This account and website were built on streaming pitchers, so who better to dive in and figure out a formula which can lead to streaming success? After asking a few people in the industry which offensive stat holds the most weight, the most common answer was wRC+. Using wRC+ combined with other stats produced some very interesting results, but before we get into that there is a caveat here.
It is impossible to find ownership rates on players for specific days in the past. For this study, I wanted to use pitchers who were owned 30% or less on the day of their start. Pitcher’s ownership fluctuates a ton, for instance, Lance Lynn was streamable in April. The subset of pitchers used was based on my streaming sheet from last year (everyone I streamed was 30% owned or under via Yahoo) and by using the obviously available streamer options such as Antonio Senzatela.
I’m a firm believer that when it comes to streaming you want to look at a team’s stats over the last 14 days. My thinking is you can find a team on a current cold streak. For instance, let’s say it was May 15th and I wanted to stream a pitcher. I would look to see which team was at the bottom of the league in wRC+ from May 1st to May 14th. From here on out, know that all thresholds are based on streamers facing offenses who met that threshold in the last 14 days. Side note: this can all be found on Fangraphs.com.
70 wRC+ Threshold
When it comes to wRC+ I played with a few different thresholds to try and create the best results using just that stat. The best threshold for 2019 seemed to be 70 wRC+. According to Fangraphs that is between a Poor rating and Awful rating. These were the results
Underwhelming results for sure, while the league average ERA was 4.50 the result of a 4.86 ERA might please some but not many. I was expecting better results here so I figured we had to start combining wRC+ with other stats.
25.0 K% and 70 wRC+ Threshold
Next, I went with K% because the more a team strikes out the better to stream against, right? After playing with a few thresholds, the best result was when I used the threshold of 25.0 K% or more and paired it with the 70 wRC+ threshold.
Slightly better ERA here as we went from 4.86 to 4.73. What shocked me here was the lower K/9 as you would think pitchers would have an easier time against strikeout prone teams. With yet again mediocre results I moved on to another combination.
12.0 SwStr% and 70 wRC+ Threshold
SwStr% is a very popular stat for plate discipline, after trying different percentages I found the best threshold to combine with the 70 wRC+ was if a team had an SwStr% of 12.0 or higher.
Like K% the ERA was better but came out the exact same at 4.73. Relying on SwStr% and wRC+ appears to give you higher strikeout potential. The WHIP was also the best we have seen so far, but once again, the overall numbers weren’t what I was expecting.
To save you from looking at too many graphs I also tried wRC+ with OPS and GB% but neither panned out. Both ended up with worse WHIP’s, worse ERA’s, and lower strikeouts.
.275 wOBA and 70 wRC+ Threshold
Eventually, I arrived at testing out 70 wRC+ thresholds with wOBA. Now, wOBA is used in the wRC calculation so maybe this would be a good way to find outliers and create better results.
It certainly gave us better results. Combining the two we were able to produce a 4.68 ERA in 801.2 innings pitched with a 1.35 WHIP. All much better than the original table using just wRC+. Now came time to refine this.
.275 wOBA, home-field advantage, and 70 wRC+ Threshold
With these better results, I decided to add in home-field advantage. Let’s take a look at what would have happened if you took streamers at home while combining our wOBA and wRC+ thresholds.
This is something we can live with. A 4.63 ERA certainly isn’t flashy but the fact that you can work the wire with scrub starting pitchers and produce an ERA just .13 above league average isn’t terrible. The only concern with this method is the very high WHIP.
I put up a poll on twitter asking people what ERA they would be happy with from their streamers. The majority said 4.50 while a good chunk said 4.75. Whether these are good results or not is up to who you are asking. Personally, coming from someone who picked streamers every day last year the results are pretty good. Especially if you think about it this way, a quality start is a 4.50 ERA. That means by using this method you are getting slightly worse than a quality start.
The best and worst streaming performances included in this study.
Touki Touissant vs. CLE: 1.1 innings pitched, 7 earned runs.
Tyler Beede vs CIN: 2.1 innings pitched, 7 earned runs.
Pablo Lopez vs NYM: 3.0 innings pitched, 10 earned runs
Ivan Nova vs TOR: 3.0 innings pitched, 8 earned runs
Antonio Senzatela vs ARI: 4.1 innings pitched, 7 earned runs
Reynaldo Lopez vs MIA: 8.0 inning pitched, 2 earned runs, 10 strikeouts
Chris Bassitt vs CHW: 7.0 innings pitched, 0 earned runs, 7 strikeouts
Elieser Hernandez vs SFG: 5.0 innings pitched, 1 earned run, 9 strikeouts
Ross Detwiler vs TEX: 6.0 innings pitched, 1 earned run, 8 strikeouts
Merrill Kelly vs. SDP: 7.0 innings pitched, 0 earned runs, 9 strikeouts
I plan to look deeper into streamers and try out different methods. I also plan to track every streamer’s ownership rate this year to really produce results with more substance. More research is certainly to be had so consider this part one of more to come. This article took some time and I would love to hear your thoughts, so please reach out to me on twitter @SPStreamer or via email SPStreamer4@gmail.com. Thanks for reading!