When (and why) to Ignore Projections

Fantasy baseball is an inherently analytical game – you are trying to use a scarce resource (either dollars or draft picks) in a wiser manner than your competitors. And so it is no surprise that fantasy players have become big fans of the projection systems from ZiPS to Steamer to PECOTA and beyond. Just last week, Justin Vibber shared his ottoneu Surplus Calculator which uses Steamer projections to find expected values for MLB players.

In the ottoneu Slack community, the surplus calculator and other similar tools have become the primary manner of determining player value – if Steamer says a player will produce $30, that is the baseline. And this creates an opportunity to profit by ignoring projections.

Building a winning team requires you to identify opportunities to add value that your opposition doesn’t. Back in the early days of ottoneu (like, a decade ago), I was pulling projections from ZiPS on Baseball Think Factory and creating fantasy values. Most of my leaguemates were relying on standard rankings on fantasy sites, and the result was a competitive advantage for me – I valued different players than they did and targeted those players at prices below what I thought they were worth.

Today, that advantage is gone. Fantasy values are published all over, for lots of formats, including ottoneu. So if the market inefficiency a decade ago was players the projections loved but rankers did not, what is the inefficiency today?

I think it’s players the projection systems are likely to miss on. The projection systems are basically algorithms. They only know what the model accounts for and that is often limited. For example, going into 2015, Steamer and ZiPS were well aware that Jason Kipnis had underperformed in 2014. They also knew he had missed time due to injury. They did not know, necessarily, that he had played extensively through that injury and that his deflated numbers might not reflect what a healthy Kipnis can do.

If you play in a league where the projection systems are taken as gospel, Kipnis was a great opportunity for profit last year, and I ended up owning him in multiple leagues.

I’ve put together my 2016 values based on Steamer and see some examples of players whose projections I won’t take at face value:

Joey Votto – Steamer projects Votto at .283/.418/.479 with 22 HR, 87 R, and 74 RBI.* Last year, he went .314/.459/.541 with 29/95/80. In 2014, he was hurt and managed only 272 PA with a .255/.390/.409 6/32/23, but in 2013 he was again better than Steamer expects this year – .305/.435/.491 24/101/73. Steamer’s projections peg him at $21 in 5×5, $36 in 4×4 and $40 in FanGraphs Points. A hybrid of his 2013 and 2015 lines (say .310/.447/.516 26/98/76), plugged into the rest of the Steamer projections, comes out at $30/$49/$73. (That’s a bit misleading, because Steamer regresses everyone closer to the mean, so other players will also improve, but you get the idea). That awful 2014 has a real impact on the projections, as does typical age related decline, but I feel confident the injury is behind Votto. Yes, he could get hurt again, yes he could regress, but if my competition pegs him at those Steamer prices, I will gladly pay him a bit more, with a clear path towards significant profit.

Robinson Cano – Cano’s first half struggles were well documented, and he ended up being trade (or cut) bait in many leagues. But his vintage second half wasn’t quite as highly publicized. The first half woes were part bad batted ball luck (.290 BABIP for a guy who regularly posts much higher numbers), part intestinal illness, part emotional struggles related to the loss of his grandfather. It was also part aging and part home park that doesn’t help offense. But the Steamer projections are more in line with his overall 2015 line than his 2013, 2014, and 2nd half of 2015 lines. It takes a small leap of faith, but if you are willing to assume away the first half – even partially – another season with a .300+ average, .370+ OBP and high-.400’s SLG isn’t that hard to see. If you competition relies on the projections, you can turn a tidy profit by “overpaying” for Cano.

Carlos Santana – Santana has been on a three year slide – his triple slash rates all peaked in 2013 and have dropped the last two years (ok, his average stayed steady at .231 the last two years). Steamer sees a small bounce back, but the bounce back sits in between the two most recent (and worst) years, but with a lot less power than he showed in 2014 (when he hit 27 HR). However, Santana revealed after the year that he was dealing with a back issue all season. If that sapped his bat speed (a big drop in HR/FB% and HH% would vouch for the drop in pop) and if he is healthy, that return to form could look more like 2013 or 2014 than 2015. Steamer calls Santana a solid util option (14th best 1B) in FanGraphs Points, but I think he can sneak up a couple spots and become a low-end starting 1B (or a top-tier util).

You can’t do this for every player. First of all, it is too much work. Second of all, the projection systems are probably smarter than you (they are definitely smarter than me) so you have to pick your spots. It’s not enough to think a guy might be better than his projection; you have to think it is extremely likely before you gamble heavily on it.

Votto, Cano and Santana are all players I will be targeting (though Santana maybe only in OBP leagues). Who would you target as likely to outpace their projections?

*Note that I set these up last week so some Steamer projections have been updated on the site and are inaccurate in my spreadsheet – I am citing my spreadsheet.