Koko The Fantasy Football Monkey

It’s that time of year when morons all over the country (like me) start to put their cheat sheets together for another season of fantasy football. I’ve resisted doing a lot of fantasy stuff on this site despite the obvious overlap between real stats and fantasy stats. I did some stuff last year on drafting strategies, but this year I’m going to break down and produce my own player rankings, but with a twist.

I’m going to make dumb rankings, in fact, as dumb as I could possibly make them. My fantasy projections are going to represent what someone would do without any knowledge of football at all. And you know what? I think there’s a good chance they’re going to end up no worse than any other ranking by the 'experts' or sophisticated ranking systems out there. I have a hunch that fantasy football is about 99% luck.

In a 10-team league, you’d ordinarily expect a 10% chance of winning. But what if you optimized your draft perfectly, read every fantasy site out there, and hawked the waiver wire every week. How high could you get your chances of winning your league? 13%? 15%? You’d need to play in literally dozens of seasons of fantasy football to really know if you’re any good or just lucky.

After last season I was struck by an analysis of the accuracy of some of the more prominent expert projections. Accuracy scores were listed for each system by position. I noticed how none of the systems were consistently near the top. One system could be #1 for QBs, but near the bottom for RBs and WRs. And no system was consistent from year to year either. If a system were any good, wouldn’t it work well in more than just one year or for more than one position? What this tells me is that no one really knows what they’re doing. They’re guessing like everyone else.

Here’s how I’m going to do my projections: I’ll take each component of a position’s fantasy score total and regress each component separately. Then make an estimate for each stat based on the regressions, add up the points, and that’s the final projection. For example, QB TD passes are less consistent from year to year than passing yards, so each component would have a different rate of regression. I’ll walk you through the QB rankings later in this post, and other positions will follow.

Blatant Rip-Off

Some of you may think that this is a blatant rip-off of baseball analyst Tom Tango’s ‘Marcel’ projections for MLB players. You’re completely right. Full credit hereby goes to him. Tango regresses hitter and pitcher stats over the last several seasons and adjusts for age. His system is completely unaware of trades, injuries, or ‘intangibles.’ It turns out Marcel is nearly as accurate as the proprietary high-power projection systems out there, some of which charge subscription fees.

These projection systems all seem to need names like CHONE or PECOTA for some reason. Marcel gets its name from the pet monkey on the sitcom Friends because the projections are what a monkey would project (a monkey that knows regression, I suppose.) But I was always more of a Seinfeld guy, so I’m leaning toward Koko:



QB Projections

I’ll walk through the QB projections. Nothing here is any more complicated than what anyone can do with an internet connection, Excel, and 30 minutes. I’m only looking at QBs who had over 200 pass attempts in a season. And after all, in fantasy we’re only interested in the top guys.

Baseball projection systems usually look back three or more years and weight recent years more heavily. But for football, I’m going to just go back one year for a couple reasons. First, using only one year is really simple and easy, and monkeys like simple and easy. Plus, in contrast to baseball, player performance is heavily dependent on the rest of the team. Teammates come and go all the time, either due to free agency, retirement, draft, or injury. It’s likely better to keep those changes to a minimum by only looking at the most similar year in terms of teammates. So I’m going to stick with one year. (And actually, going back more than one year provides almost no added predictive power based on the r-squared of the regression.)

I won’t make any age adjustments either. Going back just one year means that a player will now only be one year older than his baseline, so the effect should be minimal. I think the only glaring shortcoming with this method may be second-year starting QBs, who tend to make significant improvements. But in a way, a simple regression will adjust for some of the effect of age. Older, better players will tend to decline, regressing to the league average, while younger rookies, who often have rough first years, will tend regress up toward the league average.

For guys like Tom Brady or Carson Palmer, who missed an entire or nearly an entire season due to injury, I regress them twice--once for their estimated performance for the missing year and again for the current year projection. This might a little too pessimistic for a guy like Brady, but the method is logical and consistent.

