We have released an update to the daily MLB projections beginning with June 7th games.
This upgrade took a longer time than expected to complete as it required almost a full rewrite of the code but I am psyched to finally release it!
Thanks to all of you – especially you rabid DFSers – who have provided suggested improvement areas throughout the season. Behind the scenes, we automated several processes to ensure that daily projections are updated as early as possible (~6:15 AM EST) and then updated throughout the day regularly to reflect posted lineups (every 10 minutes. increased to 5 minutes between 4-7PM EST).
Below are some of the noteworthy upgrades with our new projection system:
- We are now incorporating both Steamer hitter and pitcher split projections. Our legacy projections used Steamer hitter split projections and then adjusted pitchers based on league average splits. So when we are projecting how, say, David Ortiz will fare against Rich Hill, the foundation is now based on David Ortiz’s projections against LHP and Rich Hill’s projections against LH hitters. Steamer’s split projections factor in past performance vs each hitter/pitcher split and then regress them to league average splits. The larger the sample of recent past performance, the less regression that needs to take place.
- We are reducing projected PA/ABs for hitters who tend to be removed before end of game for a pinch-hitter. These hitters typically fall into two camps: 1) A platoon hitter who is pulled once the starting pitcher is pulled and the platoon advantage is lost or 2) A weak hitter. The PA reductions are based on last 30 game usage of the hitter. Examples of platoon hitters who often see the bench before final pitch include Chris Coghlan (loses an average of about 0.8 PA/game), Kike Hernandez (~ 0.6 PA/game), and Ryan Raburn (~0.65 PA/game). Examples of the latter include Paul Janish and David Ross.
- Better handling of ‘swingmen’. The daily pitcher split projections we are using from Steamer now reflect a pitcher’s projection specifically as a Starter vs a Reliever. The previous method relied on Rest of Season projections which would blend the results of these two roles. The Rest of Season projection methodology has been similarly updated.
- Improvements in Run/RBI formula. Hitter Runs/RBIs are now 100% correlated to the pitcher run projections. While they had been very closely correlated (~95%), through this process, we uncovered an issue with the previous method that would lead to inflated run projections in high-scoring games and deflated run projections in low-scoring games. The new methodology fixes this.
- Improvements in SB formula. The old formula multiplied a hitter’s opponent-neutral likelihood to steal against an index based on the starting pitcher’s stolen base rates vs league average. This method works fine in most cases but overstated the SB projections in extreme cases (e.g., Billy Hamilton vs Jimmy Nelson). The new method better handles extreme matchups. Â
The below stats reflect the differences seen between the ‘legacy’ and ‘new’ projections using week of June 6th data. Most of the projections are north 0f 0.9 (on a scale of 0 to 1) which reflects that these improvements, while significant, are not dramatically different than our previous estimates. The ‘new average’ column reflects the average seen across all projected starting pitchers + hitters on the active roster. The ‘Absolute +/-‘ column reflects the average absolute value of the differences seen vs the previous projections.
The correlations for DraftKings/FanDuel hitter points were both around 0.937. The correlation for DraftKings/FanDuel pitcher points were both around 0.98.
Please let me know if you have any questions or future suggestions in the comments, via e-mail ([email protected]) or via Twitter (@rudygamble).
Streamonator/Starting Pitcher Projections (268 GS)
| Correlation | New AVG | Absolute +/- | |
| Â $ | 0.96 | 0.04 | 3.28 |
| Â QS | 0.97 | 0.52 | 0.02 |
| Â W | 0.95 | 0.35 | 0.02 |
| Â L | 0.93 | 0.33 | 0.02 |
| Â IP | 0.97 | 5.69 | 0.10 |
| Â H | 0.93 | 5.64 | 0.19 |
| Â ER | 0.91 | 2.47 | 0.11 |
| Â K | 0.98 | 5.06 | 0.16 |
| Â BB | 0.98 | 1.94 | 0.10 |
| Â HBP | 0.86 | 0.21 | 0.03 |
| Â HR | 0.93 | 0.68 | 0.04 |
| Â K/9 | 0.98 | 7.98 | 0.26 |
| Â K/BB | 0.98 | 2.74 | 0.18 |
| Â ERA | 0.94 | 3.94 | 0.19 |
| Â WHIP | 0.91 | 1.34 | 0.05 |
Hittertron/Hitter Projections (3,378 Games)
| Stat | Correlation | New AVG | Absolute +/- |
| Â $ | 0.87 | 0.92 | 5.01 |
| Â PA | 0.86 | 4.09 | 0.13 |
| Â AB | 0.93 | 3.69 | 0.12 |
| Â H | 0.93 | 0.93 | 0.05 |
| Â 1B | 0.90 | 0.63 | 0.03 |
| Â 2B | 0.90 | 0.18 | 0.01 |
| Â 3B | 0.92 | 0.02 | 0.00 |
| Â R | 0.98 | 0.44 | 0.04 |
| Â HR | 0.92 | 0.10 | 0.01 |
| Â RBI | 0.96 | 0.42 | 0.05 |
| Â SB | 0.97 | 0.06 | 0.01 |
| Â BB | 0.94 | 0.32 | 0.02 |
| Â HBP | 0.96 | 0.04 | 0.00 |
| Â SO | 0.94 | 0.83 | 0.06 |
| Â AVG | 0.95 | 0.252 | 0.01 |
| Â OBP | 0.95 | 0.314 | 0.01 |
| Â SLG | 0.95 | 0.395 | 0.02 |
| Â OPS | 0.94 | 0.710 | 0.02 |
Teamonator / Team Projections (268 Team Games / 134 Actual Games)
| Correlation | New Avg | Absolute +/- | |
| Â BF | 0.53 | 37.95 | 1.06 |
| Â IP | n/a | 8.77 | 0.23 |
| Â QS | 0.97 | 0.52 | 0.02 |
| Â W | 0.97 | 0.50 | 0.15 |
| Â H | 0.86 | 8.60 | 0.40 |
| Â HR | 0.94 | 1.02 | 0.07 |
| Â ER | 0.93 | 3.79 | 0.18 |
| Â SO | 0.95 | 7.80 | 0.57 |
| Â BB | 0.97 | 2.98 | 0.12 |
| Â ERA | 0.94 | 3.90 | 0.22 |
| Â WHIP | 0.93 | 1.32 | 0.05 |
Forgive my ignorance but didn’t the hitter daily projections show the opposing pitcher before? Now it is showing their teammate that is pitching.
