget. It's also more than enough to upgrade someone like Petracca to Fyfe ... although whether that's the best use of those funds is something to consider