Let’s start with passing yards. Normally, I don’t rely on total yardage when ranking players or teams, but that’s all that matters for fantasy. So I’ll look at total passing yards per game. Using data from 2001 through 2008, I plotted each qualifying QB’s Yds/G from a previous year with his Yds/G from the subsequent year (below). I’ll use the resulting trend line to estimate next year’s Yds/G based on last year’s.


Next, let’s look at TD passes per game. The slope is shallower, and the dispersion of the data is wider. This means TDs/G are not as consistent as Yds/G. A QB who last year led the league in fantasy points based on lots of TD passes will probably not be so hot this year.


Interceptions are even less consistent. Interceptions per game appear almost completely unrelated to those of the previous year. You might be able to get a better projection with a complex formula of interceptions per attempt and yards per attempt, but monkeys don’t want to think that hard. I’m tempted to just assign the average Ints/G to every QB. But I’ll project them according to the trend line, even if it’s almost certainly not statistically significant.


Fantasy points for QBs also come from sacks, fumbles, rushing yards and rushing TDs. The plots for year-to-year projections for each of those stats are below.






Adding up all the projections based on prior year per-game stats, we can get a reasonable fantasy point total for each QB. Top qualifying QBs play an average of 13.6 games the following year, so I’ll multiple the per game stats by 13.6 for a grand total. The table below is based on a scoring system of 6 points for every TD, 1 point for every 50 pass yards, 1 point for every 20 rush yards, -2 points for every turnover, and -1 point for every sack.

What about rookies or other players not ranked? I don’t know. I’ll just throw anyone else as tied for last. If you’re picking QBs that deep into your fantasy draft, maybe you need a smaller league.

Would I use these projections in my own league? Probably not. But these projections should serve as the bare minimum level of predictive power. If a system does any worse, it’s either really bad or really unlucky. After the season it will be fun to come back and see if these rankings were any good. At the very least, we learned something about the year-to-year consistency of a QB's performance.







































PlayerPass YdsTDsIntsSacksFumRushPts/GProj Pts
Drew Brees 2781.71.01.10.51.312.3167
Tony Romo 2511.71.01.60.63.811.5156
Peyton Manning 2421.60.91.20.42.311.2152
Kurt Warner 2581.60.91.60.61.211.0149
Jay Cutler 2511.50.91.20.511.110.8148
Philip Rivers 2321.61.01.60.55.410.6144
Aaron Rodgers 2381.61.22.00.511.410.2139
Tom Brady 2441.6

1.11.50.50.09.8

133
Donovan McNabb 2371.40.92.10.58.59.5130
Shaun Hill 2261.51.22.20.611.39.0123
Carson Palmer 2411.41.21.60.50.09.1123
Matt Schaub 2451.41.01.90.66.18.9121
Eli Manning 2121.40.91.80.51.88.7119
Matt Cassel 2271.41.22.40.514.58.6118
David Garrard 2231.30.92.30.517.08.6116
Derek Anderson 1981.30.81.40.65.68.4114
Kyle Orton 2111.41.11.90.53.88.4114
Chad Pennington 2211.30.92.00.54.38.2111
Sage Rosenfels 2061.41.21.40.50.08.1111
Ben Roethlisberger 2141.50.92.90.66.28.0109
Matt Ryan 2201.31.21.60.56.47.9108
Jake Delhomme 2151.31.21.70.52.37.6103
Jason Campbell 2121.10.82.10.513.97.5102
Joe Flacco 2051.31.22.00.610.17.4101
Trent Edwards 1981.20.91.60.67.87.4100
Matt Hasselbeck 2061.20.92.50.49.07.298
Gus Frerotte 2101.31.22.30.51.87.297
Kerry Collins 1961.21.11.30.53.77.297
JaMarcus Russell 1931.21.12.00.67.96.994
Marc Bulger 2001.10.92.70.53.45.880


  • Spread The Love
  • Digg This Post
  • Tweet This Post
  • Stumble This Post
  • Submit This Post To Delicious
  • Submit This Post To Reddit
  • Submit This Post To Mixx

31 Responses to “Koko The Fantasy Football Monkey”

  1. Anonymous says:

    I've been meaning to do this for a while, but I could never find the 30 minutes. Thanks!