@Moore Semien than a Blackmon:
I don’t think it is correct Semien. I sent him a message on Twitter.
@Moore Semien than a Blackmon: Yup – goof on my end. Fixed.
Nice!
Wow this is pretty cool. Rudy can’t fail
THANK YOU for this!
@knucks: it’s either death or glory.
@Rudy Gamble: …becomes just a Trevor Story.
PS This is awesome, Rudy — thanks!
nicely done (typed from my dingy basement)
Lots of ‘perts claim that SP wins are less predictable than other stats. Do your analyses of SON confirm or disconfirm that view?
Thanks, as always, for the great work!
Thanks for all the hard work Rudy. These tools are fantastic and the new formula looks great.
I didn’t want to make a request on this day of glory but I have a feature that I keep forgetting to ask for:
I play in NFBC leagues (weekly as you know), but you can set hitters on Mondays and again on Fridays. I usually manually add the next 3-4 days to give me an idea of who has the best matchup…
Could we get a next 3 days/series button that sums this on the Hittertron? I like to play matchups the little that I can.
Again, thanks for everything. Your hard work to bring us these projections is unquantifiable.
@Mr. Snrub: Thanks. Good idea! So what you really need, then, are ‘Mon-Thur’ and ‘Fri-Sun’ projections, right? Given I do Mon-Sun projections it’s definitely possible…just a matter of re-purposing code. Will look into it.
@Rudy Gamble: Ideally, yeah, Mon-Thurs and Friday-Sunday links would be ideal, but…
Probably easier to re-purpose the query for your next 7 days page though and just go with next 3 days…
I too am a student of the programming :D
Thanks for the prompt response per usual!
@Mr. Snrub: Next calendar week projections force me to identify days of week. So the awkward part is already done.
@Rudy Gamble: Even better! The only thing better than the best projections site is that site catering to my specific needs.
Gimme gimme gimme, I need, I need!
I guess the only question left is when will those update? Mondays I assume along with the next calendar week page?
@Mr. Snrub: ha. well, this will help other NFBCers too. Both would be first available on Fridays and then Mon-Thurs would update through Monday and Fri-Sun through Friday.
@Rudy Gamble: God damn that is beautiful. Thanks again Rudy!
@Mr. Snrub: Fri-Sunday now live (link in top menu). Mon-Thurs will go live on Friday.
@Rudy Gamble: I am using it right now. So helpful. This is gonna be a game changer.
The NFBC players of the Razzball community thank you. If finish up in the moneys I’ll make a bigger donation next year!
thx for the idea. should’ve thought of it myself.
Can you clarify what the correlation numbers are measuring? If you were able to project the night’s stats with 93-98% accuracy, you’d be a rich(er) man, so it can’t be that. I know you explained it in text but somehow I’m not grasping. You’re measuring the correlation between your new projections and… ?
@TheTinDoor: The corrleation numbers are versus the previous projection methodology. No way I can get 93-98% correlations vs actual!
@Rudy Gamble: so I think that’s the better measure though. How much better is the new method at projecting the actuals?
@TheTinDoor: it’s hard to say. only time will tell via the Ombotsman testing. They are close enough that it takes a decent-sized sample to distinguish. I tested throughout the period but used that to identify areas of improvement/bug checking.
I’m noticing that RP’s, closers and others are now very underrated. Even the top closer is shown as worth about $6 ROS. Is this a glitch? Prior to the update they used to show more in the $20 range which would be much more logical.
@Rico: good catch! fixed.
Tough first day, per the Ombotsman. Need mid-20s correlation for DFS success, typically.
@Charles: i just focus on how the correlation is vs DK/FD. Yesterday was a slightly tough night for hitters but always prefer to see us beat the DK/FD salaries.
Hi Rudy, I’m exited to use M-Th evaluator for hitting, but it keeps bombing on me. Is there a fix in the works?
excited @billy beeeen:
@billy beeeen: I just checked and it’s working fine for all subscriptions. Are you logged into CoinTent?