Has anyone figured out why seconds per shot is being recorded by everyone. Seems like such an arbitrary metric to track.

Thank you for posing this question, got a lot out of it. I did a similar version of what diced pineapples did, but used kelly criterion instead. I found that we should be wagering more of our bankroll on the 3 way line except for when the tigers win probability was => 0.5. Then we should be using the 2 way line.

Thinking I'm gonna try my hand at hits again. Also gonna bet home runs yes or no because it's the same process. But can someone tell me if it's the same process for run scored yes or no? Something in my head tells me this might not be right for calculating run scored yes or no because runs scored aren't so directly tied to PA in the same manner hits or home runs are?

Prop builder/bet builder whatever it is called has Wander Franco 1+ hits -118. Go smash that, it's way way way off.

sorry odds changed as soon as I posted this

Here's another way to think about it. Calculate the no-vig line for both the two-way and three-way matchups and use the binomial theorem to back into the implied per game win percentage for each team.

I like the "synthetic 2-way" line idea and clearly the probability spaces towards the bottom of the spreadsheet have +EV. If all I had to go on were the two sets of market prices provided in the original question I would have assumed it was a "no-bet" situation though:

Using the no-vig 3-way probabilities of a DET win/draw/lose={0.11,0.29,0.60} or the probabilities derived from Binomial(N=x,P=a,K=4) where x=(0,1,2,3,4) & a=avg(0.11,0.29,0.60)=0.33 with the proposed synthetic payouts of {win=$703.49, draw=$0.00, lose=-$146.50} I came up with an expected value of -$10.6 and -$11.6 respectively.

If I had a good reason to believe those markets were significantly mis-priced then naturally the EV changes, but I definitely don't know enough about handicapping baseball to make that call myself

I'm on a book i think has many inefficiencies, specifically pitcher strikeout props. What is best to compare to for tight lines for edge?

I originally compared to Bovada, but have always heard pinnacle is the tightest. And pinnacle lines actually differed pretty significantly from Bovada to my surprise.

Thoughts?

Whoops I said I would let everyone know how the NHL Faceoff thing went and then deleted my account. Here it is sorry guys.

This is based off the Bayesian paper we talked about that accounts for hands. I only used faceoffs that had a sample size of 40 or more vs. the hand and I only used their last 200 faceoffs. The sample size was neutral only faceoffs.

This created 691 games that met the criteria. I also set a min edge of 0.05 for a bet to be bettable.

Of those 380 were no bets and 311 were bets. I only bet each faceoff once and no playoff bets.

The total record went 184-127

Risk: $90,411.50
Net: $9,487.50

ROI: 10.48%



Not too bad. Throughout I kept track of a system that doesn't use the approach from the paper:

Net: 2462.5 on 92987.5 bet. Record: 172-141

Only two more bets not as good of a result.

It was interesting project. Going forward I doubt I will bet them. It's too much work to sit there, wait for the lineups, hope it doesn't move, etc. but it is nice to show profit in the NHL for once!

just out of curiosity i went back from 2016-present with the same restrictions (40 sample size, 0.05 edge). Since I don't have faceoff odds from then I just assumed they're all -115

Here's what it gave us (no backtesting, just pretending hey today is oct 15, 2016 what would the line be based on pre oct 15, 2016)

1398-937 (59.87%)

2887 no bets. so 5,222 games met the criteria.

Risk: $671,312.50
Net: $80,112.50 (7.46%)



It looks like this year we ran better than in the past (better ROI) but won less because of the shortened NHL season (56 games vs. 81). It looks like 2016 was a bad year. OUtside of that it looks about ~20k/year.That ~18k downswing at the end of 2017/beginning of 2018 looks rough.

Did you give each of the most recent 200 faceoffs the same weight, or was there some sort of scaling? And I don't remember the paper exactly but was it the generalized log 5 one for batters vs pitchers of different handedness?

Nice results, congrats. I also messed around betting the NHL Opening Faceoffs earlier in the season. It showed a profit although I did not track results/ROI. I'd estimate I ran somewhere in the 4%-8% range. It's a good market for people willing to put in the time for a decent-sized edge. I don't think I'll bet it much anymore myself. Too fat and lazy now. Thank you to fubster and others for showing how to beat this market. That's more satisfying than the money. All of the info is in this thread for those dedicated enough to find it.

JSkelts, the paper was "Modeling the probability of a batter/pitcher matchup event: A Bayesian approach" It's open access so it's on google.

I couldn't get it to improve vanilla log5 results, but he said he didn't use the same priors as the paper. Also I don't get why the paper puts a prior on the mean of beta's prior, why not just set B ~ Normal((1,1-1),...) instead. Don't get it.

Thinking about it now, it's a baseball paper so having different coefficients for batters and pitchers probably makes sense. But for faceoffs you probably just one coefficient that scales the difference inside the logistic function. Wonder if that's what DNRS did.

Hey guys, I am currently on an offshore book that's offering lines that are pretty different than sharp books like pinnacle, providing opportunity for value, specifically on baseball total bases and pitcher strikeout props.

Currently, I'm taking anything that is 15 points more favorable on my book than pinnacle, but tracking lines for actual % edge to eventually come up with a threshold that is hopefully profitable.

My question is, when comparing the lines I'm getting to pinnacle (which I'm assuming are "sharp", or the "real" lines) for % edge, should I be stripping the pinnacle lines of juice (roughly 7.1%)?

Again, no model is involved, I'm simply comparing prices under the assumption that pinnacle are the "real" lines (my book lags behind and barely adjusts).

Any additional insight is appreciated on the best way to test this/come up with a threshold of simply comparing prices.

Yes, this is called the vig-free line.

https://sportsbettingcalcs.com/betti...ree_calculator

Except that website is spitting out bad numbers, or I have been using bad numbers for a long time. My spreadsheet and http://sportsbettingsites.org/bettin...ig-calculator/ has different numbers.

My guess is you have been using bad numbers. The calculator I built spits the same lines out as this calculator.

My wording was a bit off. The sportsbettingsites website has the same numbers as my spreadsheet.

I was saying those numbers are different from the website that Cody listed. I believe my numbers are correct.

Also, it's good to read the actual rules for series wagers instead of just assuming books will push if a four-game series ends tied 2-2.

i've decided to tackle ncaa volleyball, will share any progress for you guys to laugh at but also happy to hear any advice from anyone about going about it

will be a slow process but google scholar has a ton of articles on it so will be reading up on it during free time this week

Checking in with the team. Have we decided who will be managing NHL Opening Faceoffs this season? In a few weeks after the initial burn-in period, Poogs will you have time to do the NHL SOGs props? That should get hockey under control. There is the NBA Team to Score First prop which has taunted us for a few seasons now mainly due to a lack of good jump ball data. Our partners over at Reddit seem to have done some good initial work on this here. Now that we have a good NFL season simulator from Unabated we have the option of betting NFL futures each week instead of individual games. So much to do, gotta love this time of the year.

What do we think about outsourcing our NBA props modeling to the DFS experts at ETR? There is no other group so willing to give away so much so cheaply as DFS players (other than this syndicate of course).

Unfortunately I don't think this will work because ETR randomly posts their picks one by one throughout the day and their picks are moving lines very quickly.

did you end up finding enough reading material to put a model together and start betting? is so how's it been going?

Our prediction market division is killing it

https://polymarketwhales.info/accounts

Why chase steam when pros prediction trades are fully public on the blockchain

If bovada lists a futures bet that you are convinced is wrong is it morally acceptable to bet it and if you do and it wins are you going to get paid out?

I think they have 2 teams mixed up in who will win a championship with one being -200 and the other +350.