Posted on

July 29, 2023 MLB Baseball Betting Cards details:

No model changes today.

I will provide a few more details into the recent 9.2 changes as it did cut the MAINS down significantly.

Expanded Model 9.2 Notes:

I’ve tracked every game this season for 2023. I have recorded the teams in each matchup, the pregame Fangraphs WP%, and the first odds I saw for the game. In addition I track the final outcome and then I have background data to crunch all these data points and calculate an adjusted trend for each team based on their L90 performance. This adjustor is the key to the model as it tries to course correct Fangraphs WP% closer to the true performance for each team.

Now let’s look at the performance of the qualifying bet parameters.

MAINS:

  • I currently have the model set at an eROI of 13%, this is my adjusted WP% for a team divided by the BE ML % of the odds in the game.
    • For example: July 26th I had MIA @ TBR, my adjusted WP% had MIA at 50.7%. The odds I recorded were +140 (this is before line shopping which I don’t record best odds in my back data, only in my actual bet log of profits/losses) which tranlsates to a BE ML % of 41.7%.
    • 50.7%/41.7% -1 = 21.5% eROI, this is the expected ROI from this bet if my projected WP% is accurate.

Backtesting Mains:

  • For backtesting I mainly look at games since May 1st as the L90 adjustor is weird in April games due to teams like TBR not losing until game 14 or so or Oakland and others losing a bunch. Data from May 1st gives the model about 30 day or so to crunch performance and things smooth out considerably once that count increases and gets closer to a full set of a rolling 90 days.
  • Since May 1st, there have been 1,122 completed games so far.

13% MAIN eROI:

  • The 13% eROI has qualified 49 games
  • With a 28-21 record (57%)
  • Resulting in 106.8 Units won
  • On the current Unit size of 7 units for 343 Units bet and a 31% ROI

Previous parameter of 5% MAIN eROI:

  • The 5% eROI has qualified 141 games
  • With a 71-70 record (50%)
  • Resulting in 87.8 Units won
  • On the current Unit size of 7 units for 987 Units bet and a 9% ROI

Overall I’m mostly focused on tailoring the model so that the maxium amount of units are won.

Higher ROI is great, but if I need to lower the ROI to capture more Units and bet more games I will do that. In this case there was nothing to be gained by keeping the qualified game count elevated with a lower ROI as the total units won, 106.8, was greater on the higher criteria of 13% eROI vs the 87.8 won at 5%.

RUNTS:

Current parameter of 18% MAIN eROI:

  • The 18% eROI has qualified 101 games
  • With a 54-48 record (53%)
  • Resulting in 257.4 Units won
  • On the current Unit size of 7 units for 707 Units bet and a 36% ROI

Both Bet Types were backtested with various eROI cutoff criteria and these proved to be the best levels to set the model at to maximize units won. Best of all the data holds in monthly blocks and doesn’t appear to be influenced by any overall streak, suggesting it should be a sustainable trend.

The June performance for MAINS is a little bit iffy, but it’s not negative and overall the total is showing 31% ROI. I think if I can even get half these results that would be amazing as 30% ROI to the books is insane value. The game quantity may take a hit here, but it could prove out that I can be a little bit more aggressive in Unit sizes going forward if these results can be replicated.

I am not too concerend about the validity of the results here as my data is set to look at each team’s performance the rolling 90 days prior to each game being analyzed. That is why I don’t count April in this data set as the adjustor swings the WP% on each pick wildly as it sees 100% or 0% win records to 40-60% baseline WP% from Fangraphs.

The only flaw that I have in the backtested data is that I can’t account for each game that may have truly qualified as I don’t have recorded the best line after line shopping. To be fair it shouldn’t swing too wildly, but it could move the needle about 5-10% in eROI…. which is likely a better test. I’ll experiment in future model adjustments to see if setting my opening lines down to about 95% of it’s current BE ML% impacts this and how to best capture that movement.

Overall though I hope this explanation proves helpful in showing why I moved the cutoffs in eROI.

Current Missing Lines/Close Lines:

These are games missing lines currently or games that I’m looking to check back on as the best lines I’ve found so far are close to qualifying for a bet. The line indicated is what will qualify the bet, parentheses is the closest odds I’ve found.

0 games missing lines:

  • CLE +110 @ CWS +175
  • BOS +120 @ SFG +160
  • OAK +160 @ COL +115

3 games are close to qualifying:

  • DET +155 (+140)
  • WSN +155 (+150)
  • MIL +205 (+180)

Last checked @ 3:30 PM, No new games qualified. Nothing is budging, in fact moving away it looks like. I’m calling the card final today. Annoying to have a full slate and such little, I’m an action junkie lol. But I’m focused on maintaining profitability as best as possible. I’ll be doing some model testing to see if I can find a way to reasonably bring game counts back up.

CARD IS FINAL

DAILY CARD:

*See Glossary for details to help explain terms and other recent model changes.

___________________________________

___

2 Replies to “July 29, 2023 – MLB Daily Sports Betting Card and Odds”

    1. For now, I do like the idea of each game having the same amount of risk. Not to say they’re gone forever, just the current iteration of the model doesn’t use them.

Leave a Reply