Hello.  The purpose of this blog is to post my game simulations and the occasional analysis of a game or topic.  I post almost everyday.  Occasionally when traveling I will take the day off.  I usually take a two-week vacation each year and will have limited post during that time.  I am not selling anything, just providing my analysis on games and bets.  I also try to post a few picks on Twitter each day.  Thanks for reading.  Good Luck!


10 Responses to About

  1. german says:

    Amazing suff. wondering if you would be willing to share your math model. it seem to be the best.

    • derekbets says:

      I would definitely be open to discussing my models. Probably won’t disclose the detail of the inner workings but would be happy to discuss the theory behind them and possibly post on how they work. Thanks for the note!

  2. Brandon Beylo says:

    This site is incredible! I am a high school senior and am taking an AP Statistics class. Believe it or not this is helping me understand that! Anyways, Derek, how would you go about making a betting log. Do you recommend free hand? Or on excel? Or just printing out your simulations?

    • derekbets says:

      Thanks for the comment and glad this site helps. As far a betting log, I would start with a process that is easy for you and fits best with how you do your best work. I would lean toward something electronic that you can easily reformat and work with to do testing/analysis. Here are a few options. 1. Set up a Google Drive account and create a Google Spreadsheet (both are free). With this you can access the spreadsheet from any PC. 2. Using Dropbox, Drive, or Box you could use excel and sync this across multiple computers or access via the web. 3. Again with a Google Drive account and IFTTT.com account (both free) you could set up a Google Spreadsheet and simply text message your bets to the the spreadsheet. Those are just a few options. I think the keys are ease of use and access, being in a flexible, electronic format, and capturing complete information (league, game, bet type, line, your predicted score/line, outcome, bet amount, etc).

      Good luck with AP Stats!

      • Brandon Beylo says:

        Thank you! I just set up a Google Drive account and made the tabs for league, game, bet type, etc. One quick question … What does ‘your predicted score/line’ mean? Does it mean what I think will happen? My pick?

      • derekbets says:

        By ‘your predicted score/line’, I am referring to the score or line you have simulated or handicapped the game at. For instance, if you simulate a football game at home team 23, away team 10. If you were betting the home team, your predicted line/score would be -13. So yes, what you think will happen. I like to record this so I can measure how I am doing based on how large/small the gap is between my line and the actual line. Hope that helps.

      • Brandon Beylo says:

        Okay thank you! What I have done is calculated the projected win percentage by using the “implied money line projections that you talk about. For example +150 is like a 40% chance of winning. However, how do you come up with your simulated score? And then your spread after that?

      • derekbets says:

        That’s the hard part. All of my simulated scores are based on models I developed using team/player statistics. For a sport like football or basketball, my model output is a score prediction so i can easily create a spread from the difference in the scores. When it comes to baseball and hockey where the money line is the primary betting form, it is trickier. For those sports, I evaluate performance over time to determine a win probability. I use this win probability against the implied odds of the money line to identify value.

        Here is a quick/simple overview of how I create a model. 1. I gather as much data as possible (e.g. game outcomes, team and player stats). Multiple seasons is preferred. 2. I will do some statistically testing on the data to identify what info is critical in the outcome of a game (e.g. regression). 3. I construct each model on how I understand teams score. This relies heavily on the results of step 2. 4. I test the model on a data set. 5. I put in place a process to gather information after each game, update a database, and run the model for the next games.

        That is a very brief overview of how I build the models I use. You could also use a ratings system such as ELO or Sagarin. Sagarin is pretty popular and available for free. Or you could use a neural network model. There are several options. I use my models and a couple of ratings systems too. Building your own model or ratings system would probably be a helpful exercise.

  3. Derek- had to chuckle as I read your steps 1-5. My modeling approach is identical. It’s funny how people converge to the same methods independently.

    I was wondering which metrics you use to evaluate your model’s performance on the test set- RMSE and line movement are my top choices. Given a small sample size (and a reasonable efficient market) I would actually prefer to see concordant line movement than a high (actual) win rate. Keep up the good work.

    • derekbets says:

      Thanks for the note! Yeah, I have spoken with a few other people who started in a completely different point but their process evolved into something similar to what you and I both do.

      I use both RMSE and line movement. I also use multiple ratings/models with my own models. Always working to improve and find new approaches.

      Good luck!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: