For what it is worth, in my MLB model, I found that my best ROI came from betting underdogs only up to plus 115 and betting favorites only down to minus 125.
@arthurkent9847Күн бұрын
Thank you for your work. Much appreciated.
@arturoburigo2 күн бұрын
Hello sir, how's your model Doing? R u counting the W/L ?
@aleferstein28914 күн бұрын
Thank you for posting this information. Are you going to post findings through the playoffs?
@numeristical4 күн бұрын
I'm going to stop at the end of the regular season. Too few games after that and I suspect the model will mostly align with the offered odds.
@puneetchahal59534 күн бұрын
Can we try building models for other sports?
@aleferstein28915 күн бұрын
Thank you for letting us know.
@zimrihidaf5 күн бұрын
Thank you
@nathanie16377 күн бұрын
Good luck on the event! Please please consider open sourcing this model. or revamping your series on how to create a model like this.
@jordang81357 күн бұрын
Yeah I would like to see that as well. Good luck in the triathlon
@puneetchahal59537 күн бұрын
You the best. 🕺
@DealHunter7048 күн бұрын
Love the series and you have almost done it the entire season. Do you think you will be able to tune the model for next year? It seems the totals predictions is solid. Great work
@KhushalTrehan-xc8bt10 күн бұрын
Sir, is there any way I can contact you? I wanted to ask something
not very good picks, keep trying, you will figure it out.
@aleferstein289113 күн бұрын
Thank you for posting this before noon. Helps me with my plays today. Would you consider adding NFL? Much data is available, and you will generate more of a following?
@aleferstein289114 күн бұрын
Thank you for posting the picks, how about NFL?
@numeristical13 күн бұрын
I probably won’t get to an NFL model this year. Maybe NBA depending on the other stuff going on in my life…
@puneetchahal595314 күн бұрын
Your the best. You have a fan for life
@numeristical13 күн бұрын
Thanks!
@richt8714 күн бұрын
Take the Rockies
@numeristical13 күн бұрын
I should have!!
@richt8713 күн бұрын
@@numeristical KC Royals seems to be the play today, thanks for your content.
@aleferstein289115 күн бұрын
are you posting today 9/13?
@aleferstein289116 күн бұрын
Thank you for letting me know. How about adding NFL football to the show?
@aleferstein289116 күн бұрын
Can you post earlier so I can make better decisions?
@numeristical16 күн бұрын
Sorry - got a late start today and once I got to it the model had no picks anyway... Will try to post earlier but sometimes life gets in the way!
@aleferstein289116 күн бұрын
@@numeristical Thank you for letting me know. How about adding NFL football to the show?
@landonpieper146717 күн бұрын
Do you think your model is over fitting for the white Sox lol ?
@numeristical17 күн бұрын
Hard to say. As I’ve mentioned before, it’s possible the model has never seen a team this bad before and so is systematically overestimating their chances. Or it could be right and we are just watching the variance play out. We had a couple big wins with them in the last month
@aleferstein289117 күн бұрын
Like your computer system set-up but obviously the machine is not learning very quickly. How about adding college and pro football to your system? This would bring in more followers. Please try to post earlier so I can make better decisions. Your totals picks seem better than the sides, yet if you throw out the huge underdogs, the results are much better. Thank you for your time and effort.
@jean-francoisbouvier857620 күн бұрын
At 8:30, we see there is 162 columns in your file. Coincidence that the number of columns is equal to the number of games a team play in a mlb season?
@numeristical19 күн бұрын
Haha - yes, total coincidence!
@jean-francoisbouvier857620 күн бұрын
Rockies and White Sox get the victory...today was a good day for AI!
@jean-francoisbouvier857619 күн бұрын
Gosh, just realized AI did not recommend them….
@jean-francoisbouvier857621 күн бұрын
I'm trying a small bet on the White Sox too at +175 tonight. Hope Crochet is prepared...
@landonpieper146722 күн бұрын
Hey do you have the videos where you create this model on your page ???
@numeristical21 күн бұрын
Look at the playlist "Baseball Prediction using Machine Learning". I don't build this exact model, but they build a similar model. My model uses this approach with some other sneaky features and tricks.
@puneetchahal595322 күн бұрын
Thank you making these videos. They help me in a weird way
@numeristical21 күн бұрын
Glad it's helpful!
@douglasscholtes544822 күн бұрын
This is awesome. Been watching these for about a month trying to teach myself python. I did this very thing back 2000-2005 trying to teach myself Java. Didn't know about Python. Had to use screen scraping to get box score and odds info. So hard getting the data. Every time they changed the web page I'd have to redo the code. All I knew how to do was linear regression and did not know how to properly analyze the results to determine how well the model was doing. Thank you for doing this.
@numeristical21 күн бұрын
Thanks for watching! Glad you find them helpful.
@LynoldMercado23 күн бұрын
so how does the module work? It is my understanding that module tends to pick who's going to win a match + have favorable odds / payour over the long run rather than this particular head to head, is this a fair assessment? I say this because looking at the white sox win over baltimore the other day, the line up I think it was bet on the White Sox at +325 not because it thought they were going to win, but because the pay of them winning vs the payout of baltimore winning was drastically higher.. And now we see a similar set up with Rockies vs Braves. Braves have won both games so far over the rockies by at least 1.5 and now the module is saying taking the Rockies money line at +266 despite Braves have a way better pitcher. Curious to hear your thoughts!
@numeristical21 күн бұрын
Right, the idea is not "bet this team because they will win". Rather, you bet a team because you believe the probability of them winning is high enough that it is a "positive Expected Value" (+EV) bet. Mathematically, if you make positive expected value bets, in the long run you should make money (though it may take a while to get to the "long run"). So, given the models predictions about a set of games, it will suggest a bet on any team if "the price is right" (i.e. if the payoff is favorable enough). If it thinks the White Sox will win only 20% of the time, but the bookies are offering +1000, it is a good bet. Despite the fact that the very likely outcome is that you will lose the bet.
