[Gaze73]

Has anyone tried anything based on that? Just using common sense like for example in NHL: Game 1 between Tampa and Washington, the away dog easily won 2:4 at +260 in regular time. Game 2 was 2 days later at the exact same odds so the common sense would say to pick the dog again, and what a shocker, they won 2:6 this time. Yet I've seen two experts pick Tampa ML and -1.5 because "they're the better team" or whatever. Now in game 3 the common sense pick would be Washington again, they're at home this time so it should be even easier and +140 in reg looks like a steal.

Same thing happened in NBA, I don't remember which team some expert recommended like a month ago but IIRC they lost hard 2 games at home and then they were on the road for game 3, why the hell would you back them? Surprise surprise they lost hard again and probably the 4th game too.

The bottom line is when you have super fresh H2H results and the dog won on the road, why would you bet against the dog at home in the next game? Because of some silly revenge factor?

Be it as it may, I'm backing Washington in Reg and -1.5.

[Waterstpub87]

It isn't correct to think this way. One game is relatively meaningless, especially in something low scoring like hockey.

I forget with hockey, but I think it is something like 8% or so shots turn to goals. Easy for a team to get a few lucky goals in a game, and despite being out shot by their opponent, end up winning.

You should instead consider that there are 1,000,000+ possible outcomes for a game. Only one is observed. Why would you put any stock in one result out of 1,000,000?

[Bsims]

You should instead consider that there are 1,000,000+ possible outcomes for a game. Only one is observed. Why would you put any stock in one result out of 1,000,000?

Exactly correct. There is a distribution of potential results. I attempt to identify this distribution and then look at each potential wager and compute what they would do against each of the potential results. 1,000,000 is pretty hard to deal with. When I was running computer simulations, I would generate 5,000 potential results for each game. Currently I'm generating 325+ potential results for each MLB game and use these.

[Waterstpub87]

Exactly correct. There is a distribution of potential results. I attempt to identify this distribution and then look at each potential wager and compute what they would do against each of the potential results. 1,000,000 is pretty hard to deal with. When I was running computer simulations, I would generate 5,000 potential results for each game. Currently I'm generating 325+ potential results for each MLB game and use these.

Similar situation. I run 1,000 sims for each games, down to pitcher vs. batter. I would run more, but I run in VBA/Excel which is kind of slow. I also keep a running tab of total bases, strike outs, Runs/RBI/Hits, Total hits ect. to bet props with. It take me roughly 45 minutes per game to run.

[ChuckyTheGoat]

Exactly correct. There is a distribution of potential results. I attempt to identify this distribution and then look at each potential wager and compute what they would do against each of the potential results. 1,000,000 is pretty hard to deal with. When I was running computer simulations, I would generate 5,000 potential results for each game. Currently I'm generating 325+ potential results for each MLB game and use these.

Nice post, bsims. Good luck.