  2. Anonymous says:

    Is the double regression for Brady screwing with the number totals? It seems that his ProjPts should be (9.9 * 13.6 =) 134.64, not 173.

    If in fact he has 173 projected points, shouldn't he be on the top of the list?

  3. Anonymous says:

    I appreciate the amount of time these take. You're surely speaking tongue-in-cheek when you claim that it took 30 minutes; the post alone would probably take double that amount.

    Apart from a few anomalies, most notably Romo's and Brady's projection, the list just reads like a 2008 year-end list. The complexity of the regression lines merely yielded what we knew about the QB's from last year's performance.

    And I suppose that is the whole point: like stock prediction gadgets that sell for tens of thousands of dollars, or mutual fund managers who garner hundreds of millions of dollars in new money based on one year's performance, we really can't know if we, as fantasy managers, are actualy that good.

    Now I have to tell my wife that I need to spend the next 25 years playing fantasy football to really see how I'm doing. Know any good divorce lawyers? ;)

  4. Brian Burke says:

    Yes, I actually do know a great divorce lawyer! Maybe he was too good. That's how I have all this time to do this stuff. I'm a single dad with custody and not much else to do after the kids go to bed.

    Yes, if you want to make pretty graphs and tables it will take much longer than 30 min. I actually did all the spreadsheets back in January.

    The Tom Brady number was a mistake. The 173 was for a single year's regression. I guess that's a good thing to point out. If you think he won't "miss a beat" or whatever, you should have him as your top QB. I'll fix the table in the post. Thanks for the catch.

  5. JB H says:

    Marcel is smarter than you think. Marcel acts pretty close to optimally given the amount of information it has. The age adjustment and the amount it weighs prior seasons are not guesses.

    I don't know a lot about the quality of other football projections, or the predictabilty of fantasy stats, but I would expect Koko to do much worse than Marcel

  6. Will says:

    Brian,

    Looks like your axis labels are probably switched on QB passing YPG. Also, are the two datapoints of QBs rushing for 50-60 ypg both Michael Vick? I don't see them in the table.

    As a pretty unsuccessful fantasy manager I definitely think it's almost all luck. You could probably predict each QB to have the league mean performance from last year and have reasonably accurate predictions, at least in terms of mean (not absolute) error.

  7. Anonymous says:

    When you did multi-year regressions, did you treat each stat as an independent point? Or did you use the mean of the last N years as a single point?

  8. Jason says:

    What would be really neat is if you could get 6 or 8 statheads to each develop their own "system" and then put them in a league together and see whose works the best. It'd only be a one-year deal, so it would vary pretty wildly due to luck, but it might still be neat to watch.

    (I tried something like this with Strat-O-Matic hockey one year, drafting six teams, each with a different focus: offense, defense, goalie play, "enforcers" (guys with lots of PIMs), and whatever the others were. I only played a few games, so it didn't get very far. Did something similar with Diamond Mind Baseball, drafting 10 teams around various stats, each team drafting the guy with the best of their chosen stat: OBP, OPS, Avg., SLG, SB, SB%, and so on. Simmed a season or two of that and it was interesting to see the results.)

  9. Brian Burke says:

    Yes, the two data points up around 60 yds/G are Vick. The table should say Year Y+1 on the vertical axis. Sorry.

    Don't get me wrong about Marcel. It's great, and it's partially why I did this. I would also expect Koko to do far worse than Marcel. But I would also expect any football projection system to do far worse than any baseball projection system for a number of reasons. The biggest three are: 162 games vs 16 games, dependence on teammate performance, and opponent strength. I doubt 'Koko' would do much worse than say, ESPN's projections, but I could be wrong. For example, RB projections are tougher than QBs. They depend far more on how many carries they'll get, which an expert might be able to project much better than a regression from the year prior. Rookie RBs can have impacts too.