@landonpieper146724 күн бұрын
More please
@numeristical23 күн бұрын
yes, I'd like to get back to more of this kind of content...
@jean-francoisbouvier857625 күн бұрын
No bets on White Sox this time? 😅
@numeristical23 күн бұрын
Finally hit the one at +325 !
@jean-francoisbouvier857623 күн бұрын
@@numeristical AI is patient, most people would have lost faith at this point!
@jordang813525 күн бұрын
Great series and info thanks for all of these videos
@numeristical23 күн бұрын
Thanks!
@jean-francoisbouvier857626 күн бұрын
Thanks for mentioning some of my previous comments about White Sox being such a bad team and AI maybe not having enough similar MLB historical data to assess how bad the team is. That being said, I laid a small bet on White Sox today at +350. Not really expecting them to win, but I feel at the corresponding 22,22 % required win probability to be in the money, it is a fair bet. Personally, I don't see how we can bet the White Sox at odds under 2 to 1 and expect to make money. AI or no AI involved. Even if it's Garrett Crochet on the mound. The team is just unable to score more than 1 or 2 runs per game, and even if their starter has a great outing, odds are the bullpen will crumble in the late innings to let the win slip from their fingers.
@numeristical21 күн бұрын
Yeah, the White Sox hitting is pretty terrible. Still, they managed to score 8 runs against a pretty good starter / bullpen for the Orioles. It's all about trying to figure out those probabilities.
@jean-francoisbouvier857621 күн бұрын
@@numeristical I agree, they are always alive as long as the game aint over officially. But they really reshaped the meaning of mediocrity in baseball.
@jean-francoisbouvier857627 күн бұрын
You pout aside the Whit Sox bets, and most of the other picks were right. If not all. AI model seems to need some improvement with such a bad team.
@nathanie163727 күн бұрын
Would you be open to open-sourcing the current model you're using?
@jean-francoisbouvier857627 күн бұрын
Nice underdogs picks for for the Rockies and Angels. Unfortunate for the White Sox, I assume the AI had never encountered such a bad team in previous history. We'll see if he learns...
@carl0228Ай бұрын
I just found your channel. I will start making some bets using your machine learning. Do you have the year-to-date percentage of your progress? Also, do you have a model for football? Great content!!
@numeristical21 күн бұрын
I give the "yearly progress" at the beginning of each video in terms of units won/lost to date for each of the three models.
@nathanie1637Ай бұрын
Are the picks generated from the model developed in your ML baseball prediction videos? If so, this is very promising. I was able to hit two tickets mix and matching some moneyline and over/under picks.
@numeristicalАй бұрын
It's not the model from the videos, but it uses the same approach. I added a few improvements in features and a couple of extra tricks.
@nathanie1637Ай бұрын
Rays, Cubs, and Rays money line parlay hit for me
@numeristicalАй бұрын
Nice! you listed Rays twice - did you mean the Nationals?
@nathanie1637Ай бұрын
@@numeristical sry meant to say padres
@nathanie1637Ай бұрын
pitcher strikeouts, please do a dataviz notebook on how to develop an ML for strikes
@numeristicalАй бұрын
I'll note the request :)
@IllChannelАй бұрын
Since I'm not touching Padres/Cards U 8 I am confident it will hit today.
@numeristicalАй бұрын
no such luck for me. Glad it didn't get chosen today.
@IllChannelАй бұрын
Padres love when i do Unders.
@jean-francoisbouvier8576Ай бұрын
Good pick on the Athlectics.
@numeristicalАй бұрын
Thanks - hopefully the luck starts to turn our way again. Very roller coaster season so far.
@LynoldMercadoАй бұрын
are you still using oddsshark for past odds ? looks like they have not updated the histrocal odds since aug 22
@numeristicalАй бұрын
No, I bought a historical odds data source and use the odds api for the real time stuff
@puneetchahal5953Ай бұрын
Fix the echo. Good call on the rays last night.
@numeristicalАй бұрын
yeah, sorry about the echo....
@numeristicalАй бұрын
Sorry for the echo! Haven't figured out how that occassionally happens....
@LynoldMercadoАй бұрын
Betting on the white Sox was just perfect
@numeristicalАй бұрын
yeah, too bad everything else lost that day!
@wisetom6575Ай бұрын
I might have missed this when you were setting up your model, but I was curious: considering you're using fairly straightforward and widely available stats (hitting, pitching, etc.), where does your model's edge come from compared to the LV models? I imagine they're using the same stats, plus more detailed insights like player behavior, well-being, and so on. If a more 'basic' model based on raw performance stats is outperforming theirs, wouldn't they adjust for that as well? I'm really enjoying the series-thanks for putting in the effort!
@numeristicalАй бұрын
A deep question to answer here, but the short reply is that LV is not using sophisticated machine learning and is relying on more "heuristic" judgements. They are quite good at it! But there are cases where they may be off by a few percentage points because they overvalue or undervalue certain things. They also respond to market forces which may contain their own biases.
@Pak02Ай бұрын
I can't run the code %matplotlib inline error: invalid syntax. Can u help me to fix it
@numeristicalАй бұрын
It's only used if you are in a jupyter notebook. You can just comment it out if it is giving you trouble.
@Pak02Ай бұрын
@@numeristical thank you so much, i forgot to open jupyter notebook in vs code
@jean-francoisbouvier8576Ай бұрын
A little bit of echo, but always like your picks!
@numeristicalАй бұрын
Thanks - yeah, sometimes the echo creeps in. Haven't figured out waht causes it, but will try to avoid it.