  10. Brian Burke says:

    Also, if I understand the question correctly, for multi-year regressions, I used each player's year-long per-game average as a single data point. But I limited the cases to seasons with at least 200 carries.

  11. JB H says:

    I think it would be smart to regress the projections to the player's current team's performance the year before. That would help solve the player dependency problem, and would allow you to do more reasonable projections for players without much or any playing time.

    I don't think it would be against the spirit of Marcel's "dumbness", it's just making a concession to the differences between the sports.

  12. Brian Burke says:

    That's a great idea!

  13. Zach says:

    Brian, you said you only used one year of data to control for personnel and teammates moving each year. Would it help if you use however many years each player's been on the same team? (

    For instance, you'd look at Tom Brady's career numbers, weighted by season, but you'd only look at Drew Brees' numbers since he arrived in New Orleans.)

  14. bruddog says:

    I've played fantasy football for many years and my conclusions are the following:

    1. There is some "skill" to fantasy football. That " skill" comes int he form of being on top of your leagues scoring format, on top of player moves, changes in potential starting roles. especially early on in the season.

    2. The "theoretical" benefit gained from the skills above drops exponentially from the time put in. Spending just a minimal amount of time produces most of the benefit.

    3. Like you said these sites try to include as many variables as possible but in the end they are still guesses. Maybe I should go into the fantasy football business. Make up some proprietary formula with zero visibility as to how it works and get people to subcribe. Like these guys

    http://www.fantasyfootballchamps.com/index.cfm?page=ffcpi

    4. Nothing de-rails your team like a random unpredictable injury.

  15. Anonymous says:

    could you post your spreadsheet?

  16. Brian Burke says:

    Zach-That's a good thought, but the point is that the surrounding cast changes enough, even without a change in the entire team, to negate the added accuracy of using earlier years. Lineman, receivers, coordinators, etc. Even the defense has a big impact on a QB's fantasy production--generating turnovers and influencing field position.

    Yes, I'll put up a spreadsheet when I get around to doing the other positions.

  17. Anonymous says:

    If its 99% luck, when it comes to fantasy football I must be the luckiest man alive. I've had two teams a year since 03. Here is how I finished: 1st, 1st, 1st, 2nd, 1st, 1st, 4th, 3rd, 1st, 1st, 1st, 5th.

    There is certainly luck involved, possibly a lot. But I think one of the reasons for my success is just as much being attentative to free agents as it is good drafting.

  18. ben says:

    What about numbers of games played. Shouldn't they be regressed toward the mean also instead of using an average? On an elite player (i.e. the only ones we care about for fantasy football) starts are inverse related to injuries which is one of those luck factors that we are regressing for.

  19. Brian Burke says:

    Ben-Good question. You'd think there would be a correlation, but there wasn't. It was a blob, a lot like the interception graph. At least, with my simple methodology that's how it was. For the most part, these guys either get injured or they don't. And even if you can project if they're injured (which I don't think is possible), you can't project when.

  20. Chase Stuart says:

    Great post as usual, Brian. If you click on my name, I wrote a post that came to the same conclusion you did regarding interception rates.

    The most interesting thing to me was that the slope was higher for sacks than passing TDs. Do you have the R^2 for those two statistics? I like including sacks in my QB rating formulas, but some people argue that sacks are more a factor of the OL than the QB. While this doesn't necessarily say otherwise, it does indicate that QB sacks are a repeatable "skill", at least as much as QB TDs are. And it certainly appears that almost all INTs are due to randomness (although they deserve to be a part of any QB rating formula that is retrodictive in nature).

  21. John says:

    All you need to do is take a couple of sites' projected stats and a big pool of projected rankings, split the players up by position, regress the stats and (logistically) the rankings, toss out the stats with high p-values (like apparently INTs), and you have as good a fantasy football projection system as you can get. You'd need at least 3 years worth of projected rankings and stats for a decent sample size, but with a decent mastery of Excel it could be done.

  22. Withnail says:

    Very interesting stuff. When you have another 30 minutes(!) how about doing the same analysis as at the start of the 08/09 season? Then compare Koko's predictions with those of some of the experts, and see who was most accurate.

  23. darrint says:

    Anyone care to post a tutorial covering the following: 1. Where all these apparently free for the downloading stats live? 2. What a sample Excel session might look like?

    I've been trying to learn to handle stats and draft day prep sounds like a good place to practice.

  24. Koko Damonkey says:

    Interesting article. I'm one of the many who enjoy deluding myself into thinking that preseason statistical analysis really helps.

    Although I would agree that there is quite a bit of luck in fantasy football, I think it would be extremely hard to measure how good pre-draft rankings are in terms of fantasy team wins/losses. The reason, IMO, is that the accuracy of one's draft rankings accounts for much less than 50% of how well one's team does over the season. Draft strategy (which position to pick in each round), drop/adds, trades, and bench/start decisions play a huge role, too. More than enough to overshadow pre-season rankings.

    I think the unusually large number of comments here shows that your readers would love for statistical analysis to have more of a fantasy impact, though.

  25. Anonymous says:

    darrint,

    Try this link for free stats:
    http://www.pro-football-reference.com/years/2008/passing.htm

    The specific link points to 2008 passing stats, but they have pretty much everything you'd want for standard fantasy analysis. And it's all free.

  26. Dave says:

    Speaking for myself, I actually prefer that most of the articles don't have much to do with fantasy.

    THe cost/benfit analysis for most fantasy leagues means the extra amount of time/research you put into "possibly" improving your chances is hardly worth the benefit. The more people in your league that do the same thing the more inefficient your time becomes...

  27. Nate the Great says:

    I liked the article above! I do quite a bit of excel sheets to help my predraft cheatsheet. From what I've found in the countless years I've played, its almost all luck.
    The first year player has won 3 out of the last 6 times in one league I'm in.

    The best strategy I have come up with is this:
    Since my league has 2 rb slots and a flex Wr/Rb, I am going to spend over half of my roster on the "sleeper"ish Rbs throughout the draft.

    I won when I took a risk with westbrook in the first but wasted 6 picks on the likes of Rookie Adrian Peterson, Edge James, J Addai and some busts. What happened? 3 of those 6 broke out and played some football.

    All in All, You Win in the LUCK of your PICKS from round 5-9. Peterson can only score so many points. Brady can only throw so many TDS.

  28. Chris says:

    I agree that simply knowing the rankings of the players, and having a good grasp of them, is only a small part of a successful draft.

    Draft strategy plays the largest role when determining how to spend your picks by looking at the draft as a whole. Positional depth plays a part, as does choosing a mixture of high-risk/high-reward players and players who have a better chance of coming close to their projections. Another strategy is handcuffing running backs who share time.

    I've noticed that no team in our league has a significantly higher rate of picking individual players successfully, but that some teams have a better plan when going into the draft.

    This does result in some managers having consistently better drafts than others, mainly because they budget for some picks to fail going into it and they have a backup plan.

  29. Alchemist says:

    Great stuff, as usual. I know from reading your previous posts that you're familiar with game theory and I'd love to see an analysis of how game theory plays into fantasy draft strategies. I've done pretty well in my fantasy leagues - certainly better than a "random walk" would suggest - and one thing I usually key on is when to draft certain positions based upon A) the relative value of each position, given that league's scoring system, and B) the number of "impact" players available at each position.

  30. Brian Burke says:

    Here are some articles. In particular, check out the "Game Theory and Fantasy Draft Theory" article. That's how I draft.

  31. Alchemist says:

    Thanks - I guess I should have checked your archives a bit more thoroughly! Your VONA analysis is actually what I was trying to describe in my post above. In fact, my drafts often look like your sample did, with me often taking a WR or TE higher than others would normally expect.

Leave a Reply